Additive Value of Texture Analysis Based on Breast MRI for Distinguishing Between Benign and Malignant Non-mass Enhancement in Premenopausal Women

Additive Value of Texture Analysis Based on Breast MRI for Distinguishing Between Benign and Malignant Non-mass Enhancement in Premenopausal Women | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Help Center Sign In Submit a Preprint Cite Share Download PDF Research article Additive Value of Texture Analysis Based on Breast MRI for Distinguishing Between Benign and Malignant Non-mass Enhancement in Premenopausal Women Yu Tan, Hui Mai, Zhiqing Huang, Li Zhang, Chengwei Li, Songxin Wu, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-92781/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 12 Mar, 2021 Read the published version in BMC Medical Imaging → Version 1 posted 10 You are reading this latest preprint version Abstract Background: Non-mass enhancement (NME) is a diagnostic dilemma. Texture analysis (TA) could serve as an objective method to quantify tumor characteristics and growth patterns. However, there are few reports about TA use in NME diagnosis. To our knowledge, NME diagnosis based on the combination of the features noted on routine MRI and TA has not been reported.. The purpose of this study was to explore the value of TA in distinguishing between benign and malignant NME in premenopausal women. Methods: Women in whom NME was histologically proven (n = 147) were enrolled (benign: 58; malignant: 89) was retrospective. Then, 102 and 45 patients were classified as the training and validation groups, respectively. Scanning sequences included Fat-suppressed T2-weighted and fat-suppressed contrast-enhanced T1-weighted which were acquired on a 1.5T MRI system. Clinical and routine MR characteristics (CRMC) were evaluated by two radiologists according to the Breast Imaging and Reporting and Data system (2013). Texture features were extracted from all post-contrast sequences in the training group. The combination model was built and then assessed in the validation group. Pearson’s chi-square test and Mann-Whitney U test were used to compare categorical variables and continuous variables, respectively. Logistic regression analysis and receiver operating characteristic curve were employed to assess the diagnostic performance of CRMC, TA, and their combination model in NME diagnosis. Results: The combination model showed a superior diagnostic performance in differentiating between benign and malignant NME compared to that of CRMC or TA alone (AUC, 0.887 vs 0.832 vs 0.74). Moreover, compared to CRMC, the model showed high specificity (72.5% vs 80%). The results obtained in the validation group confirmed the model was promising. Conclusion: The combined use of TA and CRMC could afford an improved diagnostic performance in differentiating between benign and malignant NME. Nuclear Medicine & Medical Imaging breast non-mass enhancement texture analysis additive value Figures Figure 1 Figure 2 Figure 3 Figure 4 Background According to the Breast Imaging-Reporting and Data System (BI-RADS) magnetic resonance imaging (MRI) lexicon (2013), non-mass enhancement (NME) is defined as a special MRI enhancement mode, which is different from the surrounding enhanced breast parenchyma. It has no space occupation effect and typically contains scattered adipose and glandular tissues[ 1 , 2 ]. NME is often encountered on MRI screening. It might appear in benign breast lesions, such as focal adenosis or fibrocystic and inflammatory changes and can also manifest in malignant lesions, such as lobular carcinoma, diffuse invasive breast cancer, invasive ductal carcinoma, ductal carcinoma in situ (DCIS), and occasionally, some special types of breast cancers[ 3 – 5 ]. Distinguishing between benign and malignant NME is a challenge in breast MRI-based diagnosis. Since biopsy guided by MRI is not popular, over or delayed surgery is frequent[ 6 , 7 ]. Previous studies have shown that specific morphological MRI features and kinetic curve patterns of NME could offer some guidance[ 8 , 9 ]. However, in daily clinical practice, the use of these methods is considered rather limited and controversial. Recent studies have claimed that MRI feature reports have significant inter-observer variability and a lack of quantitative indicators and repeatability[ 10 ]. Moreover, breast tissue affected by hormone effects would add to the difficulty of diagnosis of NME, especially in premenopausal women[ 11 ]. Texture analysis (TA) uses a computer-assisted approach to analyze the statistical difference in the grey-level pixel intensity in extracted medical images, thereby providing an objective way to quantify tumor characteristics and growth patterns[ 12 , 13 ], which is not feasible in traditional radiology evaluations. TA based on breast MRI has shown great potential in terms of offering molecular biology information. It has been used as a “digital biopsy” to distinguish between malignant tumor and benign lesions[ 14 ], to predict outcomes for patients[ 15 ], and to assess treatment responses[ 16 ]. For lesions presenting as NME, it remains unclear what role TA plays in a predictive model based on routine MRI characteristics. The purpose of this study, therefore, was to explore the value of TA in distinguishing between benign and malignant NME in premenopausal women. Methods Study population We searched all breast MRI examination reports in our radiology information systems from January 2015 to March 2019 and selected “NME” as the retrieval keyword. MRI data for 394 female patients were found. Two hundred and ninety-three of the patients met the following criteria: (1) NME confirmed by pathological analysis; (2) MRI performed within 1 week before the surgery and during 7–15 days of the patients’ menstrual cycle to decrease the false-positive results provided by background enhancement (BPE)[ 17 ]; and (3) define the shortest part of its measuring diameter is greater than 0.5 cm, such that the possible adverse effects on the texture features extracted from DICOM data were minimized. The exclusion criteria were as follows: (1) severe motion artifacts in the contrast-enhanced images or the use of different 1.5T machines for scanning (n = 31); (2) a history of treatment for breast cancer, i.e., surgery, biopsy, radiotherapy, or chemotherapy (n = 76); (3) a history of hormone therapy (n = 13); and (4) NME and mass enhancement both existed on the ipsilateral breast simultaneously (n = 26). Finally, 147 patients were included in this retrospective study. Among these patients, 58 had benign lesions while 89 had malignant lesions. Figure 1 shows a flow chart of the inclusion and exclusion criteria for this study. Table 1 summarizes the histological types of these two groups of lesions. Table 1 Histological types of lesions in the two groups: benign and malignant non-mass enhancement Tumor group Number (cases) Percentage Benign non-mass enhancement 58 39.5% Fibrocystic changes 33 22.4% Inflammation 20 13.6% Mix 6 4.1% Malignant non mass enhancement 89 60.5% Invasive ductal carcinoma 16 10.9% Atypical ductal hyperplasia 18 12.2% Ductal carcinoma in situ 11 7.5% Invasive ductal carcinoma 13 8.8% Invasive micropapillary carcinoma 3 2% Apocrine carcinoma 1 0 .7% Mix 26 17.7% MRI protocol All MRI studies were conducted using 1.5 T (T) dedicated breast MRI system (Aurora Imaging Technology, North Andover, MA), equipped with an integrated breast-specific coil. The patients were scanned in the prone position. Dynamic enhanced imaging included a total of five phases performed using a T1-weighted fat-suppression sequence in the axial plane with TR = 29 ms, TE = 4.8 ms, flip angle = 45°, FOV = 36 × 36 cm, slice thickness = 1.12 mm, and gap = 0. A total of 160 slices were used to cover the entire breast. After acquiring one set of pre-contrast images, the contrast medium (gadobenate diethylenetriamine pentaacetic acid, Gd-DTPA, Magnevist) was administered as a bolus injection (infusion rate: 2 ml/s; dose: 0.2 mmol/kg per patient weight), followed by flushing with 20 ml of normal saline. Both the contrast medium and normal saline were injected into the vein through an automated contrast delivery system (Sonic Shot GX; Nemoto Kyorindo, Japan). Four sets of post-contrast enhanced images were obtained. The acquisition time for each was 3 min. In addition, a fat-suppressed T2-weighted sequence was performed with the following parameters: TR = 6680 ms, TE = 5.3 ms, matrix size = 320 × 192, FOV = 36 cm, slice thickness = 3 mm, and gap = 0. All images were further analyzed in using a dedicated workstation equipped with computer-aided detection for further analysis. The time-intensity curve (TIC) was generated by analyzing different color codes of fluid and edema. Clinical and routine MRI characteristic assessment Clinical and routine MRI characteristics (CRMC) were used to distinguish between benign and malignant NME. The clinical variable assessed was age. Routine MRI characteristics were visually assessed by two breast radiologists (reader 1, with over 10 years of experience; reader 2, with more than 13 years of experience) separately and independently. Both readers had access to the patients’ previous clinical and/or imaging information, except for the histopathological diagnosis, during their initial reading. The NME evaluation involved a comparison of both breasts to avoid false-positive results caused by BPE. The final diagnosis was based on the consensus between the two radiologists. The selection of image characteristics was based on the BI-RADS-MRI (2013) diagnostic guidelines, including lesion distribution pattern (focal/linear/segment/regional/multiple regions/diffuse) appeared or not and internal enhanced mode (homogeneous/ heterogeneous/clumped/clustered ring) presence or absence. Figure 2 shown the MRI examples of these features. The relationship between lesion signal intensity and time was evaluated by TIC, which was categorized into persistent, plateau, and washout patterns[ 2 ]. TA The 2nd to 5th contrast sequences were input into Mazda 4.6 (a public software developed by the Institute of Electronics in Lodz Technical University, Poland) for TA. For each case, a region of interest (ROI) was manually delineated by reader 1. The slice of the 2nd contrast phase that was selected to draw the ROI met the following criteria: (1) slice showing the largest cross-section area of the NME and with no visible necrotic areas; and(2) selection of the largest slice when multiple lesions were found on one slice of the same breast. Then, reader 2 double-checked the ROI setting. If there was a disagreement on the boundary, the readers resolved it by discussing between themselves. The same ROI was placed on the same slice for the 3rd to the 5th contrast phases. In addition, to decrease the impact of image brightness and contrast variation on the TA results, the grey-level intensity was normalized within µ + 3σ (µ, mean grey value; σ, mean standard deviation[ 18 ]. T he MaZda TA report could offer almost 300 texture parameters for each ROI. There are six texture feature categories included in this analysis: run-length matrix (RLM), autoregressive model (ARM), wavelet, absolute gradient (GrM), histogram, and the co-occurrence matrix parameters (COM). Additionally, for each ROI, the RLM algorithm was computed in the vertical, horizontal, 45-degree, and 135-degree directions, i.e., four times in all. The COM algorithm was derived from four directions (θ = 0, 45, 90, and 135), and the distance of the pixels ranged from 1 to 5, i.e., for each ROI, each of the five distances was counted separately in the four directions, making up a total of 20. As shown in Table 2 [ 19 ], the combined use of the feature extraction algorithms, including the Fisher coefficient, mutual information, classification error probability, and average correlation coefficients (POE + ACC), afforded the screening of the top 30 texture features with the strongest ability to distinguish between benign and malignant NME. The TA workflow chart for NME is shown in Fig. 3 . Table 2 Texture parameters computed by MaZda Texture Feature Algorithm Parameters RLM Grey-level/run-length nonuniformity, long/short run emphasis, fraction of image in runs. ARM Model parameter vector includes 4 parameters; Sigma: standard deviation of the noise Wavelet Energy of the wavelet coefficients in subbands GrM Kurtosis, skewness, variance, mean, percentage of pixels with a nonzero gradient Histogram Skewness; mean; kurtosis; variance; and perc. 01%, perc. 10%, perc. 50%, perc. 90%, and perc. 99% COM Angular second moment, correlation, contrast, sum of squares, inverse difference moment, sum variance, sum average, sum entropy, entropy, difference variance, difference entropy Note: RLM, run-length matrix; ARM, Auto-regressive model; GrM, absolute gradient; COM, co-occurrence matrix parameters. Statistical analysis Mann-Whitney U test was used to compare continuous variables. Categorical variables evaluated by Pearson’s chi-square test (n > 40, TRC > 5) or Yates’s correction for continuity (n > 40, 1 ≤ TRC < 5). Univariate logistic regression was performed initially on each variable, and the variables showing statistical significance in the univariate logistic regression were further analyzed using multiple logistic regression to establish a discriminating model. For assessing the diagnostic efficacy of each approach, the receiver operating characteristic (ROC) and the area under the curve (AUC) were evaluated. All data analyses were performed on SPSS 22.0(Windows version), and a P value less than 0.05 was considered statistically significant. Validation study To evaluate the diagnostic performance of the combined model, the data were divided into a training dataset of 102 cases and a validation set of 45 cases by simple random sampling with an approximate method in SPSS 22.0. The ratio of the two was 7:3. The mean ages of the training and validation cohorts were 38.7 + 6.8 and 38.1 + 8.7 years, respectively. The number of cases of benign and malignant NME in the training data set were 40 and 62, respectively; the corresponding numbers for the validation set were 18 and 27. The holdout cross-validation method was used to verify the diagnostic performance of the discriminating model constructed in multivariate logistic regression. AUROC values were applied as a measure of success. P < 0.05 was considered statistically different. Results CRMC Among the 102 cases with pathologically proven NME, 40 cases showed benign findings, and 62 cases showed malignancy. Patient age in cases showing benign findings (36.1 ± 6.8) was lower than that in the cases showing malignancy (40.4 ± 6.2). The difference between the two groups was statistically significant (p < 0.001). With the respect to the conventional MRI features of NME, a linear, multiple-region distribution and the washout time-intensity pattern were significantly more frequent in malignant lesions, whereas a distribution of focal areas and a plateau time-intensity pattern were common findings in benign lesions (p < 0.05). In contrast to the distribution (regional, segmental, diffuse) and internal enhancement modes, the persistence time-intensity patterns of NME did not differ significantly between benign and malignant NME (p > 0.05) (Table 3 ). Table 3 Statistical results of clinical and routine MRI findings Clinical and routine MR characteristics Benign (n = 40) Malignant (n = 62) P-value Age (years) 36.1 ± 6.8 (22–49) 40.4 ± 6.2 (25–54) < 0.001 * Distribution Focal area 9(22.5%) 5(8.1%) 0.039 * Linear 10(25%) 41(66.1%) < 0.001 * Segment 33(82.5%) 56(90.3%) 0.247 Regional 23(57.5%) 26(41.9%) 0.125 Multiple regions 12(30.0%) 32(51.6%) 0.031 * Diffuse 3(7.5%) 4(6.5%) 0.838 Internal enhancement Homogeneous 9(22.5%) 13(20.1%) 0.854 Heterogeneous 31(77.5%) 49(79.1%) 0.854 Clumped 32(80.0%) 52(83.9%) 0.617 Clustered ring 9(22.5%) 21(33.9%) 0.218 TIC pattern Persistent 16(40%) 22(35.5%) 0.645 Plateau 13(32.5%) 6(9.7%) 0.004 * Washout 11(27.5%) 34(54.8%) 0.007 * Note: *P-value, statistically significant (P < 0.05) Multivariate logistic regression analysis of CRMC showed 3 independent indicators with statistical significance to discriminate benign and malignant NME, namely, age, linear distribution, and multiple-region distribution (P < 0.05). For ROC analysis, the AUC was 83.7% (CI, 0.76–0.91) and standard error was 0.04 (p < 0.001). The sensitivity was 80.6% and the specificity was 72.5%. Texture features One, four, and eight statistically significant texture features were selected from 2nd, 3rd, and 5th contrast phases respectively, and no statistically significant texture features were found in the 4th contrast phase (Table 4 ). Multivariate logistic regression analysis of TA found that three statistically significant texture features could discriminate benign and malignant NME, which were as follows: S (5, 5) Correlate (p = 0.01) from the second contrast phase, Perc.90% (p = 0.002), and S (4,-4) Correlate (p = 0.001) from the fifth contrast phase. For ROC analysis, the AUC was 74% (CI,0.64–0.84) and standard error was 0.05, (p < 0.001). The sensitivity was 64.5% and the specificity was 70%. Table 4 Statistically significant texture features in the FMC method of contrast phases Dynamic enhanced phases Texture parameters Z-value P-value Algorithm model 2nd phase * S (5,5) Correlat -2.467 0.01 COM 3rd phase Perc.99% -2.20 0.03 Histogram Mean -2.32 0.02 Histogram Perc.50% -2.28 0.02 Histogram Perc.90% -2.40 0.02 Histogram 5th phase Perc.99% -2.29 0.02 Histogram * Perc.90% -2.55 0.01 Histogram Perc.50% -2.31 0.02 Histogram Mean -2.31 0.02 Histogram Teta 3 -2.05 0.04 ARM * S (4, -4) Correlat -2.41 0.02 COM S (5, -5) Correlat -2.51 0.01 COM Variance -2.02 0.04 GRM Note: FMC, methods included Fisher coefficient, mutual information, classification error probability, and average correlation coefficients algorithms; * Data, the statistically significant texture features in the multiple regression analysis, which would be input into the combined diagnosis model to distinguish between benign and malignant NME. Combined model Multiple logistic regression was used to create a combined model to predict malignant NME by using age, linear, multiple regions distribution, Perc.90%, S (5,5) Correlate and S (4, -4) Correlate, which were statistically significant and independent factors (P < 0.05) (Table 5 ). Table 5 Logistic regression results of identifying benign and malignant NME in the training dataset CRMC and texture features B-value P-value Odds ratio 95% confidence level a Age 0.162 < 0.01 1.176 1.07–1.292 a Multiple regions of distribution 1.431 0.015 4.181 1.313–13.311 a Linear 2.283 < 0.01 9.81 2.959–32.521 b S(5,5)Correlat 0.697 0.034 2.009 1.053–3.832 c S(4,-4)Correlat -0.92 < 0.01 0.399 0.199–0.797 c Perc. 90% -0.612 0.044 0.542 0.299–0.984 Note: a Data, features of CRMC; b Data, texture feature from the 2nd contrast phase; C Data, texture features from the 5th contrast phase. For discriminate benign and malignant NME, the combined model shown the best diagnostic efficiency, in comparison to the efficiencies of CRMC and TA alone. Its AUC was 88.7% (CI 0.83–0.95) and standard error was 0.03 (p < 0.001). The sensitivity was 82.3% and specificity was 80% (Table 6 , Fig. 4 ). Table 6 ROC results for CRMC, TA, and combination model AUC (95% CI) Sensitivity Specificity P-value CRMC 83.7% (0.76,0.91) 80.6% 72.5% < 0.01 TA 74% (0.64,0.84) 64.5% 70% < 0.01 Combine training set 88.7% (0.83,0.95) 82.3% 80% < 0.01 Validation set 81.9% (0.68,0.96) 77.8% 72.2% < 0.01 Validation study results . The validation set included 18 benign and 27 malignant cases of NME, with a mean patient age of 38.1 + 8.7 years (range, 16 to 52 years; p < 0.001). To verify the repeatability of the combined model constructed by multiple logistic regression, the holdout cross-validation method was used. Its AUROC was 81.9% (CI 0.68–0.92), sensitivity was 77.8%, and specificity was 72.2%, as shown in Table 6 . Discussion In this study, we assessed the diagnostic value of texture features in discriminating benign and malignant NME. To this end, we compared three diagnostic methods: models using TA or CRMC alone and a model using a combination of these. The diagnostic efficacy obtained with TA alone was not significantly higher than that with CRMC (74% vs 83.2%), but their combination resulted in additive effects and improved diagnostic performance (AUC = 0.887, p < 0.05) (Table 6 , Fig. 4 ). At the same time, the combined model was successfully verified as a promising diagnostic model in the validation set (AUC = 0.819). Our results also indicated that reducing the influence of BPE could improve the diagnostic specificity of CRMC, and this study yielded more information about the use of TA for assessment of NME in premenopausal women. The morphological features and dynamic contrast enhanced (DCE) parameters in benign and malignant NME have been studied extensively. Many investigators confirmed the results obtained by Tozaki et al for the NME internal enhancement and distribution patterns, and they suggested that most benign NMEs appeared with a linear distribution and homogeneous internal enhancement, whereas lesions exhibiting a heterogeneous and clustered ring internal enhancement with segmental distribution were highly suggestive of malignant NME[ 20 – 22 ]. For TIC, previous studies have demonstrated no statistically significant differences between benign and malignant lesions in any type of enhancement pattern[ 23 ]. However, the results of the present study were inconsistent with these findings, since the present study showed that a linear, segmental, and multiple-regions enhancement distribution and washout kinetic pattern were detected more frequently in malignancy, whereas a focal area-enhanced distribution with a plateau kinetic curve pattern was more likely to appear in benign lesions. Moreover, this study showed no evidence that a clumped, cluster ring with a homogeneous or heterogeneous structure was statistically significant in identifying benign or malignant status. For ROC analysis, Z.Z.S et al performed a meta-analysis of diagnostic performance based on morphological characteristics and enhanced parameters by using pooled weighted estimates, and their results indicated low sensitivity (50%) and high (80%) specificity. In contrast, the results of this study indicated high sensitivity (80.6%), while the specificity was not high (72.5%)[ 24 ]. The discrepancy might be attributed to the following reasons. First, the inclusion criteria were different. The criteria for this study included measures to reduce the interference of BPE in NME diagnosis. Since some investigators believed that when BPE manifests as asymmetric, regional, or focal distribution, it was difficult to distinguish BPE from NME[ 25 – 27 ]. Moreover, BPE might interfere with the delineation of tumor boundaries[ 28 ]. However, this major factor that affected the diagnostic accuracy of NME in premenopausal women was ignored by previous studies; Second, the interpretation of morphologic features in MR images was highly dependent on the radiologist’s experience level and lacked reproducibility. This might account for the different sensitivities and specificities of NME diagnosis with routine MRI features. At present, texture analysis by extracting the features of the particular area in an image is considered to be a repeatable and efficient auxiliary diagnostic method, the principle of which is based on the spatial distribution of the intensity level in each pixel[ 10 , 13 ]. Unfortunately, to the best of our knowledge, few studies used TA in NME. Newell D et al. first used TA to diagnose NME, and their ROAUC was not high (0.76)[ 29 ]. The results of subsequent studies were similar, and our TA results were no exception, with the AUC, sensitivity, and specificity all lower than those with CRMC, indicating that the diagnostic efficiency of TA alone in NME diagnosis was not high. Some investigators had used TA combined with breast MRI morphology features to distinguish between phyllodes and fibroadenomas tumors, while others had combined TA with DWI parameters to predict the response to neoadjuvant chemotherapy for breast cancer, and their results demonstrated that combined TA could improve the diagnostic performance[ 30 , 31 ]. On the basis of previous studies, we tried to use the combination of TA and CRMC in NME diagnosis. Our results showed that the diagnostic performance of the model combining TA and CRMC was greater than that achieved with CRMC or TA alone (AUC: 0.887 vs. 0.832 vs. 0.74). Furthermore, in comparison with CRMC, the combined model also showed greater specificity (72.5% vs 80%). In addition, this study found that features from the 2nd and 5th contrast sequences were more meaningful in discriminating benign and malignant NME, which was consistent with previous results showing that the time to enhancement (TTE) and maximum slope (MS) in DCE-MRI could distinguish benign and malignant NME. The pathological and pharmacokinetic mechanisms differed in benign and malignant lesions. Malignant tumors had abundant vascularity and highly permeable vessel walls that allowed easier transfer of the contrast agent from vessels to the extravascular space was easier; thus, malignant lesions had shorter TTE and larger MS, while the benign lesions showed the opposite findings[ 32 , 33 ]. This could explain why in the combined model, the texture features extracted from the 2nd and 5th contrast sequences were independently relevant to discriminate benign and malignant NME. The limitations of our studies should be noted: First, we used a small-sized retrospective database, which is subject to potential bias. Further studies using larger datasets and validating the combined model on other equipment should be attempted in the future. Moreover, manual ROI segmentation led to inevitable measurement errors; thus, the next step is to develop artificial intelligence tools that can accurately recognize these lesions. Conclusions In summary, the addition of TA to CRMC could improve the diagnostic performance in NME, providing a noninvasive quantitative approach for NME diagnosis that could distinguish malignant and benign lesions and decrease the excessive surgery or benign NME core needle biopsy. Abbreviations MRI: Magnetic resonance imaging; NME: Non-mass enhancement; TA: Texture analysis; CRMC: Clinical and routine MR characteristics; DCIS: Ductal carcinoma in situ; BPE: background enhancement; TIC:time-intensity curve; BI-RADS-MRI: Breast Imaging-Reporting and Data System magnetic resonance imaging lexicon; ROI: Region of interest; ROC: receiver operating characteristic; AUC：Area under the curve；DCE：Dynamic contrast enhanced; TTE: the time to enhancement; MS: maximum slope Declarations Authors’ contributions Y.T and H.M：conception and manuscript writing. K.M.J: guarantor of integrity of entire study. Z.Q.H, Z.L, C.W.L, H.H, W.T, Y.X.L: data acquisition and interpretation. S.X.W and W.T: performed the statistical analysis. All authors read the approved the final manuscript. Funding This work was funding by the Third Affiliated Hospital of Guangzhou Medical University Youth Research Project (Grant Number 2017Q07) Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Ethics approval and consent to participate This retrospective study was approved by the ethics committee of Guangdong Women and children hospital. Patient consent for publication Not applicable. Competing interests All authors declare that they have no competing interests. References Mercado CL: BI-RADS update. Radiol Clin North Am 2014, 52(3):481-487. Edwards SD, Lipson JA, Ikeda DM, Lee JM: Updates and revisions to the BI-RADS magnetic resonance imaging lexicon. Magn Reson Imaging Clin N Am 2013, 21(3):483-493. Giess CS, Raza S, Birdwell RL: Patterns of nonmasslike enhancement at screening breast MR imaging of high-risk premenopausal women. Radiographics 2013, 33(5):1343-1360. Milosevic ZC, Nadrljanski MM, Milovanovic ZM, Gusic NZ, Vucicevic SS, Radulovic OS: Breast Dynamic Contrast Enhanced MRI: Fibrocystic Changes Presenting as a Non-mass Enhancement Mimicking Malignancy. Radiol Oncol 2017, 51(2):130-136. Chadashvili T, Ghosh E, Fein-Zachary V, Mehta TS, Venkataraman S, Dialani V, Slanetz PJ: Nonmass enhancement on breast MRI: review of patterns with radiologic-pathologic correlation and discussion of management. AJR Am J Roentgenol 2015, 204(1):219-227. Santoso MR, Yang PC: Magnetic Nanoparticles for Targeting and Imaging of Stem Cells in Myocardial Infarction. Stem Cells Int 2016, 2016:4198790. Dratwa C, Jalaguier-Coudray A, Thomassin-Piana J, Gonin J, Chopier J, Antoine M, Trop I, Darai E, Thomassin-Naggara I: Breast MR biopsy: Pathological and radiological correlation. Eur Radiol 2016, 26(8):2510-2519. Gity M, Ghazi Moghadam K, Jalali AH, Shakiba M: Association of Different MRI BIRADS Descriptors With Malignancy in Non Mass-Like Breast Lesions. Iran Red Crescent Med J 2014, 16(12):e26040. Sakamoto N, Tozaki M, Higa K, Tsunoda Y, Ogawa T, Abe S, Ozaki S, Sakamoto M, Tsuruhara T, Kawano N et al : Categorization of non-mass-like breast lesions detected by MRI. Breast Cancer 2008, 15(3):241-246. Pinker K, Chin J, Melsaether AN, Morris EA, Moy L: Precision Medicine and Radiogenomics in Breast Cancer: New Approaches toward Diagnosis and Treatment. Radiology 2018, 287(3):732-747. Giess CS, Yeh ED, Raza S, Birdwell RL: Background parenchymal enhancement at breast MR imaging: normal patterns, diagnostic challenges, and potential for false-positive and false-negative interpretation. Radiographics 2014, 34(1):234-247. Marino MA, Pinker K, Leithner D, Sung J, Avendano D, Morris EA, Jochelson M: Contrast-Enhanced Mammography and Radiomics Analysis for Noninvasive Breast Cancer Characterization: Initial Results. Mol Imaging Biol 2020, 22(3):780-787. Holli K, Laaperi AL, Harrison L, Luukkaala T, Toivonen T, Ryymin P, Dastidar P, Soimakallio S, Eskola H: Characterization of breast cancer types by texture analysis of magnetic resonance images. Acad Radiol 2010, 17(2):135-141. Chitalia RD, Kontos D: Role of texture analysis in breast MRI as a cancer biomarker: A review. J Magn Reson Imaging 2019, 49(4):927-938. Cao K, Zhao B, Li XT, Li YL, Sun YS: Texture Analysis of Dynamic Contrast-Enhanced MRI in Evaluating Pathologic Complete Response (pCR) of Mass-Like Breast Cancer after Neoadjuvant Therapy. J Oncol 2019, 2019:4731532. Fan M, Wu G, Cheng H, Zhang J, Shao G, Li L: Radiomic analysis of DCE-MRI for prediction of response to neoadjuvant chemotherapy in breast cancer patients. Eur J Radiol 2017, 94:140-147. Kajihara M, Goto M, Hirayama Y, Okunishi S, Kaoku S, Konishi E, Shinkura N: Effect of the menstrual cycle on background parenchymal enhancement in breast MR imaging. Magn Reson Med Sci 2013, 12(1):39-45. Waugh SA, Purdie CA, Jordan LB, Vinnicombe S, Lerski RA, Martin P, Thompson AM: Magnetic resonance imaging texture analysis classification of primary breast cancer. Eur Radiol 2016, 26(2):322-330. Szczypinski PM, Strzelecki M, Materka A, Klepaczko A: MaZda--a software package for image texture analysis. Comput Methods Programs Biomed 2009, 94(1):66-76. Tozaki M, Fukuda K: High-spatial-resolution MRI of non-masslike breast lesions: interpretation model based on BI-RADS MRI descriptors. AJR Am J Roentgenol 2006, 187(2):330-337. Chen QL, Luo Z, Zheng JL, Li XD, Liu CX, Zhao YH, Gong Y: Protective effects of calcium on copper toxicity in Pelteobagrus fulvidraco: copper accumulation, enzymatic activities, histology. Ecotoxicol Environ Saf 2012, 76(2):126-134. Chikarmane SA, Michaels AY, Giess CS: Revisiting Nonmass Enhancement in Breast MRI: Analysis of Outcomes and Follow-Up Using the Updated BI-RADS Atlas. AJR Am J Roentgenol 2017, 209(5):1178-1184. El Khouli RH, Macura KJ, Jacobs MA, Khalil TH, Kamel IR, Dwyer A, Bluemke DA: Dynamic contrast-enhanced MRI of the breast: quantitative method for kinetic curve type assessment. AJR Am J Roentgenol 2009, 193(4):W295-300. Shao Z, Wang H, Li X, Liu P, Zhang S, Cao S: Morphological distribution and internal enhancement architecture of contrast-enhanced magnetic resonance imaging in the diagnosis of non-mass-like breast lesions: a meta-analysis. Breast J 2013, 19(3):259-268. Hegenscheid K, Schmidt CO, Seipel R, Laqua R, Ohlinger R, Hosten N, Puls R: Contrast enhancement kinetics of normal breast parenchyma in dynamic MR mammography: effects of menopausal status, oral contraceptives, and postmenopausal hormone therapy. Eur Radiol 2012, 22(12):2633-2640. DeMartini WB, Liu F, Peacock S, Eby PR, Gutierrez RL, Lehman CD: Background parenchymal enhancement on breast MRI: impact on diagnostic performance. AJR Am J Roentgenol 2012, 198(4):W373-380. Brooks JD, Sung JS, Pike MC, Orlow I, Stanczyk FZ, Bernstein JL, Morris EA: MRI background parenchymal enhancement, breast density and serum hormones in postmenopausal women. Int J Cancer 2018, 143(4):823-830. Amano Y, Woo J, Amano M, Yanagisawa F, Yamamoto H, Tani M: MRI Texture Analysis of Background Parenchymal Enhancement of the Breast. Biomed Res Int 2017, 2017:4845909. Newell D, Nie K, Chen JH, Hsu CC, Yu HJ, Nalcioglu O, Su MY: Selection of diagnostic features on breast MRI to differentiate between malignant and benign lesions using computer-aided diagnosis: differences in lesions presenting as mass and non-mass-like enhancement. Eur Radiol 2010, 20(4):771-781. Mai H, Mao Y, Dong T, Tan Y, Huang X, Wu S, Huang S, Zhong X, Qiu Y, Luo L et al : The Utility of Texture Analysis Based on Breast Magnetic Resonance Imaging in Differentiating Phyllodes Tumors From Fibroadenomas. Front Oncol 2019, 9:1021. Eun NL, Kang D, Son EJ, Park JS, Youk JH, Kim JA, Gweon HM: Texture Analysis with 3.0-T MRI for Association of Response to Neoadjuvant Chemotherapy in Breast Cancer. Radiology 2020, 294(1):31-41. Goto M, Sakai K, Yokota H, Kiba M, Yoshida M, Imai H, Weiland E, Yokota I, Yamada K: Diagnostic performance of initial enhancement analysis using ultra-fast dynamic contrast-enhanced MRI for breast lesions. Eur Radiol 2019, 29(3):1164-1174. Yang X, Dong M, Li S, Chai R, Zhang Z, Li N, Zhang L: Diffusion-weighted imaging or dynamic contrast-enhanced curve: a retrospective analysis of contrast-enhanced magnetic resonance imaging-based differential diagnoses of benign and malignant breast lesions. Eur Radiol 2020. Cite Share Download PDF Status: Published Journal Publication published 12 Mar, 2021 Read the published version in BMC Medical Imaging → Version 1 posted Editorial decision: Major revision 06 Jan, 2021 Review # 2 received at journal 05 Jan, 2021 Reviewer # 2 agreed at journal 01 Jan, 2021 Review # 1 received at journal 09 Dec, 2020 Reviewer # 1 agreed at journal 17 Nov, 2020 Editor assigned by journal 15 Oct, 2020 Reviewers invited by journal 15 Oct, 2020 Submission checks completed at journal 14 Oct, 2020 Editor invited by journal 14 Oct, 2020 First submitted to journal 09 Oct, 2020 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-92781","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research article","associatedPublications":[],"authors":[{"id":3580872,"identity":"4375ee62-2322-4426-8a5a-248855816f93","order_by":0,"name":"Yu Tan","email":"","orcid":"https://orcid.org/0000-0001-5148-3311","institution":"Guangdong Women's and Children's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Tan","suffix":""},{"id":3580873,"identity":"79b95538-bdf7-428e-bbf2-9a73305817d2","order_by":1,"name":"Hui Mai","email":"","orcid":"","institution":"Guangzhou Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Hui","middleName":"","lastName":"Mai","suffix":""},{"id":3580874,"identity":"ed58b97c-21c7-445b-bab2-6ba0ad150c76","order_by":2,"name":"Zhiqing Huang","email":"","orcid":"","institution":"GuangdongWomen's and Children's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Zhiqing","middleName":"","lastName":"Huang","suffix":""},{"id":3580875,"identity":"74e541b1-4eeb-4868-b044-90cd99e2547d","order_by":3,"name":"Li Zhang","email":"","orcid":"","institution":"guangdong women and children hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Zhang","suffix":""},{"id":3580876,"identity":"e54ddbbb-17d2-47f1-9193-b2b4d5e551a8","order_by":4,"name":"Chengwei Li","email":"","orcid":"","institution":"guangdong women and children hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Chengwei","middleName":"","lastName":"Li","suffix":""},{"id":3580877,"identity":"292f4235-cad7-42a0-accb-b88494b500bb","order_by":5,"name":"Songxin Wu","email":"","orcid":"","institution":"guangdong women and chileren hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Songxin","middleName":"","lastName":"Wu","suffix":""},{"id":3580878,"identity":"a0bffe1d-141d-4772-8f07-d34a2b40d769","order_by":6,"name":"Huang Huang","email":"","orcid":"","institution":"guangdong women and children hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Huang","middleName":"","lastName":"Huang","suffix":""},{"id":3580879,"identity":"95df24be-3e61-4410-9c0c-761d20015030","order_by":7,"name":"Wen Tang","email":"","orcid":"","institution":"guangdong women and children hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wen","middleName":"","lastName":"Tang","suffix":""},{"id":3580880,"identity":"d3f708d8-2b12-4863-8798-310acafa589b","order_by":8,"name":"Yongxi Liu","email":"","orcid":"","institution":"guangdong women and children hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yongxi","middleName":"","lastName":"Liu","suffix":""},{"id":3580881,"identity":"45fe0b1a-5d7c-4bfe-9963-3584af61d5bf","order_by":9,"name":"Kuiming Jiang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1klEQVRIiWNgGAWjYBACCQaGhANAWo6xvYHhAElajJl7DhCvBQwS2WckEOkwyWkHHh74uaM2gXfm84eHC2oY5PnFCFgmLZ2QcLD3zPE8ydk5BodnHGMwnDmbgHVyQC0HeNuOFRvOzmE4zMPGkGBwmwgtB/+2HUvcf/P4g8M8/4jQAnLYYd62msTGGQwGQAYRWiRnA7XIth0wZuwB+oW3T4KwXyRu5yR/fNtWB4zK448/83yzkeeXJqCFgYEHpOIw3AhCykGA/QCQqCNG5SgYBaNgFIxUAAAT90sBh1a4UAAAAABJRU5ErkJggg==","orcid":"","institution":"guangdong women and children hospital","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Kuiming","middleName":"","lastName":"Jiang","suffix":""}],"badges":[],"createdAt":"2020-10-14 19:24:41","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-92781/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-92781/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12880-021-00571-x","type":"published","date":"2021-03-12T15:00:29+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":3063102,"identity":"76c67c87-67ac-49f5-a5dc-ce377654685b","added_by":"auto","created_at":"2020-10-19 15:43:37","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":602906,"visible":true,"origin":"","legend":"The workflow of the inclusion and exclusion criteria of this study.","description":"","filename":"Fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-92781/v1/aa0d35e9dd4256619666a86d.jpg"},{"id":3063103,"identity":"ec9334da-48a2-4526-a46f-983c4e2dae43","added_by":"auto","created_at":"2020-10-19 15:43:37","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":107213,"visible":true,"origin":"","legend":"The sequence of dynamic imaging performed using axial T1-weighted fat-suppression. (a) The second contrast-phase image shows an NME with homogeneous internal enhancement patterns (red irregular shape) and a linear distribution (red arrow). (b) The third contrast-phase image shows that the entire lesion enhancement was heterogeneous, with intermingling local, cluster-ring, and clumped enhancement (purple irregular shapes) and the appearance of segmental distribution (yellow arrow). (c) and (d) respectively show the focal (green irregular shape) and diffuse (green arrow) distribution patterns of NME in the fifth contrast-phase image. (e) and (f) respectively show the regional (orange irregular shape) and multiple-region (blue arrow) distributions in the second contrast phase.","description":"","filename":"Fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-92781/v1/cda8c458d1c875ce601cf75f.jpg"},{"id":3063104,"identity":"aef32103-5c33-4d8f-9dac-7df1d30efe4c","added_by":"auto","created_at":"2020-10-19 15:43:37","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":110193,"visible":true,"origin":"","legend":"Workflow for identification of benign and malignant non-mass enhancement based on texture analysis. Processes in blue boxes were performed in MaZda.","description":"","filename":"Fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-92781/v1/310a13ff7bf343d7d61910b1.jpg"},{"id":3063105,"identity":"44fa4817-4562-4cd0-994b-df30b69ad129","added_by":"auto","created_at":"2020-10-19 15:43:37","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":396560,"visible":true,"origin":"","legend":"The ROC curves in different diagnostic methods to distinguish benign and malignant NME. (a) CRMC, TA, and combination diagnostic models in the training dataset; (b) the combination model in the validation data set.","description":"","filename":"Fig4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-92781/v1/f96345e1ed2962d7a6445a56.jpg"},{"id":13604966,"identity":"8c5d559a-9176-450c-a78f-ac1537f31a31","added_by":"auto","created_at":"2021-09-17 06:02:10","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":657207,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-92781/v1/c9de12aa-d8dc-40e0-868e-094b2505d6ab.pdf"}],"financialInterests":"","formattedTitle":"\u003cp\u003eAdditive Value of Texture Analysis Based on Breast MRI for Distinguishing Between Benign and Malignant Non-mass Enhancement in Premenopausal Women\u003c/p\u003e","fulltext":[{"header":"Background","content":" \u003cp\u003eAccording to the Breast Imaging-Reporting and Data System (BI-RADS) magnetic resonance imaging (MRI) lexicon (2013), non-mass enhancement (NME) is defined as a special MRI enhancement mode, which is different from the surrounding enhanced breast parenchyma. It has no space occupation effect and typically contains scattered adipose and glandular tissues[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. NME is often encountered on MRI screening. It might appear in benign breast lesions, such as focal adenosis or fibrocystic and inflammatory changes and can also manifest in malignant lesions, such as lobular carcinoma, diffuse invasive breast cancer, invasive ductal carcinoma, ductal carcinoma in situ (DCIS), and occasionally, some special types of breast cancers[\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eDistinguishing between benign and malignant NME is a challenge in breast MRI-based diagnosis. Since biopsy guided by MRI is not popular, over or delayed surgery is frequent[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Previous studies have shown that specific morphological MRI features and kinetic curve patterns of NME could offer some guidance[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. However, in daily clinical practice, the use of these methods is considered rather limited and controversial. Recent studies have claimed that MRI feature reports have significant inter-observer variability and a lack of quantitative indicators and repeatability[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Moreover, breast tissue affected by hormone effects would add to the difficulty of diagnosis of NME, especially in premenopausal women[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTexture analysis (TA) uses a computer-assisted approach to analyze the statistical difference in the grey-level pixel intensity in extracted medical images, thereby providing an objective way to quantify tumor characteristics and growth patterns[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], which is not feasible in traditional radiology evaluations. TA based on breast MRI has shown great potential in terms of offering molecular biology information. It has been used as a \u0026ldquo;digital biopsy\u0026rdquo; to distinguish between malignant tumor and benign lesions[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], to predict outcomes for patients[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], and to assess treatment responses[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. For lesions presenting as NME, it remains unclear what role TA plays in a predictive model based on routine MRI characteristics. The purpose of this study, therefore, was to explore the value of TA in distinguishing between benign and malignant NME in premenopausal women.\u003c/p\u003e "},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n\u003ch2\u003eStudy population\u003c/h2\u003e\n\u003cp\u003eWe searched all breast MRI examination reports in our radiology information systems from January 2015 to March 2019 and selected \u0026ldquo;NME\u0026rdquo; as the retrieval keyword. MRI data for 394 female patients were found. Two hundred and ninety-three of the patients met the following criteria: (1) NME confirmed by pathological analysis; (2) MRI performed within 1 week before the surgery and during 7\u0026ndash;15 days of the patients\u0026rsquo; menstrual cycle to decrease the false-positive results provided by background enhancement (BPE)[\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e]; and (3) define the shortest part of its measuring diameter is greater than 0.5\u0026nbsp;cm, such that the possible adverse effects on the texture features extracted from DICOM data were minimized. The exclusion criteria were as follows: (1) severe motion artifacts in the contrast-enhanced images or the use of different 1.5T machines for scanning (n\u0026thinsp;=\u0026thinsp;31); (2) a history of treatment for breast cancer, i.e., surgery, biopsy, radiotherapy, or chemotherapy (n\u0026thinsp;=\u0026thinsp;76); (3) a history of hormone therapy (n\u0026thinsp;=\u0026thinsp;13); and (4) NME and mass enhancement both existed on the ipsilateral breast simultaneously (n\u0026thinsp;=\u0026thinsp;26). Finally, 147 patients were included in this retrospective study. Among these patients, 58 had benign lesions while 89 had malignant lesions. Figure\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e shows a flow chart of the inclusion and exclusion criteria for this study. Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes the histological types of these two groups of lesions.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab1\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eHistological types of lesions in the two groups: benign and malignant non-mass enhancement\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eTumor group\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eNumber (cases)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003ePercentage\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eBenign non-mass enhancement\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e58\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e39.5%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eFibrocystic changes\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e33\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e22.4%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eInflammation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e20\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e13.6%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMix\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e6\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e4.1%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eMalignant non mass enhancement\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e89\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e60.5%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eInvasive ductal carcinoma\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e16\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e10.9%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAtypical ductal hyperplasia\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e18\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e12.2%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDuctal carcinoma in situ\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e11\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e7.5%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eInvasive ductal carcinoma\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e13\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e8.8%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eInvasive micropapillary carcinoma\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eApocrine carcinoma\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e1\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e0\u0026nbsp;.7%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMix\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e26\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e17.7%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n\u003ch2\u003eMRI protocol\u003c/h2\u003e\n\u003cp\u003eAll MRI studies were conducted using 1.5\u0026nbsp;T (T) dedicated breast MRI system (Aurora Imaging Technology, North Andover, MA), equipped with an integrated breast-specific coil. The patients were scanned in the prone position. Dynamic enhanced imaging included a total of five phases performed using a T1-weighted fat-suppression sequence in the axial plane with TR\u0026thinsp;=\u0026thinsp;29\u0026nbsp;ms, TE\u0026thinsp;=\u0026thinsp;4.8\u0026nbsp;ms, flip angle\u0026thinsp;=\u0026thinsp;45\u0026deg;, FOV\u0026thinsp;=\u0026thinsp;36\u0026thinsp;\u0026times;\u0026thinsp;36\u0026nbsp;cm, slice thickness\u0026thinsp;=\u0026thinsp;1.12\u0026nbsp;mm, and gap\u0026thinsp;=\u0026thinsp;0. A total of 160 slices were used to cover the entire breast. After acquiring one set of pre-contrast images, the contrast medium (gadobenate diethylenetriamine pentaacetic acid, Gd-DTPA, Magnevist) was administered as a bolus injection (infusion rate: 2\u0026nbsp;ml/s; dose: 0.2\u0026nbsp;mmol/kg per patient weight), followed by flushing with 20\u0026nbsp;ml of normal saline. Both the contrast medium and normal saline were injected into the vein through an automated contrast delivery system (Sonic Shot GX; Nemoto Kyorindo, Japan). Four sets of post-contrast enhanced images were obtained. The acquisition time for each was 3\u0026nbsp;min. In addition, a fat-suppressed T2-weighted sequence was performed with the following parameters: TR\u0026thinsp;=\u0026thinsp;6680\u0026nbsp;ms, TE\u0026thinsp;=\u0026thinsp;5.3\u0026nbsp;ms, matrix size\u0026thinsp;=\u0026thinsp;320\u0026thinsp;\u0026times;\u0026thinsp;192, FOV\u0026thinsp;=\u0026thinsp;36\u0026nbsp;cm, slice thickness\u0026thinsp;=\u0026thinsp;3\u0026nbsp;mm, and gap\u0026thinsp;=\u0026thinsp;0.\u003c/p\u003e\n\u003cp\u003eAll images were further analyzed in using a dedicated workstation equipped with computer-aided detection for further analysis. The time-intensity curve (TIC) was generated by analyzing different color codes of fluid and edema.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch2\u003eClinical and routine MRI characteristic assessment\u003c/h2\u003e\n\u003cp\u003eClinical and routine MRI characteristics (CRMC) were used to distinguish between benign and malignant NME. The clinical variable assessed was age. Routine MRI characteristics were visually assessed by two breast radiologists (reader 1, with over 10\u0026nbsp;years of experience; reader 2, with more than 13\u0026nbsp;years of experience) separately and independently. Both readers had access to the patients\u0026rsquo; previous clinical and/or imaging information, except for the histopathological diagnosis, during their initial reading. The NME evaluation involved a comparison of both breasts to avoid false-positive results caused by BPE. The final diagnosis was based on the consensus between the two radiologists.\u003c/p\u003e\n\u003cp\u003eThe selection of image characteristics was based on the BI-RADS-MRI (2013) diagnostic guidelines, including lesion distribution pattern (focal/linear/segment/regional/multiple regions/diffuse) appeared or not and internal enhanced mode (homogeneous/ heterogeneous/clumped/clustered ring) presence or absence. Figure\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e shown the MRI examples of these features. The relationship between lesion signal intensity and time was evaluated by TIC, which was categorized into persistent, plateau, and washout patterns[\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e\n\u003ch2\u003eTA\u003c/h2\u003e\n\u003cp\u003eThe 2nd to 5th contrast sequences were input into Mazda 4.6 (a public software developed by the Institute of Electronics in Lodz Technical University, Poland) for TA. For each case, a region of interest (ROI) was manually delineated by reader 1. The slice of the 2nd contrast phase that was selected to draw the ROI met the following criteria: (1) slice showing the largest cross-section area of the NME and with no visible necrotic areas; and(2) selection of the largest slice when multiple lesions were found on one slice of the same breast. Then, reader 2 double-checked the ROI setting. If there was a disagreement on the boundary, the readers resolved it by discussing between themselves. The same ROI was placed on the same slice for the 3rd to the 5th contrast phases. In addition, to decrease the impact of image brightness and contrast variation on the TA results, the grey-level intensity was normalized within \u0026micro;\u0026thinsp;+\u0026thinsp;3\u0026sigma; (\u0026micro;, mean grey value; \u0026sigma;, mean standard deviation[\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eT\u003c/strong\u003ehe MaZda TA report could offer almost 300 texture parameters for each ROI. There are six texture feature categories included in this analysis: run-length matrix (RLM), autoregressive model (ARM), wavelet, absolute gradient (GrM), histogram, and the co-occurrence matrix parameters (COM). Additionally, for each ROI, the RLM algorithm was computed in the vertical, horizontal, 45-degree, and 135-degree directions, i.e., four times in all. The COM algorithm was derived from four directions (\u0026theta;\u0026thinsp;=\u0026thinsp;0, 45, 90, and 135), and the distance of the pixels ranged from 1 to 5, i.e., for each ROI, each of the five distances was counted separately in the four directions, making up a total of 20. As shown in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e [\u003cspan class=\"CitationRef\"\u003e19\u003c/span\u003e], the combined use of the feature extraction algorithms, including the Fisher coefficient, mutual information, classification error probability, and average correlation coefficients (POE\u0026thinsp;+\u0026thinsp;ACC), afforded the screening of the top 30 texture features with the strongest ability to distinguish between benign and malignant NME. The TA workflow chart for NME is shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab2\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eTexture parameters computed by MaZda\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eTexture Feature Algorithm\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eParameters\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRLM\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGrey-level/run-length nonuniformity, long/short run emphasis, fraction of image in runs.\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eARM\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eModel parameter vector includes 4 parameters; Sigma: standard deviation of the noise\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eWavelet\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEnergy of the wavelet coefficients in subbands\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGrM\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eKurtosis, skewness, variance, mean, percentage of pixels with a nonzero gradient\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSkewness; mean; kurtosis; variance; and perc. 01%, perc. 10%, perc. 50%, perc. 90%, and perc. 99%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCOM\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAngular second moment, correlation, contrast, sum of squares, inverse difference moment, sum variance, sum average, sum entropy, entropy, difference variance, difference entropy\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003ctfoot\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"2\"\u003eNote: RLM, run-length matrix; ARM, Auto-regressive model; GrM, absolute gradient; COM, co-occurrence matrix parameters.\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tfoot\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n\u003ch2\u003eStatistical analysis\u003c/h2\u003e\n\u003cp\u003eMann-Whitney U test was used to compare continuous variables. Categorical variables evaluated by Pearson\u0026rsquo;s chi-square test (n\u0026thinsp;\u0026gt;\u0026thinsp;40, TRC\u0026thinsp;\u0026gt;\u0026thinsp;5) or Yates\u0026rsquo;s correction for continuity (n\u0026thinsp;\u0026gt;\u0026thinsp;40, 1\u0026thinsp;\u0026le;\u0026thinsp;TRC\u0026thinsp;\u0026lt;\u0026thinsp;5). Univariate logistic regression was performed initially on each variable, and the variables showing statistical significance in the univariate logistic regression were further analyzed using multiple logistic regression to establish a discriminating model.\u003c/p\u003e\n\u003cp\u003eFor assessing the diagnostic efficacy of each approach, the receiver operating characteristic (ROC) and the area under the curve (AUC) were evaluated. All data analyses were performed on SPSS 22.0(Windows version), and a P value less than 0.05 was considered statistically significant.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch2\u003eValidation study\u003c/h2\u003e\n\u003cp\u003eTo evaluate the diagnostic performance of the combined model, the data were divided into a training dataset of 102 cases and a validation set of 45 cases by simple random sampling with an approximate method in SPSS 22.0. The ratio of the two was 7:3. The mean ages of the training and validation cohorts were 38.7\u0026thinsp;+\u0026thinsp;6.8 and 38.1\u0026thinsp;+\u0026thinsp;8.7 years, respectively. The number of cases of benign and malignant NME in the training data set were 40 and 62, respectively; the corresponding numbers for the validation set were 18 and 27. The holdout cross-validation method was used to verify the diagnostic performance of the discriminating model constructed in multivariate logistic regression. AUROC values were applied as a measure of success. P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically different.\u003c/p\u003e"},{"header":"Results","content":"\u003ch2\u003eCRMC\u003c/h2\u003e\n\u003cp\u003eAmong the 102 cases with pathologically proven NME, 40 cases showed benign findings, and 62 cases showed malignancy. Patient age in cases showing benign findings (36.1\u0026thinsp;\u0026plusmn;\u0026thinsp;6.8) was lower than that in the cases showing malignancy (40.4\u0026thinsp;\u0026plusmn;\u0026thinsp;6.2). The difference between the two groups was statistically significant (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\n\u003cp\u003eWith the respect to the conventional MRI features of NME, a linear, multiple-region distribution and the washout time-intensity pattern were significantly more frequent in malignant lesions, whereas a distribution of focal areas and a plateau time-intensity pattern were common findings in benign lesions (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). In contrast to the distribution (regional, segmental, diffuse) and internal enhancement modes, the persistence time-intensity patterns of NME did not differ significantly between benign and malignant NME (p\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab3\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eStatistical results of clinical and routine MRI findings\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eClinical and routine MR characteristics\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eBenign\u003c/p\u003e\n\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;40)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMalignant\u003c/p\u003e\n\u003cp\u003e(n\u0026thinsp;=\u0026thinsp;62)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eP-value\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAge (years)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e36.1\u0026thinsp;\u0026plusmn;\u0026thinsp;6.8\u003c/p\u003e\n\u003cp\u003e(22\u0026ndash;49)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e40.4\u0026thinsp;\u0026plusmn;\u0026thinsp;6.2\u003c/p\u003e\n\u003cp\u003e(25\u0026ndash;54)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eDistribution\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eFocal area\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e9(22.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e5(8.1%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.039\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eLinear\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e10(25%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e41(66.1%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSegment\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e33(82.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e56(90.3%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.247\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRegional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e23(57.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e26(41.9%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.125\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMultiple regions\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e12(30.0%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e32(51.6%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.031\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDiffuse\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3(7.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e4(6.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.838\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eInternal enhancement\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHomogeneous\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e9(22.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e13(20.1%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.854\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHeterogeneous\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e31(77.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e49(79.1%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.854\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eClumped\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e32(80.0%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e52(83.9%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.617\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eClustered ring\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e9(22.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e21(33.9%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.218\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eTIC pattern\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePersistent\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e16(40%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e22(35.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.645\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePlateau\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e13(32.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e6(9.7%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.004\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eWashout\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e11(27.5%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e34(54.8%)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.007\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003ctfoot\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"4\"\u003eNote: *P-value, statistically significant (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05)\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tfoot\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMultivariate logistic regression analysis of CRMC showed 3 independent indicators with statistical significance to discriminate benign and malignant NME, namely, age, linear distribution, and multiple-region distribution (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). For ROC analysis, the AUC was 83.7% (CI, 0.76\u0026ndash;0.91) and standard error was 0.04 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The sensitivity was 80.6% and the specificity was 72.5%.\u003c/p\u003e\n\u003ch2\u003eTexture features\u003c/h2\u003e\n\u003cp\u003eOne, four, and eight statistically significant texture features were selected from 2nd, 3rd, and 5th contrast phases respectively, and no statistically significant texture features were found in the 4th contrast phase (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e). Multivariate logistic regression analysis of TA found that three statistically significant texture features could discriminate benign and malignant NME, which were as follows: S (5, 5) Correlate (p\u0026thinsp;=\u0026thinsp;0.01) from the second contrast phase, Perc.90% (p\u0026thinsp;=\u0026thinsp;0.002), and S (4,-4) Correlate (p\u0026thinsp;=\u0026thinsp;0.001) from the fifth contrast phase. For ROC analysis, the AUC was 74% (CI,0.64\u0026ndash;0.84) and standard error was 0.05, (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The sensitivity was 64.5% and the specificity was 70%.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab4\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eStatistically significant texture features in the FMC method of contrast phases\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eDynamic enhanced\u003c/p\u003e\n\u003cp\u003ephases\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eTexture parameters\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eZ-value\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eP-value\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eAlgorithm model\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2nd phase\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003e*\u003c/sup\u003eS (5,5) Correlat\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.467\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCOM\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3rd \u0026nbsp;phase\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePerc.99%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.20\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.03\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMean\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.32\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePerc.50%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.28\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePerc.90%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.40\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e5th phase\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePerc.99%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.29\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003e*\u003c/sup\u003ePerc.90%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.55\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePerc.50%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.31\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMean\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.31\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHistogram\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTeta 3\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.05\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.04\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eARM\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003e*\u003c/sup\u003eS (4, -4) Correlat\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.41\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCOM\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eS (5, -5) Correlat\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.51\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCOM\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eVariance\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-2.02\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.04\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGRM\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003ctfoot\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\"\u003eNote: FMC, methods included Fisher coefficient, mutual information, classification error probability, and average correlation coefficients algorithms;\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\"\u003e\u003csup\u003e*\u003c/sup\u003eData, the statistically significant texture features in the multiple regression analysis, which would be input into the combined diagnosis model to distinguish between benign and malignant NME.\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tfoot\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003c/div\u003e\n\u003ch2\u003eCombined model\u003c/h2\u003e\n\u003cp\u003eMultiple logistic regression was used to create a combined model to predict malignant NME by using age, linear, multiple regions distribution, Perc.90%, S (5,5) Correlate and S (4, -4) Correlate, which were statistically significant and independent factors (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab5\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eLogistic regression results of identifying benign and malignant NME in the training dataset\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eCRMC and texture features\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eB-value\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eP-value\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eOdds ratio\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003e95% confidence level\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003ea\u003c/sup\u003eAge\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.162\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e1.176\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e1.07\u0026ndash;1.292\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003ea\u003c/sup\u003e Multiple regions of distribution\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e1.431\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.015\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e4.181\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e1.313\u0026ndash;13.311\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003ea\u003c/sup\u003eLinear\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e2.283\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e9.81\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e2.959\u0026ndash;32.521\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003eb\u003c/sup\u003eS(5,5)Correlat\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.697\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.034\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e2.009\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e1.053\u0026ndash;3.832\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003ec\u003c/sup\u003eS(4,-4)Correlat\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-0.92\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.399\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.199\u0026ndash;0.797\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003csup\u003ec\u003c/sup\u003ePerc. 90%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e-0.612\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.044\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.542\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.299\u0026ndash;0.984\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003ctfoot\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\"\u003eNote: \u003csup\u003ea\u003c/sup\u003eData, features of CRMC;\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\"\u003e\u003csup\u003eb\u003c/sup\u003eData, texture feature from the 2nd contrast phase;\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"5\"\u003e\u003csup\u003eC\u003c/sup\u003eData, texture features from the 5th contrast phase.\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tfoot\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor discriminate benign and malignant NME, the combined model shown the best diagnostic efficiency, in comparison to the efficiencies of CRMC and TA alone. Its AUC was 88.7% (CI 0.83\u0026ndash;0.95) and standard error was 0.03 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The sensitivity was 82.3% and specificity was 80% (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab6\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eROC results for CRMC, TA, and combination model\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n\u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eAUC\u003c/p\u003e\n\u003cp\u003e(95% CI)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eSensitivity\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eSpecificity\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eP-value\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCRMC\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e83.7%\u003c/p\u003e\n\u003cp\u003e(0.76,0.91)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e80.6%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e72.5%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e74%\u003c/p\u003e\n\u003cp\u003e(0.64,0.84)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e64.5%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e70%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd rowspan=\"2\" align=\"left\"\u003e\n\u003cp\u003eCombine\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003etraining set\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e88.7%\u003c/p\u003e\n\u003cp\u003e(0.83,0.95)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e82.3%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e80%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eValidation\u003c/p\u003e\n\u003cp\u003eset\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e81.9%\u003c/p\u003e\n\u003cp\u003e(0.68,0.96)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e77.8%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e72.2%\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eValidation study results\u003c/em\u003e. The validation set included 18 benign and 27 malignant cases of NME, with a mean patient age of 38.1\u0026thinsp;+\u0026thinsp;8.7\u0026nbsp;years (range, 16 to 52 years; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). To verify the repeatability of the combined model constructed by multiple logistic regression, the holdout cross-validation method was used. Its AUROC was 81.9% (CI 0.68\u0026ndash;0.92), sensitivity was 77.8%, and specificity was 72.2%, as shown in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003c/p\u003e"},{"header":"Discussion","content":" \u003cp\u003eIn this study, we assessed the diagnostic value of texture features in discriminating benign and malignant NME. To this end, we compared three diagnostic methods: models using TA or CRMC alone and a model using a combination of these. The diagnostic efficacy obtained with TA alone was not significantly higher than that with CRMC (74% vs 83.2%), but their combination resulted in additive effects and improved diagnostic performance (AUC\u0026thinsp;=\u0026thinsp;0.887, p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). At the same time, the combined model was successfully verified as a promising diagnostic model in the validation set (AUC\u0026thinsp;=\u0026thinsp;0.819). Our results also indicated that reducing the influence of BPE could improve the diagnostic specificity of CRMC, and this study yielded more information about the use of TA for assessment of NME in premenopausal women.\u003c/p\u003e \u003cp\u003eThe morphological features and dynamic contrast enhanced (DCE) parameters in benign and malignant NME have been studied extensively. Many investigators confirmed the results obtained by Tozaki et al for the NME internal enhancement and distribution patterns, and they suggested that most benign NMEs appeared with a linear distribution and homogeneous internal enhancement, whereas lesions exhibiting a heterogeneous and clustered ring internal enhancement with segmental distribution were highly suggestive of malignant NME[\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. For TIC, previous studies have demonstrated no statistically significant differences between benign and malignant lesions in any type of enhancement pattern[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. However, the results of the present study were inconsistent with these findings, since the present study showed that a linear, segmental, and multiple-regions enhancement distribution and washout kinetic pattern were detected more frequently in malignancy, whereas a focal area-enhanced distribution with a plateau kinetic curve pattern was more likely to appear in benign lesions. Moreover, this study showed no evidence that a clumped, cluster ring with a homogeneous or heterogeneous structure was statistically significant in identifying benign or malignant status. For ROC analysis, Z.Z.S et al performed a meta-analysis of diagnostic performance based on morphological characteristics and enhanced parameters by using pooled weighted estimates, and their results indicated low sensitivity (50%) and high (80%) specificity. In contrast, the results of this study indicated high sensitivity (80.6%), while the specificity was not high (72.5%)[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe discrepancy might be attributed to the following reasons. First, the inclusion criteria were different. The criteria for this study included measures to reduce the interference of BPE in NME diagnosis. Since some investigators believed that when BPE manifests as asymmetric, regional, or focal distribution, it was difficult to distinguish BPE from NME[\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Moreover, BPE might interfere with the delineation of tumor boundaries[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. However, this major factor that affected the diagnostic accuracy of NME in premenopausal women was ignored by previous studies; Second, the interpretation of morphologic features in MR images was highly dependent on the radiologist\u0026rsquo;s experience level and lacked reproducibility. This might account for the different sensitivities and specificities of NME diagnosis with routine MRI features.\u003c/p\u003e \u003cp\u003eAt present, texture analysis by extracting the features of the particular area in an image is considered to be a repeatable and efficient auxiliary diagnostic method, the principle of which is based on the spatial distribution of the intensity level in each pixel[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Unfortunately, to the best of our knowledge, few studies used TA in NME. Newell D et al. first used TA to diagnose NME, and their ROAUC was not high (0.76)[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The results of subsequent studies were similar, and our TA results were no exception, with the AUC, sensitivity, and specificity all lower than those with CRMC, indicating that the diagnostic efficiency of TA alone in NME diagnosis was not high. Some investigators had used TA combined with breast MRI morphology features to distinguish between phyllodes and fibroadenomas tumors, while others had combined TA with DWI parameters to predict the response to neoadjuvant chemotherapy for breast cancer, and their results demonstrated that combined TA could improve the diagnostic performance[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. On the basis of previous studies, we tried to use the combination of TA and CRMC in NME diagnosis. Our results showed that the diagnostic performance of the model combining TA and CRMC was greater than that achieved with CRMC or TA alone (AUC: 0.887 vs. 0.832 vs. 0.74). Furthermore, in comparison with CRMC, the combined model also showed greater specificity (72.5% vs 80%).\u003c/p\u003e \u003cp\u003eIn addition, this study found that features from the 2nd and 5th contrast sequences were more meaningful in discriminating benign and malignant NME, which was consistent with previous results showing that the time to enhancement (TTE) and maximum slope (MS) in DCE-MRI could distinguish benign and malignant NME. The pathological and pharmacokinetic mechanisms differed in benign and malignant lesions. Malignant tumors had abundant vascularity and highly permeable vessel walls that allowed easier transfer of the contrast agent from vessels to the extravascular space was easier; thus, malignant lesions had shorter TTE and larger MS, while the benign lesions showed the opposite findings[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This could explain why in the combined model, the texture features extracted from the 2nd and 5th contrast sequences were independently relevant to discriminate benign and malignant NME.\u003c/p\u003e \u003cp\u003eThe limitations of our studies should be noted: First, we used a small-sized retrospective database, which is subject to potential bias. Further studies using larger datasets and validating the combined model on other equipment should be attempted in the future. Moreover, manual ROI segmentation led to inevitable measurement errors; thus, the next step is to develop artificial intelligence tools that can accurately recognize these lesions.\u003c/p\u003e "},{"header":"Conclusions","content":" \u003cp\u003eIn summary, the addition of TA to CRMC could improve the diagnostic performance in NME, providing a noninvasive quantitative approach for NME diagnosis that could distinguish malignant and benign lesions and decrease the excessive surgery or benign NME core needle biopsy.\u003c/p\u003e "},{"header":"Abbreviations","content":"\u003cp\u003eMRI: Magnetic resonance imaging; NME: Non-mass enhancement; TA: Texture analysis; CRMC: Clinical and routine MR characteristics; DCIS: Ductal carcinoma in situ; BPE: background enhancement; TIC:time-intensity curve; BI-RADS-MRI: Breast Imaging-Reporting and Data System magnetic resonance imaging lexicon; ROI: Region of interest; ROC: receiver operating characteristic; AUC：Area under the curve；DCE：Dynamic contrast enhanced; TTE: the time to enhancement; MS: maximum slope\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthors\u0026rsquo; contributions\u003c/h2\u003e\n\u003cp\u003eY.T and H.M：conception and manuscript writing. K.M.J: guarantor of integrity of entire study. Z.Q.H, Z.L, C.W.L, H.H, W.T, Y.X.L: data acquisition and interpretation. S.X.W and W.T: performed the statistical analysis. All authors read the approved the final manuscript.\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis work was funding by the Third Affiliated Hospital of Guangzhou Medical University Youth Research Project (Grant Number 2017Q07)\u003c/p\u003e\n\u003ch2\u003eAvailability of data and materials\u003c/h2\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003ch2\u003eEthics approval and consent to participate\u003c/h2\u003e\n\u003cp\u003eThis retrospective study was approved by the ethics committee of Guangdong Women and children hospital.\u003c/p\u003e\n\u003ch2\u003ePatient consent for publication\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eAll authors declare that they have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMercado CL: BI-RADS update. \u003cem\u003eRadiol Clin North Am \u003c/em\u003e2014, 52(3):481-487.\u003c/li\u003e\n\u003cli\u003eEdwards SD, Lipson JA, Ikeda DM, Lee JM: Updates and revisions to the BI-RADS magnetic resonance imaging lexicon. \u003cem\u003eMagn Reson Imaging Clin N Am \u003c/em\u003e2013, 21(3):483-493.\u003c/li\u003e\n\u003cli\u003eGiess CS, Raza S, Birdwell RL: Patterns of nonmasslike enhancement at screening breast MR imaging of high-risk premenopausal women. \u003cem\u003eRadiographics \u003c/em\u003e2013, 33(5):1343-1360.\u003c/li\u003e\n\u003cli\u003eMilosevic ZC, Nadrljanski MM, Milovanovic ZM, Gusic NZ, Vucicevic SS, Radulovic OS: Breast Dynamic Contrast Enhanced MRI: Fibrocystic Changes Presenting as a Non-mass Enhancement Mimicking Malignancy. \u003cem\u003eRadiol Oncol \u003c/em\u003e2017, 51(2):130-136.\u003c/li\u003e\n\u003cli\u003eChadashvili T, Ghosh E, Fein-Zachary V, Mehta TS, Venkataraman S, Dialani V, Slanetz PJ: Nonmass enhancement on breast MRI: review of patterns with radiologic-pathologic correlation and discussion of management. \u003cem\u003eAJR Am J Roentgenol \u003c/em\u003e2015, 204(1):219-227.\u003c/li\u003e\n\u003cli\u003eSantoso MR, Yang PC: Magnetic Nanoparticles for Targeting and Imaging of Stem Cells in Myocardial Infarction. \u003cem\u003eStem Cells Int \u003c/em\u003e2016, 2016:4198790.\u003c/li\u003e\n\u003cli\u003eDratwa C, Jalaguier-Coudray A, Thomassin-Piana J, Gonin J, Chopier J, Antoine M, Trop I, Darai E, Thomassin-Naggara I: Breast MR biopsy: Pathological and radiological correlation. \u003cem\u003eEur Radiol \u003c/em\u003e2016, 26(8):2510-2519.\u003c/li\u003e\n\u003cli\u003eGity M, Ghazi Moghadam K, Jalali AH, Shakiba M: Association of Different MRI BIRADS Descriptors With Malignancy in Non Mass-Like Breast Lesions. \u003cem\u003eIran Red Crescent Med J \u003c/em\u003e2014, 16(12):e26040.\u003c/li\u003e\n\u003cli\u003eSakamoto N, Tozaki M, Higa K, Tsunoda Y, Ogawa T, Abe S, Ozaki S, Sakamoto M, Tsuruhara T, Kawano N\u003cem\u003e et al\u003c/em\u003e: Categorization of non-mass-like breast lesions detected by MRI. \u003cem\u003eBreast Cancer \u003c/em\u003e2008, 15(3):241-246.\u003c/li\u003e\n\u003cli\u003ePinker K, Chin J, Melsaether AN, Morris EA, Moy L: Precision Medicine and Radiogenomics in Breast Cancer: New Approaches toward Diagnosis and Treatment. \u003cem\u003eRadiology \u003c/em\u003e2018, 287(3):732-747.\u003c/li\u003e\n\u003cli\u003eGiess CS, Yeh ED, Raza S, Birdwell RL: Background parenchymal enhancement at breast MR imaging: normal patterns, diagnostic challenges, and potential for false-positive and false-negative interpretation. \u003cem\u003eRadiographics \u003c/em\u003e2014, 34(1):234-247.\u003c/li\u003e\n\u003cli\u003eMarino MA, Pinker K, Leithner D, Sung J, Avendano D, Morris EA, Jochelson M: Contrast-Enhanced Mammography and Radiomics Analysis for Noninvasive Breast Cancer Characterization: Initial Results. \u003cem\u003eMol Imaging Biol \u003c/em\u003e2020, 22(3):780-787.\u003c/li\u003e\n\u003cli\u003eHolli K, Laaperi AL, Harrison L, Luukkaala T, Toivonen T, Ryymin P, Dastidar P, Soimakallio S, Eskola H: Characterization of breast cancer types by texture analysis of magnetic resonance images. \u003cem\u003eAcad Radiol \u003c/em\u003e2010, 17(2):135-141.\u003c/li\u003e\n\u003cli\u003eChitalia RD, Kontos D: Role of texture analysis in breast MRI as a cancer biomarker: A review. \u003cem\u003eJ Magn Reson Imaging \u003c/em\u003e2019, 49(4):927-938.\u003c/li\u003e\n\u003cli\u003eCao K, Zhao B, Li XT, Li YL, Sun YS: Texture Analysis of Dynamic Contrast-Enhanced MRI in Evaluating Pathologic Complete Response (pCR) of Mass-Like Breast Cancer after Neoadjuvant Therapy. \u003cem\u003eJ Oncol \u003c/em\u003e2019, 2019:4731532.\u003c/li\u003e\n\u003cli\u003eFan M, Wu G, Cheng H, Zhang J, Shao G, Li L: Radiomic analysis of DCE-MRI for prediction of response to neoadjuvant chemotherapy in breast cancer patients. \u003cem\u003eEur J Radiol \u003c/em\u003e2017, 94:140-147.\u003c/li\u003e\n\u003cli\u003eKajihara M, Goto M, Hirayama Y, Okunishi S, Kaoku S, Konishi E, Shinkura N: Effect of the menstrual cycle on background parenchymal enhancement in breast MR imaging. \u003cem\u003eMagn Reson Med Sci \u003c/em\u003e2013, 12(1):39-45.\u003c/li\u003e\n\u003cli\u003eWaugh SA, Purdie CA, Jordan LB, Vinnicombe S, Lerski RA, Martin P, Thompson AM: Magnetic resonance imaging texture analysis classification of primary breast cancer. \u003cem\u003eEur Radiol \u003c/em\u003e2016, 26(2):322-330.\u003c/li\u003e\n\u003cli\u003eSzczypinski PM, Strzelecki M, Materka A, Klepaczko A: MaZda--a software package for image texture analysis. \u003cem\u003eComput Methods Programs Biomed \u003c/em\u003e2009, 94(1):66-76.\u003c/li\u003e\n\u003cli\u003eTozaki M, Fukuda K: High-spatial-resolution MRI of non-masslike breast lesions: interpretation model based on BI-RADS MRI descriptors. \u003cem\u003eAJR Am J Roentgenol \u003c/em\u003e2006, 187(2):330-337.\u003c/li\u003e\n\u003cli\u003eChen QL, Luo Z, Zheng JL, Li XD, Liu CX, Zhao YH, Gong Y: Protective effects of calcium on copper toxicity in Pelteobagrus fulvidraco: copper accumulation, enzymatic activities, histology. \u003cem\u003eEcotoxicol Environ Saf \u003c/em\u003e2012, 76(2):126-134.\u003c/li\u003e\n\u003cli\u003eChikarmane SA, Michaels AY, Giess CS: Revisiting Nonmass Enhancement in Breast MRI: Analysis of Outcomes and Follow-Up Using the Updated BI-RADS Atlas. \u003cem\u003eAJR Am J Roentgenol \u003c/em\u003e2017, 209(5):1178-1184.\u003c/li\u003e\n\u003cli\u003eEl Khouli RH, Macura KJ, Jacobs MA, Khalil TH, Kamel IR, Dwyer A, Bluemke DA: Dynamic contrast-enhanced MRI of the breast: quantitative method for kinetic curve type assessment. \u003cem\u003eAJR Am J Roentgenol \u003c/em\u003e2009, 193(4):W295-300.\u003c/li\u003e\n\u003cli\u003eShao Z, Wang H, Li X, Liu P, Zhang S, Cao S: Morphological distribution and internal enhancement architecture of contrast-enhanced magnetic resonance imaging in the diagnosis of non-mass-like breast lesions: a meta-analysis. \u003cem\u003eBreast J \u003c/em\u003e2013, 19(3):259-268.\u003c/li\u003e\n\u003cli\u003eHegenscheid K, Schmidt CO, Seipel R, Laqua R, Ohlinger R, Hosten N, Puls R: Contrast enhancement kinetics of normal breast parenchyma in dynamic MR mammography: effects of menopausal status, oral contraceptives, and postmenopausal hormone therapy. \u003cem\u003eEur Radiol \u003c/em\u003e2012, 22(12):2633-2640.\u003c/li\u003e\n\u003cli\u003eDeMartini WB, Liu F, Peacock S, Eby PR, Gutierrez RL, Lehman CD: Background parenchymal enhancement on breast MRI: impact on diagnostic performance. \u003cem\u003eAJR Am J Roentgenol \u003c/em\u003e2012, 198(4):W373-380.\u003c/li\u003e\n\u003cli\u003eBrooks JD, Sung JS, Pike MC, Orlow I, Stanczyk FZ, Bernstein JL, Morris EA: MRI background parenchymal enhancement, breast density and serum hormones in postmenopausal women. \u003cem\u003eInt J Cancer \u003c/em\u003e2018, 143(4):823-830.\u003c/li\u003e\n\u003cli\u003eAmano Y, Woo J, Amano M, Yanagisawa F, Yamamoto H, Tani M: MRI Texture Analysis of Background Parenchymal Enhancement of the Breast. \u003cem\u003eBiomed Res Int \u003c/em\u003e2017, 2017:4845909.\u003c/li\u003e\n\u003cli\u003eNewell D, Nie K, Chen JH, Hsu CC, Yu HJ, Nalcioglu O, Su MY: Selection of diagnostic features on breast MRI to differentiate between malignant and benign lesions using computer-aided diagnosis: differences in lesions presenting as mass and non-mass-like enhancement. \u003cem\u003eEur Radiol \u003c/em\u003e2010, 20(4):771-781.\u003c/li\u003e\n\u003cli\u003eMai H, Mao Y, Dong T, Tan Y, Huang X, Wu S, Huang S, Zhong X, Qiu Y, Luo L\u003cem\u003e et al\u003c/em\u003e: The Utility of Texture Analysis Based on Breast Magnetic Resonance Imaging in Differentiating Phyllodes Tumors From Fibroadenomas. \u003cem\u003eFront Oncol \u003c/em\u003e2019, 9:1021.\u003c/li\u003e\n\u003cli\u003eEun NL, Kang D, Son EJ, Park JS, Youk JH, Kim JA, Gweon HM: Texture Analysis with 3.0-T MRI for Association of Response to Neoadjuvant Chemotherapy in Breast Cancer. \u003cem\u003eRadiology \u003c/em\u003e2020, 294(1):31-41.\u003c/li\u003e\n\u003cli\u003eGoto M, Sakai K, Yokota H, Kiba M, Yoshida M, Imai H, Weiland E, Yokota I, Yamada K: Diagnostic performance of initial enhancement analysis using ultra-fast dynamic contrast-enhanced MRI for breast lesions. \u003cem\u003eEur Radiol \u003c/em\u003e2019, 29(3):1164-1174.\u003c/li\u003e\n\u003cli\u003eYang X, Dong M, Li S, Chai R, Zhang Z, Li N, Zhang L: Diffusion-weighted imaging or dynamic contrast-enhanced curve: a retrospective analysis of contrast-enhanced magnetic resonance imaging-based differential diagnoses of benign and malignant breast lesions. \u003cem\u003eEur Radiol \u003c/em\u003e2020.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"breast, non-mass enhancement, texture analysis, additive value","lastPublishedDoi":"10.21203/rs.3.rs-92781/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-92781/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Non-mass enhancement (NME) is a diagnostic dilemma. Texture analysis (TA) could serve as an objective method to quantify tumor characteristics and growth patterns. However, there are few reports about TA use in NME diagnosis. To our knowledge, NME diagnosis based on the combination of the features noted on routine MRI and TA has not been reported.. The purpose of this study was to explore the value of TA in distinguishing between benign and malignant NME in premenopausal women. \u003c/p\u003e\u003cp\u003eMethods: Women in whom NME was histologically proven (n = 147) were enrolled (benign: 58; malignant: 89) was retrospective. Then, 102 and 45 patients were classified as the training and validation groups, respectively. Scanning sequences included Fat-suppressed T2-weighted and fat-suppressed contrast-enhanced T1-weighted which were acquired on a 1.5T MRI system. Clinical and routine MR characteristics (CRMC) were evaluated by two radiologists according to the Breast Imaging and Reporting and Data system (2013). Texture features were extracted from all post-contrast sequences in the training group. The combination model was built and then assessed in the validation group. Pearson’s chi-square test and Mann-Whitney U test were used to compare categorical variables and continuous variables, respectively. Logistic regression analysis and receiver operating characteristic curve were employed to assess the diagnostic performance of CRMC, TA, and their combination model in NME diagnosis.\u003c/p\u003e\u003cp\u003eResults: The combination model showed a superior diagnostic performance in differentiating between benign and malignant NME compared to that of CRMC or TA alone (AUC, 0.887 vs 0.832 vs 0.74). Moreover, compared to CRMC, the model showed high specificity (72.5% vs 80%). The results obtained in the validation group confirmed the model was promising.\u003c/p\u003e\u003cp\u003eConclusion: The combined use of TA and CRMC could afford an improved diagnostic performance in differentiating between benign and malignant NME.\u003c/p\u003e","manuscriptTitle":"Additive Value of Texture Analysis Based on Breast MRI for Distinguishing Between Benign and Malignant Non-mass Enhancement in Premenopausal Women","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2020-10-19 15:43:35","doi":"10.21203/rs.3.rs-92781/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Major revision","date":"2021-01-07T00:00:00+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2021-01-06T00:00:00+00:00","index":2,"fulltext":"Recommendation: Major revisions required\nForm responses:\n---\n\nComments to Author:\n---\nThis article is surely interesting, as it portrays a new research topic on breast imaging that could be useful for further research. I do have, however some suggestion on how to improve the text. First of all, the English needs to be deeply revised. Maybe an editor or an English speaking colleague could be useful. Other points are the followings: \n- You do not need to add \"Title\", \"Abstract\" and so on as first word.\n- \"Women in whom NME was histologically proven\": what do you mean? NME is a radiological entity, what did you prove histologically?\n- Why including T2 fat-sat sequences if they are not used for TA? In the abstract you say that all the features were extracted from the post-contrast images and in the text this sequence is barely mentioned.\n- A small note: among the keywords (that are commonly 5) you could add MRI\n- \"Recent studies have claimed that MRI feature reports have significant inter-observer variability and a lack of quantitative indicators and repeatability\" I do not totally agree with this sentence for two reasons: first of all you are only citing one work, secondly, the work you cite does not imply what you wrote, nor the BIRADS would exist if that was true.\n- \"The acquisition time for each was 3 min\": what is this sentence referred to? Each post-contrast T1? The total exam?\n- Was the T2w set acquired before or after contrast injection?\n- You should specify what you mean with \"routine MRI characteristics\" and also what information the radiologist had access to.\n- In the text it is written that you acquired 3, 4 and then 4 set of post-contrast sequences, please, be careful and only write the real number of post-contrast sequences acquired.\n- Why differentiating in the text the 2nd contrast phase from the 3rd-5th if the same exact things were done? Why is the 1st contrast phase not cited? Moreover, if the contrast variation should not impact your measurement, why choosing 5 set of post-contrast imaging?\n- Figure 2. The arrows do not seem to be really useful, there is always only one arrow even when talking about multiple findings. If you are not directly pointing at the finding, I would refrain from using them, especially if the finding of the shape and the arrow is the same.\n\n* Publons Reviewer Recognition. Springer Nature can send verification of this review directly to Publons (a subsidiary of Clarivate Analytics). If you would like to take advantage of this service, please click on the “Yes” option below. Your name, email address, title of the reviewed manuscript, name of the journal, and date of your review submission (the “Review Data”) will then be transmitted to Publons upon publication of the manuscript. If you have already registered at Publons, they will notify you of the receipt of this review and update your profile as per your settings and their policy. If you are not registered with Publons, you will receive an email from them asking you to register in order for them to be able to recognize your review on your new profile page. Publons may use the Review Data to generate derivative metadata for the benefit of Publons and you as a reviewer, carefully considering the sensitivity of such information. For example, Publons may verify your record as a reviewer by updating your profile published on its webservice if you have registered for such service or help editors to identify candidate reviewers. Please find the details of processing in Publons’ privacy policy https://publons.com/about/terms: **Yes**\n* Declaration of competing interests: **I declare that I have no competing interests**\n* Reviewer Publication Consent. I agree for my report to be made available under an Open Access Creative Commons CC-BY License (http://creativecommons.org/licenses/by/4.0) if this manuscript is accepted for publication. Any comments that I do not wish to be included in the published report have been included as confidential comments to the editor, which will not be published.: **I agree to the terms of the CC-BY 4.0 license; please do not publish my name with my report. (default)**\n* Is the study design appropriate to answer the research question (including the use of appropriate controls), and are the conclusions supported by the evidence presented?: **Yes**\n* Are the methods sufficiently described to allow the study to be repeated?: **No**\n* Is the use of statistics and treatment of uncertainties appropriate?: **No**\n* Is the presentation of the work clear?: **No**\n* Are the images in this manuscript (including electrophoretic gels and blots) free from apparent manipulation?: **Yes**\n"},{"type":"reviewerAgreed","content":"","date":"2021-01-02T00:00:00+00:00","index":2,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2020-12-10T00:00:00+00:00","index":1,"fulltext":"Recommendation: Accept without revision\nForm responses:\n---\n\nComments to Author:\n---\nIt is a well prepared paper. I only have several minor comments.\n\n1. P7. Lines 47-52.\nCould the author add references for the Mazda software?\n\n2. P10. Line 12-20\nPlease explain the reasons of the difference.\n\n3. P29. Figure 1\nI would suggest to increase the font size to make the figure readable.* Publons Reviewer Recognition. Springer Nature can send verification of this review directly to Publons (a subsidiary of Clarivate Analytics). If you would like to take advantage of this service, please click on the “Yes” option below. Your name, email address, title of the reviewed manuscript, name of the journal, and date of your review submission (the “Review Data”) will then be transmitted to Publons upon publication of the manuscript. If you have already registered at Publons, they will notify you of the receipt of this review and update your profile as per your settings and their policy. If you are not registered with Publons, you will receive an email from them asking you to register in order for them to be able to recognize your review on your new profile page. Publons may use the Review Data to generate derivative metadata for the benefit of Publons and you as a reviewer, carefully considering the sensitivity of such information. For example, Publons may verify your record as a reviewer by updating your profile published on its webservice if you have registered for such service or help editors to identify candidate reviewers. Please find the details of processing in Publons’ privacy policy https://publons.com/about/terms: **Yes**\n* Declaration of competing interests: **'I declare that I have no competing interests.**\n* Reviewer Publication Consent. I agree for my report to be made available under an Open Access Creative Commons CC-BY License (http://creativecommons.org/licenses/by/4.0) if this manuscript is accepted for publication. Any comments that I do not wish to be included in the published report have been included as confidential comments to the editor, which will not be published.: **I agree to the terms of the CC-BY 4.0 license; please do not publish my name with my report. (default)**\n* Is the study design appropriate to answer the research question (including the use of appropriate controls), and are the conclusions supported by the evidence presented?: **Yes**\n* Are the methods sufficiently described to allow the study to be repeated?: **Yes**\n* Is the use of statistics and treatment of uncertainties appropriate?: **Yes**\n* Is the presentation of the work clear?: **Yes**\n* Are the images in this manuscript (including electrophoretic gels and blots) free from apparent manipulation?: **Yes**\n"},{"type":"reviewerAgreed","content":"","date":"2020-11-18T00:00:00+00:00","index":1,"fulltext":""},{"type":"editorAssigned","content":"","date":"2020-10-15T12:00:00+00:00","index":"","fulltext":""},{"type":"reviewersInvited","content":"","date":"2020-10-15T12:00:00+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2020-10-14T12:00:00+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2020-10-14T12:00:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"","date":"2020-10-09T12:00:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0d6da699-c5ce-4813-839d-856993c56ef4","owner":[],"postedDate":"October 19th, 2020","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":814943,"name":"Nuclear Medicine \u0026 Medical Imaging"}],"tags":[],"updatedAt":"2021-03-14T15:00:43+00:00","versionOfRecord":{"articleIdentity":"rs-92781","link":"https://doi.org/10.1186/s12880-021-00571-x","journal":{"identity":"bmc-medical-imaging","isVorOnly":false,"title":"BMC Medical Imaging"},"publishedOn":"2021-03-12 15:00:29","publishedOnDateReadable":"March 12th, 2021"},"versionCreatedAt":"2020-10-19 15:43:35","video":"","vorDoi":"10.1186/s12880-021-00571-x","vorDoiUrl":"https://doi.org/10.1186/s12880-021-00571-x","workflowStages":[]},"version":"v1","identity":"rs-92781","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-92781","identity":"rs-92781","version":["v1"]},"buildId":"-D5TCW68w8eVRRLyjaTIo","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}