Prediction of clinical outcome for high-intensity focused ultrasound ablation of adenomyosis based on non-enhanced MRI radiomics

Objectives:  The study aimed to develop a non-enhanced MRI-based radiomics model for the preoperative prediction of the efficacy of adenomyosis after high-intensity focused ultrasound (HIFU) treatment.

The data of 130 patients with adenomyosis who underwent HIFU treatment were reviewed. Based on a non-perfused volume ratio (NPVR) of 50%, the patients were assigned to high ablation rate and low ablation rate groups. A radiomics model was constructed from the screened radiomics features and its output probability was calculated as the radiomics score (Radscore). The clinical-imaging model was constructed from the independent predictors of clinical-imaging characteristics. The combined model was constructed by integrating Radscore and clinical-imaging independent predictors. Receiver operating characteristic (ROC) curves, the Delong test, and decision curve analysis (DCA) were used to evaluate the models.

The combined model had the best overall performance among the three models. The AUC (95% CI), specificity, sensitivity, accuracy, and precision of the combined model were 0.860 (0.786–0.935), 0.780, 0.756, 0.769, 0.738 in the training set, and 0.878 (0.774–0.983), 0.859, 0.667, 0.769, 0.800 in the test set, respectively. The Delong test showed that the performance of both the radiomics and combined models differed significantly from the clinical-imaging model. But the performance of the combined and the radiomics model was statistically equivalent. The DCA indicated that the combined model had better clinical net benefit.

The combined model based on non-enhanced MRI radiomics was effective in predicting the outcome of HIFU ablation of adenomyosis before surgery. 1.  Introduction Adenomyosis is a common benign gynecological disease in which the endometrial glands and mesenchyme invade and grow into the myometrium of the uterus [ 1,2]. It is primarily characterized by the following symptoms: dysmenorrhea, abnormal menstrual cycles, increased vaginal discharge, infer - tility, and abnormal menstrual flow, which significantly affect the patient’s quality of life [ 3,4]. Traditional treatments for adenomyosis, such as medication or surgery, have limited effectiveness due to the disease’s chronic estrogen-dependent nature and as the lesions tend to be poorly differentiated from the normal myometrium or are diffusely distributed, resulting in a high recurrence rate [ 3,5]. Hysterectomy is a radical treatment; however, it is not the best choice for women who want to keep their uterus or are trying to con - ceive [ 6]. Therefore, high-intensity focused ultrasound (HIFU), a noninvasive ablation procedure, is often used to treat ade - nomyosis [ 7–9]. Its benefits include uterine preservation, safety, efficacy, and minimal side effects [ 7,10]. However, due to individual differences and variations in the lesion tissue, not all adenomyosis treatments are equally effective. Currently, the most popular way to measure the short-term effective - ness of HIFU is the non-perfused volume ratio (NPVR) [ 11,12]. The NPVR is associated with volume reduction and symptom relief after treatment [ 13–15]. The accurate prediction of the NPVR after HIFU treatment is therefore important for the selection of suitable patients, cost-saving, and the develop - ment of treatment plans. MRI is a significant tool for the diagnosis and evaluation of adenomyosis [16,17]. Several studies have found that imaging features such as T2WI signal intensity, the number of T2 high-signal lesions, and the kind of T1WI enhancement can be used to predict the NPVR when using HIFU [ 6,18,19]. However, traditional clinical-imaging features have shown limited predic- tive power (AUC = 0.720) [ 19]. Moreover, the visual interpreta - tion of medical images is dependent on the experience of the observer and the results thus lack objectivity. Radiomics can extract quantitative image features that are not apparent to human eyes to establish relevant predictive models [ 20,21]. Most previous radiomics-based studies [2,22] on the prediction of HIFU efficacy in treating adenomyosis have focused on single © 2025 t he a uthor(s). p ublished with license by taylor & f rancis Group, ll C CONTACT Xiaohua Huang [email protected] Department of r adiology, a ffiliated Hospital of north Sichuan m edical College, no. 1, m aoyuan South r oad, Shunqing District, nanchong 637000, China. *t hese authors have contributed equally to this work. https://doi.org/10.1080/02656736.2025.2468766 t his is an o pen a ccess article distributed under the terms of the Creative Commons a ttribution license ( http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. t he terms on which this article has been published allow the posting of the a ccepted manuscript in a repository by the author(s) or with their consent. ARTICLE HISTORY r eceived 15 o ctober 2024 r evised 27 January 2025 a ccepted 13 f ebruary 2025

magnetic resonance imaging; radiomics; adenomyosis; high-intensity focused ultrasound; prediction 2 Z. LIU ET AL. T2WI images, and did not consider the differences and comple - mentarity between different sequences. Although CE-T1WI images can provide blood perfusion data, the injection of gad - olinium contrast agents not only increases the financial burden on the patient, but may also have some adverse effects. DWI-based ADC maps can provide an effective reflection of the degree of diffusion of tissue water molecules [23], which can be used as an effective complement to T2WI sequences. Therefore, this study aimed to construct a radiomics model based on non-enhanced images (ADC and T2WI images) to pre- dict the NPVR in the HIFU treatment of adenomyosis. Furthermore, this study combined radiomics and clinical-imaging features to improve the predictive performance. 2.  Materials and methods 2.1.  Patients Retrospective data were collected from 336 patients who under- went HIFU for adenomyosis at the Affiliated Hospital of North Sichuan Medical College between September 2021 and December 2023. The inclusion criteria were: (1) women experi - encing symptoms of adenomyosis; (2) no prior history of surgery or medication therapy related to the condition; (3) an MRI scan conducted no more than three days before and after ablation treatment; (4) a diagnosis of adenomyosis confirmed by clinical and radiological examinations. The exclusion criteria were: (1) insufficient or absent imaging data; (2) the presence of other gynecological disorders, such as pelvic inflammatory disease or uterine fibroids; (3) poor image quality affecting the drawing of the target region; (4) pregnancy or breastfeeding; (5) below the age of 18 years. Figure 1 shows that 130 patients who met the inclusion and exclusion criteria for the research were included in the cohort. The patients’ ages ranged from 45 to 48 years. 2.2.  MRI scanning protocol The uMR790 3.0 T, 12-channel body phased array coil from United Imaging was used to conduct the pelvic MRI scans. To minimize artifacts caused by respiratory motion, the patient was given breathing training before the test and a bandage was used to compress the abdomen. GD-DTPA was injected at a flow rate of 1.0 ml/s, with a dosage of 0.1 mmol/kg used to achieve enhanced scanning. The arteries were imaged at 15, 30, and 45 s after injection of the GD-DTPA, representing the early, middle, and late time points, respectively. The b-values, diffusion sensitivity coefficients, were set at 50 and 800 s/mm2. Table 1 provides details of the scanning order and primary parameters. 2.3.  Patient grouping When analyzing adenomyosis lesions that were found to be generally regular, the sagittal CE-T1WI and T2WI images were used to assess the postoperative ablation volume and adeno - myosis volume, respectively. The ellipsoid formula (0.5233 × longitudinal diameter × anteroposterior diameter × transverse diameter) [ 24] was used to calculate the non-perfused vol - ume (NPV) and adenomyosis volume (V) of the lesion. For diffuse adenomyosis or lesions with unclear boundaries, man - ual layer-by-layer delineation of the region of interest (ROI) was performed, with automatic determination of V and NPV using the 3D-slicer software. We estimated the ablation rate as NPVR, which is equal to NPV/V*100%. Using NPVR values of 50% as the threshold [25], the patients were allocated to high ablation rate (NPVR ≥ 50%, n = 59) and low ablation rate (NPVR< 50%, n = 71) groups. The groups were then randomly divided into training and test sets with a ratio of 7:3. The training set included 50 cases with low ablation and 41 cases with high ablation, while the test set comprised 21 cases with low ablation and 18 cases with high ablation. 2.4.  Clinical–imaging features Clinical-imaging features that were likely to affect the NPVR in adenomyosis were assessed. These included age, adeno - myosis volume, type of adenomyosis (diffuse/focus), location of adenomyosis (anterior, posterior, anterior and posterior), location of uterus (anteverted, retroverted), the distance from Figure 1. f low chart of patient recruitment. INTERNATIONAL JOURNAL OF HYPERTHERMIA 3 the anterior side of adenomyosis to skin, abdominal wall thickness, T2 signal intensity (hypointensity: lesions with lower signal intensity than the normal myometrium, isointen - sity: lesions with signal intensity similar to that of the normal myometrium), the number of hyperintense foci on T2WI (few or multiple hyperintense foci: the number of hyperintense points was ≤ or >5 on a single slice, respectively). Laboratory test data were collected, including leukocyte count, red blood cell count, hemoglobin, and platelet count. HIFU treat - ment parameters were collected, including treatment power, treatment time and energy. Univariate logistic regression was used to analyze the clinical-imaging characteristics to identify significant factors (p < 0.05). Multivariate logistic regression analysis was then performed to identify the independent predictors associated with NPVR after HIFU treatment. 2.5.  Radiomics feature extraction and ROI segmentation Two radiologists with 10 years’ experience in the diagnosis of gynecological disease and who were blinded to the specifics of the cases manually segmented the ROIs of the two sequences (ADC and T2WI), layer by layer using a 3D Slicer (version 5.6.1), making sure to include all of the adenomyosis layers. The 3D volume of interest (VOI) of adenomyosis was generated by fuzing the ROIs of the different image layers. For diffuse adenomyosis involving the entire uterine wall, the edge of the ROI was maintained at a specific distance from the endometrial and plasma layers to avoid the involvement of the normal tissue. The original images were processed by Laplacian Gaussian filtering and wavelet transform filtering, and the radiomics features of each VOI were extracted using the built-in radiomics plugin. To guarantee the reproducibility of the radiomics features, one-third of the ADC and T2WI images and outlined lesions were re-analyzed. The consistency between the observers for the extracted features was then assessed by computing the intergroup correlation coefficient (ICC). Only characteristics that demonstrated strong consistency and stability (ICC > 0.75) were retained, while the rest were discarded. 2.6.  Radiomics feature selection The following three steps were used for feature selection using R (version 4.3.3) and uAI Research Portal (version 730): (1) The feature data were preprocessed using Z ⁃score normal - ization to eliminate the dimensional effects of different fea - tures, (2) Features with variances below 0.8 were discarded using variance thresholding, after which select K Best was used to remove features that did not show substantial differ - ences between the two groups; (3) Least absolute shrinkage and selection operator (LASSO) was used to select the most relevant features for analysis, which further reduced the dimensionality. 2.7.  Model establishment Three models were constructed using logistic regression to predict the NPVR after HIFU treatment. The radiomics model was constructed from the screened radiomics features and its output probability was calculated as the radiomics score (Radscore). The clinical-imaging model was constructed from clinical-imaging features that had been found to be inde - pendently predictive. The combined model was developed by integration of the Radscore and the independently predic - tive clinical-imaging features. 2.8.  Evaluation of model performance The receiver operating characteristic (ROC) curve and the area under the curve (AUC), specificity, sensitivity, accuracy, and precision, were used to evaluate the predictive efficacy of the different models. Differences between the models were compared using the Delong test. The clinical benefits of each model were evaluated using decision curve analysis (DCA). To assess how well the model matched the data, cali - bration curves and the Hosmer-Lemeshow test were used. Figure 2 illustrates the workflow of the radiomics analysis. 2.9.  Statistical analysis A statistical significance level of p < 0.05 was employed for the analysis, which was conducted using SPSS (version 27.0) and R (version 4.3.3) software. The distribution of the quanti - tative data was determined using the Shapiro-Wilk test. Data that followed a normal distribution are presented as ( xs± ), whereas data that followed a skewed distribution are given as M (Q25, Q75). The independent samples t-test was used for comparing normally distributed data, while the Mann- Whitney U test was used for analyzing non-normally distrib - uted data. Qualitative data were analyzed using either Fisher’s exact test or the chi-square test. 3.  Results 3.1.  Clinical–imaging features The training and test sets were compared based on their clinical-imaging characteristics ( Table 2 ). Hemoglobin exhib - ited a statistically significant difference ( p = 0.041), whereas other characteristics did not ( p > 0.05). Univariate logistic regression analysis indicated that loca - tion of adenomyosis, the number of hyperintense foci on T2WI and the distance from the anterior side of adenomyosis to skin differed significantly between the high ablation rate group and the low ablation rate group. Following this, multi - variate logistic regression analysis determined the number of hyperintense foci on T2WI (OR = 0.337, p = 0.005) and the distance from the anterior side of adenomyosis to skin (OR = 0.981, p = 0.037) were independent predictors of NPVR (Table 3). Table 1. m agnetic resonance sequences and parameters. parameters t2WI-fS Ce-t1WI DWI tr (ms) 3300 4.23 2863 te (ms) 88.40 1.72 80 t hickness (mm) 4 5 4 Spacing (mm) 2 0 2 foV (mm) 260 × 260 350 × 300 280 × 260 matrix 256 × 256 304 × 75 128 × 100 note: tr: r epetition time; te: e cho time; foV: f ield of view. 4 Z. LIU ET AL. 3.2.  Radiomics features Of the overall 2446 radiomics features extracted from the ADC and T2WI images, 1909 features (752 from ADC and 1157 from T2WI) were retained according to the results of the ICC test (ICC > 0.75). Ultimately, 11 radiomics features, including 5 from ADC images and 6 from T2WI images, were retained after the three screening steps. Figure 2. f lowchart of radiomics. Table 2. Comparison of clinical and imaging features between training and test sets. training Set test Set p value a ge (years) 43(39, 47) 43(36, 46) 0.326 Volume (cm 3) 64.97(38.47, 112.27) 62.80(36.00, 108.35) 0.657 a bdominal wall thickness (mm) 29.48 ± 7.65 27.88 ± 7.31 0.271 Distance from the anterior side of adenomyosis to skin (mm) 56.00(42.61, 78.77) 67.89(49.55, 73.89) 0.435 l eukocyte (10 9·l −1 ) 5.97 ± 1.61 5.64 ± 1.35 0.263 r ed blood (10 12·l −1 ) 4.36(4.02, 4.72) 4.27(4.01, 4.58) 0.162 Hemoglobin (g·l −1 ) 115(95, 130) 103(85, 124) 0.041* platelet (10 9·l −1 ) 262(210, 320) 252(192, 327) 0.419 type of adenomyosis (n) 0.567 Diffuse 28 14 f ocus 63 25 l ocation of the uterus (n) 0.292 a nteverted 77 30 r etroverted 14 9 l ocation of adenomyosis (n) 0.890 a nterior 25 11 p osterior 48 19 a nterior and p osterior 18 9 t2 signal intensity (n) 0.661 Hypointensity 57 26 Isointensity 34 13 t he number of hyperintense foci on t2WI (n) 0.788 ≤5 42 17 >5 49 22 treatment power (W) 400(400,400) 400(400,400) 0.933 treatment time (s) 614(374,905) 627(405,828) 0.847 energy (kJ) 361.20(245.60,466.80) 331.20(250.80,399.20) 0.863 *p value represents the according parameter is of statistical significance. Table 3. univariate and multivariate logistic regression analyses between high and low ablation rate groups. univariate multivariate or (95%CI) p or (95%CI) p age 1.027 (0.967, 1.090) 0.386 —— —— Volume 1.000 (1.000, 1.000) 0.112 —— —— a bdominal wall thickness 0.959 (0.915, 1.006) 0.087 —— —— Distance from the anterior side of adenomyosis to skin 0.980 (0.963, 0.997) 0.018* 0.981 (0.963, 0.999) 0.037* l eukocyte 0.841 (0.666, 1.061) 0.144 —— —— r ed blood 0.601 (0.297, 1.213) 0.155 —— —— Hemoglobin 1.001 (0.997, 1.005) 0.608 —— —— platelet 0.995 (0.979, 1.011) 0.510 —— —— t ype of adenomyosis 0.860 (0.410, 1.804) 0.689 —— —— l ocation of uterus 1.708 (0.668, 4.363) 0.264 —— —— l ocation of adenomyosis 0.350 (0.139, 0.883) 0.026* 2.285 (0.756, 6.904) 0.143 t2 signal intensity 1.573 (0.759, 3.259) 0.223 —— —— t he number of hyperintense foci on t2WI 2.502 (1.231, 5.089) 0.011* 0.337 (0.156, 0.726) 0.005* treatment power 1.017 (0.955, 1.039) 0.126 —— —— treatment time 1.000 (0.999, 1.001) 0.889 —— —— energy 1.000 (0.998, 1.003) 0.813 —— —— *p value represents the according parameter is of statistical significance. INTERNATIONAL JOURNAL OF HYPERTHERMIA 5 3.3.  Evaluation of model performance The results of the ROC curve analysis of the three models are shown in Figure 3 and Table 4 . The AUC of clinical- imaging model, radiomics model, and combined model were 0.692, 0.838, and 0.860 in the training set, and 0.646, 0.868, and 0.878 in the test set, respectively. The combined model showed the best overall performance among the three models. Its AUC (95% CI), specificity, sensitivity, accuracy, and precision were 0.860 (0.786–0.935), 0.780, 0.756, 0.769, and 0.738 in the training set, and 0.878 (0.774–0.983), 0.859, 0.667, 0.769, and 0.800 in the test set. The results of the Delong test showed that the perfor - mance of clinical-imaging model was significantly lower than that of radiomics model ((training set: p = 0.048, test set: p = 0.047)) and combined model (training set: p = 0.004, test set: p = 0.001). There was no significant difference between the radiomics model and the combined model (training set: p = 0.253, test set: p = 0.760). According to the DCA ( Figure 4A ), for the majority of the threshold probabilities, the combined model provided greater clinical net benefit in predicting the NPVR of HIFU treatment. The results of the Hosmer-Lemeshow test and the calibration curves demonstrated that the combined model was well-corrected (0.152 for the training set, 0.147 for the test set), as shown in Figure 4B . 4.  Discussion In this study, we constructed the clinical-imaging model, radiomics model and combined model, and we found that the combined model that integrated the Radscore from the non-enhanced MRI model along with the independent pre - dictors used in the clinical-imaging model had better predic - tive efficacy and clinical value. The AUC (95% CI), specificity, sensitivity, accuracy, and precision of the combined model were 0.860 (0.786–0.935), 0.780, 0.756, 0.769, 0.738 in the training set, and 0.878 (0.774–0.983), 0.859, 0.667, 0.769, 0.800 in the test set, respectively. This method will help clini - cians to predict the NPVR of HIFU treatment before surgery and screen patients suitable for HIFU treatment. Radiomics provides large amounts of information from medical images and can identify heterogeneity in the spatial distribution of lesions [ 26–28]. Radiomics-based MRI has been used in studies on adenomyosis to predict the long and short-term efficacy of HIFU and identify adenomyosis [1,2,22,29]. Li et  al. [ 2] utilized T2WI-based radiomics to predict the long-term outcome of HIFU treatment in adenomyosis. The AUC, specificity, sensitivity, and accuracy of their radiomics-clinical model in the test set were 0.81, 0.71, 0.86, and 0.76, respectively. However, the sample size of their study was small, with only 69 cases, and the model only incorpo - rated 4 radiomics features, which may have increased the risk of model overfitting and instability. Ying et  al. [ 22] constructed Figure 3. roC curves for clinical-imaging model, radiomics model, and combined model in ( a ) the training and (B) test sets. Table 4. p erformance comparison of three models in training and test sets. auC (95% CI) Specificity Sensitivity a ccuracy precision Clinical-imaging model training 0.692(0.583–0.802) 0.820 0.439 0.648 0.667 test 0.646(0.468–0.823) 0.857 0.278 0.590 0.625 r adiomics model training 0.838(0.751–0.924) 0.840 0.683 0.769 0.778 test 0.868(0.756–0.980) 0.762 0.722 0.744 0.722 Combined model training 0.860(0.786–0.935) 0.780 0.756 0.769 0.738 test 0.878(0.774–0.983) 0.859 0.667 0.769 0.800 6 Z. LIU ET AL. a model based on T2WI radiomics and deep learning to pre - dict adenomyosis lesion ablation by HIFU treatment. The AUC, accuracy, precision, recall, and F-score in the test set of their model were 0.861, 0.814, 0.832, 0.795, and 0.813, respectively. However, these studies focused only on radiomics features from a single T2WI image and did not address the variability and complementarity between different sequences. ADC images obtained by the post-processing of DWI images with different b-values can provide an accurate reflec - tion of the true diffusion properties of tissues by eliminating the influence of the T2 transmission effect, presenting infor - mation on cells and the microcirculation [ 28,30,31]. Therefore, radiomics features that incorporate both T2WI and ADC sequences can more comprehensively and accurately reflect the pathophysiology of adenomyosis, overcoming the limita - tion of the single sequence that provides only limited data on the lesion. In this study, 11 features (6 from T2WI images and 5 from ADC images) were used to construct the radiom - ics model. The results of this study suggest that both T2WI and ADC images are indispensable in predicting the NPVR for adenomyosis treatment with HIFU. This study also analyzed the relationship between clinical- imaging characteristics and NPVR. The number of hyperin - tense foci on T2WI and the distance from the anterior side of adenomyosis to skin were found to be independent predic - tors of NPVR in adenomyosis. A greater distance from the ventral side of the adenomyosis to the skin and the presence of multiple hyperintense foci were not conducive to effective HIFU ablation in patients with adenomyosis, which is consis - tent with the findings of previous research [ 6,32]. By combin - ing the independent clinical-imaging predictors and Radscore, increased the overall performance and clinical benefit of the combined model relative to those of the single radiomics and single clinical-imaging models. Consistent with the

of recent studies, the predictive performance of the model improved when radiomics was combined with clinical-imaging features [ 2,33]. This study has several limitations. Firstly, due to the limited availability of data, external validation was not performed. Further research is necessary to acquire external validation data to confirm the model’s viability. Secondly, the study was retrospective with a small sample size and is thus potentially subject to bias; further studies with larger sample sizes are needed for verification of the findings. Finally, the ADC image might not be as high-quality as the T2WI image. Even though the outlining process used combined T2WI images to deter - mine the ROI and the ICC test excluded features with poor sta - bility and consistency, there might still be some outlining errors. 5.  Conclusion The findings showed that a combined model based on non-enhanced MRI radiomics was effective in predicting the efficacy of HIFU ablation of adenomyosis before surgery. This model will assist clinicians in treatment decisions and the screening of patients who are likely to benefit from HIFU. Author contributions Ziyi Liu and Ziyan Liu contributed equally to this work. (1) Conception and design: Ziyi Liu, Ziyan Liu and Xiaohua Huang. (2) Collection and assembly of data: Xiyao Wan and Yuan Wang. (3) Data analysis: Ziyi Liu and Ziyan Liu. (4) Manuscript writing: Ziyi Liu. (5) Final approval of man - uscript: All authors. Ethical approval The Affiliated Hospital of North Sichuan Medical College’s Medical Ethics Committee eliminated the need for informed consent after approving the research plan (IRB no.2024ER282-1). Disclosure statement No potential conflict of interest was reported by the author(s). Funding This study has received funding by Nanchong City School Cooperation Project [No. 19SXHZ0429] and This work was supported by Bureau of Science and Technology Nanchong Municipality. Figure 4. (a ) Decision curves of clinical-imaging model, radiomics model, and combined model; (B) Combined model calibration curves in the training set and test set. INTERNATIONAL JOURNAL OF HYPERTHERMIA 7 ORCID Xiaohua Huang http://orcid.org/0000-0002-3490-4142 Data availability statement The data used to support the findings of this study are available from the corresponding author upon request. The data are not publicly avail - able because they contain information that can compromise the privacy of the research participants.

[ 1] Jin W, Wang S, Wang T, et  al. Multi-machine learning model based on habitat subregions for outcome prediction in adenomyosis treated by uterine artery embolization. Acad Radiol. 2024;31(12): 4985–4995. doi: 10.1016/j.acra.2024.05.037. [ 2] Li Z, Zhang J, Song Y, et  al. Utilization of radiomics to predict long-term outcome of magnetic resonance-guided focused ultra - sound ablation therapy in adenomyosis. Eur Radiol. 2021;31(1):392– 402. doi: 10.1007/s00330-020-07076-1. [ 3] Bae JS, Lee JY, Chung HH, et  al. Optimized treatment parameter by computer simulation for high-intensity focused ultrasound treat - ment of uterine adenomyosis: short-term and long-term results. PLoS One. 2024;19(3):e0301193. doi: 10.1371/journal.pone.0301193. [ 4] Loring M, Chen TY, Isaacson KB. A systematic review of adenomyo - sis: it is time to reassess what we thought we knew about the disease. J Minim Invasive Gynecol. 2021;28(3):644–655. doi: 10.1016/ j.jmig.2020.10.012. [ 5] Tan J, Yong P , Bedaiwy MA. A critical review of recent advances in the diagnosis, classification, and management of uterine adenomy - osis. Curr Opin Obstet Gynecol. 2019;31(4):212–221. doi: 10.1097/ gco.0000000000000555. [ 6] Du CC, Wang YQ, Qu DC, et  al. Magnetic resonance imaging T2WI hyperintense foci number and the prognosis of adenomyosis after high-intensity focused ultrasound treatment. Int J Gynaecol Obstet. 2021;154(2):241–247. doi: 10.1002/ijgo.13587. [ 7] Lee JS, Hong GY, Lee KH, et  al. Safety and efficacy of ultrasound-guided high-intensity focused ultrasound treatment for uterine fibroids and adenomyosis. Ultrasound Med Biol. 2019;45(12): 3214–3221. doi: 10.1016/j.ultrasmedbio.2019.08.022. [ 8] Liu Y, Zhang WW, He M, et  al. Adverse effect analysis of high-intensity focused ultrasound in the treatment of benign uter - ine diseases. Int J Hyperthermia. 2018;35(1):56–61. doi: 10.1080/ 02656736.2018.1473894. [ 9] Yao R, Hu J, Zhao W, et  al. A review of high-intensity focused ultrasound as a novel and non-invasive interventional radiology technique. J Interv Med. 2022;5(3):127–132. doi: 10.1016/j.jimed. 2022.06.004 . [10] Guo Q, Xu F, Ding Z, et  al. High intensity focused ultrasound treat - ment of adenomyosis: a comparative study. Int J Hyperthermia. 2018;35(1):505–509. doi: 10.1080/02656736.2018.1509238. [11] Mindjuk I, Trumm CG, Herzog P , et  al. MRI predictors of clinical success in MR-guided focused ultrasound (MRgFUS) treatments of uterine fibroids: results from a single centre. Eur Radiol. 2015;25(5):1317–1328. doi: 10.1007/s00330-014-3538-6. [12] Si M, Lv F, Tang M, et  al. Non-contrast enhanced MRI for efficiency evaluation of high-intensity focused ultrasound in adenomyosis ablation. Int J Hyperthermia. 2024;41(1):2295813. doi: 10.1080/ 02656736.2023.2295813. [13] Fan TY, Zhang L, Chen W, et  al. Feasibility of MRI-guided high in - tensity focused ultrasound treatment for adenomyosis. Eur J Radiol. 2012;81(11):3624–3630. doi: 10.1016/j.ejrad.2011.05.036. [14] Lee JS, Hong GY, Park BJ, et  al. Ultrasound-guided high-intensity focused ultrasound treatment for uterine fibroid & adenomyosis: a single center experience from the Republic of Korea. Ultrason Sonochem. 2015;27:682–687. doi: 10.1016/j.ultsonch.2015.05.033. [15] Shui L, Mao S, Wu Q, et  al. High-intensity focused ultrasound (HIFU) for adenomyosis: two-year follow-up results. Ultrason Sonochem. 2015;27:677–681. doi: 10.1016/j.ultsonch.2015.05.024. [16] Bazot M, Daraï E. Role of transvaginal sonography and magnetic resonance imaging in the diagnosis of uterine adenomyosis. Fertil Steril. 2018;109(3):389–397. doi: 10.1016/j.fertnstert.2018.01.024. [17] Sudderuddin S, Helbren E, Telesca M, et  al. MRI appearances of be - nign uterine disease. Clin Radiol. 2014;69(11):1095–1104. doi: 10.1016/j.crad.2014.05.108. [18] Gong C, Setzen R, Liu Z, et  al. High intensity focused ultrasound treatment of adenomyosis: the relationship between the features of magnetic resonance imaging on T2 weighted images and the therapeutic efficacy. Eur J Radiol. 2017;89:117–122. doi: 10.1016/j. ejrad.2017.02.001. [19] Yu JW, Yang MJ, Jiang L, et  al. Factors influencing USgHIFU abla - tion for adenomyosis with NPVR ≥ 50. Int J Hyperthermia. 2023;40(1):2211753. doi: 10.1080/02656736.2023.2211753. [20] Lafata KJ, Wang Y, Konkel B, et  al. Radiomics: a primer on high-throughput image phenotyping. Abdom Radiol (NY). 2022;47(9):2986–3002. doi: 10.1007/s00261-021-03254-x. [21] Mayerhoefer ME, Materka A, Langs G, et  al. Introduction to radio - mics. J Nucl Med. 2020;61(4):488–495. doi: 10.2967/jnumed. 118.222893. [22] Ying J, Jing X, Gao F, et  al. Prediction of ablation rate for high-intensity focused ultrasound therapy of adenomyosis in MR images based on multi-model fusion. J Imaging Inform Med. 2024;37(4):1579–1590. doi: 10.1007/s10278-024-01063-4. [23] Jha RC, Zanello PA, Ascher SM, et  al. Diffusion-weighted imaging (DWI) of adenomyosis and fibroids of the uterus. Abdom Imaging. 2014;39(3):562–569. doi: 10.1007/s00261-014-0095-z. [ 24] Keserci B, Duc NM. Magnetic resonance imaging features influencing high-intensity focused ultrasound ablation of adenomyosis with a non- perfused volume ratio of ≥90% as a measure of clinical treatment suc- cess: retrospective multivariate analysis. Int J Hyperthermia. 2018;35(1): 626–636. doi: 10.1080/02656736.2018.1516301. [25] Zhou M, Chen JY, Tang LD, et  al. Ultrasound-guided high-intensity focused ultrasound ablation for adenomyosis: the clinical experi - ence of a single center. Fertil Steril. 2011;95(3):900–905. doi: 10.1016/j.fertnstert.2010.10.020. [26] Lambin P , Leijenaar RTH, Deist TM, et  al. Radiomics: the bridge be - tween medical imaging and personalized medicine. Nat Rev Clin Oncol. 2017;14(12):749–762. doi: 10.1038/nrclinonc.2017.141. [27] Rizzo S, Botta F, Raimondi S, et  al. Radiomics: the facts and the challenges of image analysis. Eur Radiol Exp. 2018;2(1):36. doi: 10.1186/s41747-018-0068-z. [28] Zhou J, Yu X, Wu Q, et  al. Radiomics analysis of intratumoral and different peritumoral regions from multiparametric MRI for evaluat - ing HER2 status of breast cancer: a comparative study. Heliyon. 2024;10(7):e28722. doi: 10.1016/j.heliyon.2024.e28722. [29] Burla L, Sartoretti E, Mannil M, et  al. MRI-based radiomics as a promising noninvasive diagnostic technique for adenomyosis. J Clin Med. 2024;13(8):2344. doi: 10.3390/jcm13082344. [30] Luo Y, Sun X, Kong X, et  al. A DWI-based radiomics-clinical machine learning model to preoperatively predict the futile recanalization af - ter endovascular treatment of acute basilar artery occlusion patients. Eur J Radiol. 2023;161(2023):110731. doi: 10.1016/j.ejrad.2023.110731. [31] Shih IL, Yen RF, Chen CA, et  al. PET/MRI in cervical cancer: associ - ations between imaging biomarkers and tumor stage, disease pro - gression, and overall survival. J Magn Reson Imaging. 2021;53(1):305–318. doi: 10.1002/jmri.27311. [ 32] Gong C, Yang B, Shi Y, et  al. Factors influencing the ablative efficiency of high intensity focused ultrasound (HIFU) treatment for adenomyo- sis: a retrospective study. Int J Hyperthermia. 2016;32(5):496–503. doi: 10.3109/02656736.2016.1149232. [33] Qin S, Jiang Y, Wang F, et  al. Development and validation of a combined model based on dual-sequence MRI radiomics for predicting the efficacy of high-intensity focused ultrasound abla - tion for hysteromyoma. Int J Hyperthermia. 2023;40(1):2149862. doi: 10.1080/02656736.2022.2149862.