{"paper_id":"ec8adea6-7848-4201-b807-53ee37345695","body_text":"973\nCopyright © 2025 The Korean Society of Radiology\nQuantitative Time-Dependent Diffusion MRI for Diagnosis \nand Aggressiveness Assessment of Endometrial Cancer: \nA Prospective Study\nWenyi Yue1,2, Ruxue Han3, Junzhong Xu4,5, Chaoyang Jin4,5, Xiaoyu Jiang4,5, Dandan Zheng6, Jing Peng6, \nJun Lu7, Qiming Liu1,2, Ning Xu8, Dan Zhao9, Hua Li3, Qi Yang1,2\n1Department of Radiology, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China \n2Laboratory for Clinical Medicine, Capital Medical University, Beijing, China \n3Department of Gynecology and Obstetrics, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China \n4Institute of Imaging Science, Vanderbilt University Medical Center, Nashville, TN, USA \n5Department of Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, TN, USA \n6Clinical & Technique Support, Philips Healthcare, Beijing, China \n7Department of Pathology, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China \n8Department of Pathology, No. 984 Hospital of Chinese PLA Logistical Support Force, Beijing, China \n9Department of Gynecology Oncology, National Cancer Center, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China\nObjective: Preoperative differentiation of benign and malignant endometrial lesions, along with the identification of \naggressive histological types of endometrial cancer (EC), is crucial for guiding treatment strategies. Time-dependent diffusion \nmagnetic resonance imaging (TDD-MRI), which allows the characterization of tissue microstructure at the cellular level, is \nnot currently applied for endometrial lesions. This study aimed to evaluate TDD-MRI-derived microstructural parameters for \nnoninvasively distinguishing benign and malignant endometrial lesions and predicting aggressive histological types of EC.\nMaterials and Methods: This prospective study enrolled 177 patients with clinically suspected EC who underwent TDD-MRI \nbetween January 2024 and March 2025. The Imaging Microstructural Parameters Using Limited Spectrally Edited Diffusion \nmethod was used to extract microstructural parameters, including the cell diameter (d), intracellular volume fraction (v\nin), \ncellularity (number of cells per unit area), cellularity index (vin/d), and extracellular diffusivity (Dex), along with three apparent \ndiffusion coefficient measurements. The area under the receiver operating characteristic curve (AUC) was used to assess \ndiagnostic performance. The Pearson correlation coefficient between the microstructural parameters and histopathological \nmeasurements was calculated.\nResults: A total of 130 women (mean ± standard deviation age: 56 ± 14 years) administered uterine curettage or surgery \nwere included in the final analysis. All microstructural parameters showed significant differences between benign \nendometrial lesions and EC (P < 0.05), as well as between nonaggressive and aggressive EC (P < 0.05). Cellularity exhibited \nthe highest AUC of 0.86 for distinguishing benign endometrial lesions from EC, whereas the cellularity index showed the \nhighest AUC of 0.88 for distinguishing aggressive histological types. D\n0Hz was positively correlated with Dex (P < 0.05) and \nnegatively correlated with diameter ( P < 0.05), cellularity index ( P < 0.01) and vin (P < 0.001) in patients with benign \nendometrial lesions. D0Hz was positively correlated with Dex (P < 0.001) and negatively correlated with vin (P < 0.001) in \npatients with EC. Microstructural parameters strongly correlated with corresponding pathological features ( r = 0.77–0.83; \nP < 0.001).\nKorean J Radiol 2025;26(10):973-985\neISSN 2005-8330\nhttps://doi.org/10.3348/kjr.2025.0633\nOriginal Article | Oncologic Imaging\nReceived: May 19, 2025   Revised: July 1, 2025   Accepted: July 31, 2025\nCorresponding author: Qi Yang, MD, PhD, Department of Radiology, Beijing Chaoyang Hospital, Capital Medical University, No. 8 Gongti \nSouth Road, Chaoyang District, Beijing 100020, China\n• E-mail: yangyangqiqi@gmail.com\nCorresponding author: Hua Li, MD, PhD, Department of Gynecology and Obstetrics, Beijing Chaoyang Hospital, Capital Medical \nUniversity, No. 8 Gongti South Road, Chaoyang District, Beijing 100020, China\n• E-mail: hual_gyn@163.com\nThis is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://\ncreativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, \nprovided the original work is properly cited. \n\n974\nYue et al.\nhttps://doi.org/10.3348/kjr.2025.0633 kjronline.org\nand biophysical models to characterize tissues at the \ncellular level. Notably, Imaging Microstructural Parameters \nUsing Limited Spectrally Edited Diffusion (IMPULSED) is \nclinically feasible, enabling cellular microstructure mapping \nin just 5–7 minutes [11]. Such novel microstructural \ninformation has been clinically applied for prostate [12], \nbreast [11,13,14], brain [15], and ovarian [16] cancers. \nPrevious evidence has shown the capability of TDD-MRI \nin differentiating benign and malignant tumors as well as \npathological grading [12,13,15-17]. However, it remains \nunclear whether TDD-MRI can characterize endometrial \ntissues, noninvasively differentiate benign and malignant \nlesions, and classify aggressive histological subtypes of EC.\nTherefore, this study aimed to evaluate the diagnostic \nfeasibility of TDD-MRI in endometrial disease, comparing \nit with conventional DWI measurements for differentiating \nbenign and malignant endometrial lesions and identifying \naggressive histological types of EC. \nMATERIALS AND METHODS\nStudy Participants\nFollowing the Declaration of Helsinki, this prospective \nstudy was approved by the Institutional Review Board of \nBeijing Chaoyang hospital of Capital Medical University \n(IRB No. 2023-11-13-1) and registered with the Chinese \nClinical Trial Registry (ChiCTR2500095674). Totally, 177 \npatients with clinically suspected EC were enrolled to \nundergo MRI between January 2024 and March 2025. \nThe inclusion criteria were as follows: 1) scheduled \nconventional contrast-enhanced pelvic MRI and TDD-MRI, \nand 2) willingness and ability to undergo MRI and provide \ninformed consent. As the enrolled patients progressed \nthrough the study procedure, some were excluded from the \nfinal analysis based on the following criteria: 1) lack of \npathological confirmation, 2) prior treatment for endometrial \ndisease before MRI, 3) Insufficient MRI quality, and 4) final \nhistology showing atypical endometrial hyperplasia. The \nparticipant flowchart is shown in Figure 1.\nINTRODUCTION\nEndometrial cancer (EC) is a common gynecological \nmalignancy with an increasing incidence and a younger \ndemographic [1]. Abnormal uterine bleeding (AUB) is a \ncommon symptom of EC. However, benign endometrial \nlesions (e.g., endometrial polyps and hyperplasia without \natypia) may also cause AUB [2]. Benign endometrial lesions \nare often managed with diagnostic curettage or conservative \ntreatment, whereas EC typically requires hysterectomy. \nThe International Federation of Gynecology and Obstetrics \n(FIGO) staging of EC identifies aggressive histological types \n[3]. Therefore, preoperative differentiation of benign and \nmalignant endometrial lesions, along with the identification \nof aggressive EC histological types, is crucial for guiding \nsubsequent patient treatment.\nEvidence suggests significant differences between the \nmicroenvironments of benign endometrium and EC [4,5]. In \nEC, the number of epithelial and endometrial stromal cells \nincreases and decreases, respectively. Additionally, changes \nin immune cell populations are observed, with a reduced \nproportion of cytotoxic and naïve CD8 lymphocytes, and an \nincreased proportion of CD4+ T regulatory cells in EC [6]. \nThis highlights the significant cellular differences between \nendometrial diseases, emphasizing the value of cell-level \ndifferentiation for diagnosis. Currently, no noninvasive \ntechnique can detect these diseases at the cellular level \nbefore pathological confirmation. \nDiffusion-weighted imaging (DWI) is a noninvasive \nimaging technique that offers functional insights \nand enhances the morphological details provided by \nconventional magnetic resonance imaging (MRI) [6-8]. \nDWI allows the quantification of diffusion through the \napparent diffusion coefficient (ADC), which reflects the \nphysiological characteristics of tissue microcirculation [9]. \nHowever, DWI cannot provide microstructural parameters, \nsuch as intra- and extracellular space, cell size, and \npermeability [10]. Recent advancements in time-dependent \ndiffusion (TDD)-MRI have demonstrated its unique ability \nto depict cellular microstructures. This diffusion MRI-based \ntechnique leverages multi-diffusion times, multi-b values, \nConclusion: TDD-MRI-derived microstructural parameters demonstrated high performance in differentiating benign from \nmalignant endometrial diseases and identifying aggressive types of EC.\nKeywords: Magnetic resonance imaging; Time-dependent diffusion MRI; Endometrial cancer; Histological types; \nMicrostructural parameters\n\n975\nTime-Dependent Diffusion MRI of Endometrial Cancer\nhttps://doi.org/10.3348/kjr.2025.0633kjronline.org\nImage Acquisition\nThe IMPULSED strategy was used for diffusion MRI by \napplying oscillating gradient spin-echo (OGSE) and pulsed \ngradient spin-echo (PGSE) sequences [11]. Scanning was \nperformed using a Philips 3T Elition MR scanner (Philips \nHealthcare, Best, the Netherlands) with an external \npelvic phased-array coil. OGSE data were acquired at \noscillating frequencies of 33 Hz (effective diffusion time, \n7.5 ms; 2 cycles; b = 0, 80, 160, 250 s/mm\n2) and 17 Hz \n(effective diffusion time, 15.0 ms; 1 cycle; b = 0, 200, \n400, 600, 800 s/mm\n2). PGSE data were acquired with a \ndiffusion duration/separation = 60/82.3 ms at b-value \nof 500/1,000/1,500 s/mm\n2. The PGSE sequence used \nrepresents conventional DWI, with the diffusivity value \nderived from this acquisition (at 0 Hz) corresponding to \nthe standard ADC commonly used in clinical diagnosis \n[12,14]. Both sequences used the following parameters: \nrepetition time/echo time, 3,000/143 ms; field of view, \n220 mm x 220 mm; voxel reconstruction size, 2.80 x \n2.80 x 6.00 mm\n3; single-shot echo planar imaging; half \nscan factor, 0.684; water-fat shift (pixels)/bandwidth \n(Hz), 14.610 pixel/29.7 Hz; fat suppression with spectral \nadiabatic inversion recovery. Dynamic stabilization was \nused to minimize the DWI signal drifts. The scanning time \nfor TDD-MRI was approximately 9 minutes 8 seconds and its \nprotocol is summarized in Figure 2.\nImage Analysis\nThe image analysis method used has been described \npreviously [18]. Briefly, diffusion signals are modeled as \narising from two distinct compartments so that S = v\nin x Sin \n+ (1 - vin ) x Sex, where vin denotes the water volume fraction \nof the intracellular space, and Sin and Sex are intracellular \nand extracellular diffusion MRI signals, respectively. Cancer \ncells were modeled as impermeable spheres so that Sin \ncan be described using analytical expressions and the cell \ndiameter (d) can be estimated to represent the mean cell \nsize [18,19]. Extracellular diffusivity (D\nex) was assumed to \nbe hindered diffusion time so that Sex = exp (-b x Dex) [20]. \nMRI-derived cellularity (number of cells per unit area) was \ncalculated as 2 x (3vin\n2π )\n2\n3 ⁄d\n2\n [19]. A previous study defined \nan unconventional “cellularity” as vin⁄(d x 100), with \nsome clinical potential [12]. We defined the latter as the \ncellularity index to avoid any confusion with conventional \ncellularity. Additionally, ADC maps were obtained at each \ndiffusion time according to S⁄S\n0 = exp(-bD), where D is the \ndiffusivity, using b = 250 s/mm2 for the 33-Hz OGSE data, b = \n800 s/mm2 for the 17-Hz OGSE data, and b = 1,200 s/mm2 \nfor the PGSE data to obtain diffusivity at 33 Hz (D33Hz), 17 \nHz (D17Hz), and 0 Hz (D0Hz), respectively.\nDiffusion images were coregistered to the corresponding \nT2-weighted S (b = 0) images to correct for subject motion. \nDenoising employs a local principal component analysis \nfilter [21]. Data fitting employed MATLAB R2024b (The \nMathWorks) to generate DWI parametric maps on a voxel-\nwise basis. Microstructural parameters (d, v\nin, and Dex) were \nfitted using the MATI package [22]. The fitting was repeated \n100 times per sample, and the analysis with the smallest \nfitting residual was chosen as the final result. \nIn patients with endometrial lesions, radiologist 1 (with \nWomen with clinical suspicion of EC were enrolled to undergo MRI, including \ntime-dependent diffusion MRI, between January 2024 and March 2025 (n = 177)\n Excluded\n    •   N o pathology confirmation (n = 22) \n    •   A typical endometrial hyperplasia (n = 13)\n    •   History o f prior treatment for endometrial \ndisease before MRI (n = 3)\n    •   In sufficient MRI quality (n = 9)\n Benign lesions (n = 68)\n    •\n   U terine curettage (n = 48)\n    •   Sur gery (n = 20)\n EC (n = 62)\n    •\n   Sur gery (n = 62)\n130 patients with complete time-dependent diffusion MRI sequence were \nincluded in the final analysis\nFig. 1. Flowchart shows participant enrollment. EC = endometrial cancer\n\n976\nYue et al.\nhttps://doi.org/10.3348/kjr.2025.0633 kjronline.org\n10 years of experience in gynecological imaging) manually \ndelineated the regions of interest (ROIs) on each slice of the \nlesions using PGSE DWI images with a b-value of 1,200 s/mm², \ncarefully excluding the surrounding tissues. To enhance \nthe representativeness and reduce potential measurement \nbias, the final value for each patient was calculated as the \naverage across all lesion-containing slices. To further improve \nrobustness, all delineations were independently resegmented \nby radiologist 2 (with 8 years of experience). The values used \nin the main analysis were the averages of the measurements \ntaken by the two radiologists. Interobserver agreement \nbetween the two radiologists was assessed using intraclass \ncorrelation coefficients (ICCs) and Bland–Altman plots.\nHistopathological Analysis\nBiopsy specimens and histopathological slides were \nreviewed under the supervision of two experienced \ngynecological oncology pathologists with 15 and 10 years of \nexperience. Benign endometrial lesions included endometrial \npolyps and hyperplasia. The pathological states of EC were \nrecorded. According to the 2023 FIGO guidelines, low-\ngrade (G1–G2) endometrioid endometrial carcinoma (EEC) \nis considered nonaggressive, whereas high-grade tumors, \nincluding EEC G3, serous carcinoma, clear cell carcinoma, \nand other rare subtypes, are classified as aggressive \n[3]. In our cohort, only EEC were present, with no non-\nendometrioid histological subtypes. Therefore, in accordance \nwith the updated FIGO classification system, we defined EEC \nG3 as aggressive EC and EEC G1–G2 as nonaggressive EC.\nA\nB\nFig. 2. Schematic shows the pulse sequences and differentiation of benign endometrial lesions and endometrial cancer using time-\ndependent diffusion magnetic resonance imaging-based microstructural mapping. A: The diagram presented illustrates the pulse \nsequences utilized for imaging microstructural parameters using a limited spectrally edited diffusion method. In addition to \nconventional PGSE ACQ, OGSE ACQ at two frequencies (n = 1 and 2) were employed. Diffusion signals, which are dependent on diffusion \ntime, can be captured using both pulsed and OGSE diffusion encoding schemes across various diffusion times. B: The diffusivity of \nwater molecules in a cellular environment is influenced by diffusion time, with this effect becoming more noticeable as cellular density \nincreases. By employing pulsed and OGSE diffusion encoding schemes at varying diffusion times, diffusion signals can be captured. These \nsignals enable the reconstruction of microstructural properties using biophysical modeling approaches. PGSE = pulsed gradient spin-echo, \nACQ = acquisitions, OGSE = oscillating gradient spin-echo, td = diffusion time, NK = natural killer\n\n977\nTime-Dependent Diffusion MRI of Endometrial Cancer\nhttps://doi.org/10.3348/kjr.2025.0633kjronline.org\nHistopathological Assessment and Correlation With MRI\nLight microscopy histology images were analyzed using a deep \nlearning-based cell pose method for cellular segmentation [23]. \nAfter all individual nuclei were identified, the histology-derived \ncellularity (the total number of nuclei divided by the total area \nof the section) was calculated. Fnuclei was calculated as the area \nfraction occupied by the nuclei. For each segmented nucleus, \nan effective diameter was calculated using d = \n(A/π) * 2, \nwhere A is the nuclear area. The mean nuclear diameter was \ncalculated for each histological section to better correlate \nwith the corresponding TDD-MRI-derived mean cell diameter. \nStatistical Analysis\nContinuous variables with normal distribution were \ncompared using the two-sample t-test, and categorical \nvariables using Pearson χ2 test and Fisher’s exact test. \nThe differences in microstructural parameters between \nbenign endometrial lesions and EC, as well as between ECs \nwith different aggressiveness levels, were assessed using \na two-sample t-test. Additionally, the Pearson correlation \ncoefficient was used to assess the relationship between the \nmicrostructural parameters and histological indices. The \ninterobserver agreement for the microstructural parameters \nwas assessed using ICCs, with ICCs >0.75 indicating good \nagreement. Bland–Altman analysis was used to demonstrate \nthe reproducibility of measurements of the microstructural \nparameters. Receiver operating characteristic (ROC) curve \nanalysis was used to assess the diagnostic performance. To \naddress the challenges posed by the small sample size and \nrarity of aggressive EC, we employed a Bayesian logistic \nregression model. This approach allows the incorporation of \nprior information and yields more stable estimates in the \npresence of limited data and rare events [24]. The model \nperformance was evaluated using standard classification \nmetrics, and the robustness of these performance measures \nwas assessed by bootstrapping with 200 resampled datasets. \nWe used the Youden index to determine the optimal threshold \nfor classifying cases. The corresponding area under the ROC \ncurve (AUC), sensitivity, specificity, positive predictive value, \nand negative predictive value were also calculated. Statistical \nanalyses were performed using SPSS 26.0 (IBM Corp., Armonk, \nNY, USA), GraphPad Prism 8.0 (San Diego, CA, USA), MedCalc \n22.0 (Oostende, Belgium) and R 4.1.0 (The R Project for \nStatistical Computing, a global open-source project). P < 0.05 \nindicated statistical significance.\nRESULTS\nBaseline Characteristics\nThe final analysis included 130 women administered \nTDD-MRI followed by uterine curettage or surgery for \nwhom complete clinical information was available. Their \nmean age was 56 ± 14 years, with an interquartile range \nof 45–66 years. Among all participants, 68 (52.3%) were \npathologically diagnosed with benign endometrial lesions \nand 62 (47.7%) were diagnosed with EC and underwent \nstaging surgery. Among these 62 patients, EC in 50 (80.6%) \nwas identified as nonaggressive, whereas in 12 (19.4%), \nit was aggressive. The baseline characteristics of the \nparticipants are summarized in Table 1. \nInterobserver Agreement\nThe ICCs for diameter, D\nex, cellularity, cellularity index, vin, D0Hz, \nD17Hz and D33Hz were 0.93 (95% confidence interval [CI]: 0.90, \n0.95), 0.96 (95% CI: 0.94, 0.97), 0.96 (95% CI: 0.94, 0.97), \n0.96 (95% CI: 0.94, 0.97), 0.97 (95% CI: 0.96, 0.98), 0.94 (95% \nCI: 0.91, 0.96), 0.94 (95% CI: 0.92, 0.96), and 0.94 (95% CI: \n0.91, 0.95) respectively. The Bland–Altman analysis showed \ngood reproducibility in the measurement of microstructural \nparameters taken by the two radiologists (Fig. 3).\nTable 1. Patient baseline characteristics\nClinical parameters Benign endometrial \nlesions (n = 68)\nEC \n(n = 62)\nNon-aggressive EC \n(n = 50)\nAggressive EC \n(n = 12) P* P†\nAge, yr   52.96 ± 14.60   59.15 ± 11.97   58.46 ± 11.77   62.00 ± 12.90 0.010 0.362\nBMI, kg/m2 25.16 ± 4.02 27.22 ± 5.25 27.59 ± 5.32 25.66 ± 4.82 0.013 0.256\nMenopausal status, yes 32 (47.1) 43 (69.4) 35 (70.0)   8 (66.7) 0.010 0.822\nAbnormal uterine bleeding, yes 29 (42.6) 48 (77.4) 38 (76.0) 10 (83.3) 0 0.585\nLive birth, yes 53 (77.9) 54 (87.1) 44 (88.0) 10 (83.3) 0.172 0.665\nDiabetes, yes 12 (17.6) 17 (27.4) 13 (26.0)   4 (33.3) 0.181 0.609\nData are presented as mean ± standard deviation or patient number with percentage in parentheses. \n*P-value for benign endometrial lesions and EC, †P-value for non-aggressive and aggressive EC.\nEC = endometrial cancer, BMI = body mass index\n\n978\nYue et al.\nhttps://doi.org/10.3348/kjr.2025.0633 kjronline.org\nComparing TDD-MRI-Derived Microstructural Parameters \nBetween Different Endometrial Pathologies\nFigure 4 shows the T2-weighted images, TDD-MRI \nmicrostructural parameter maps, and diffusivity maps at \ndifferent oscillating frequencies of benign endometrial \nlesions, nonaggressive EC, and aggressive EC. In terms of \nthe comparison between benign endometrial lesions and EC, \nthe microstructural parameters (diameter, D\nex, cellularity, \nFig. 3. Bland–Altman plots compare the reproducibility of time-dependent diffusion magnetic resonance imaging measurements from two \nindependent readers. A: Plot shows that, for diameter, the mean difference is -0.08 μm (95% CI: -0.91, 0.75). B: Plot shows that, for Dex, \nthe mean difference is 0.02 μm2/ms (95% CI: -0.13, 0.18). C: Plot shows that, for cellularity, the mean difference is 0.04 (95% CI: -0.36, \n0.44). D: Plot shows that, for the cellularity index, the mean difference is -0.00 (95% CI: -0.35, 0.34). E: Plot shows that, for vin, the \nmean difference is 0 μm2/ms (95% CI: -0.04, 0.04). F: Plot shows that, for D0Hz, the mean difference is 0.01 μm2/ms (95% CI: -0.23, 0.25). \nG: Plot shows that, for D17Hz, the mean difference is 0.03 μm2/ms (95% CI: -0.23, 0.29). H: Plot shows that, for D33Hz, the mean difference \nis -0.02 μm2/ms (95% CI: -0.27, 0.23). The dotted line represents the mean difference, and the upper and lower solid lines represent the \nupper and lower limits of agreement, respectively. CI = confidence interval, Dex = extracellular diffusivity, vin = intracellular volume fraction, \nSD = standard deviation\nFig. 4. Microstructural parameter maps of benign lesions and nonaggressive and aggressive endometrial cancer, including diameter, Dex, \nvin, cellularity, cellularity index and the diffusivity maps from pulsed gradient spin-echo (D0Hz) and oscillating gradient spin-echo (D17Hz \nand D33Hz) data. Corresponding T2W images at the similar axial locations are shown in the first column. Dex = extracellular diffusivity, \nvin, = intracellular volume fraction, T2W = T2-weighted, IMPULSED = Imaging Microstructural Parameters Using Limited Spectrally Edited \nDiffusion, ADC = apparent diffusion coefficient\nA\nE\nB\nF\nC\nG\nD\nH\n\n\n979\nTime-Dependent Diffusion MRI of Endometrial Cancer\nhttps://doi.org/10.3348/kjr.2025.0633kjronline.org\ncellularity index, and vin) showed significant differences \n(P < 0.001) (Fig. 5A-E). When comparing nonaggressive \nand aggressive EC, the diameter, Dex, cellularity, cellularity \nindex, and vin showed significant differences (P < 0.05) \n(Fig. 5I-M). In benign endometrial lesions, Dex, cellularity, \nand cellularity index were significantly higher than those \nin EC, whereas diameter and vin were significantly lower. In \nnonaggressive EC, Dex, cellularity, and cellularity index were \nsignificantly higher than those in aggressive EC, whereas \nthe diameter and v\nin were significantly lower. D0Hz, D17Hz, \nand D33Hz showed significant differences between benign \nendometrial lesions and EC (P < 0.05) (Fig. 5F-H), as well \nas between nonaggressive and aggressive EC (P < 0.05) (Fig. \n5N-P). In benign endometrial lesions, D0Hz, D17Hz, and D33Hz \nwere significantly higher than those in EC. In nonaggressive \nEC, D\n0Hz, D17Hz and D33Hz were significantly higher than those \nin aggressive EC. \nDiagnostic Performance of TDD-MRI-Derived \nMicrostructural Parameters\nIn distinguishing between benign endometrial lesions \nand EC, cellularity achieved the highest performance among \nall microstructural features, with an AUC of 0.86 (95% CI: \n0.79, 0.93), sensitivity of 74.2% (46 of 62 participants) and \nspecificity of 91.2% (62 of 68 participants), followed by \nv\nin, with an AUC of 0.83 (95% CI: 0.75, 0.90), sensitivity of \nFig. 5. Box and whisker plots show comparisons of microstructural parameters, including (A) diameter, (B) Dex, (C) cellularity, (D) cellularity \nindex, (E) vin, (F) D0Hz, (G) at D17Hz, and (H) at D33Hz among benign lesions and EC, (I) diameter, (J) Dex, (K) cellularity, (L) cellularity \nindex, (M) vin, (N) at D0Hz, (O) at D17Hz, and (P) at D33Hz among nonaggressive EC and aggressive EC. *P < 0.05, †P < 0.01, ‡P < 0.001. Dots \nrepresent individual data points, boxes indicate the standard deviation, and midlines are the median. Dex = extracellular diffusivity, vin = \nintracellular volume fraction, EC = endometrial cancer\nA\nE\nI\nM\nB\nF\nJ\nN\nC\nG\nK\nO\nD\nH\nL\nP\n‡\n‡\n‡‡\n‡\n‡\n‡\n‡\n‡\n†\n† †\n†\n\n980\nYue et al.\nhttps://doi.org/10.3348/kjr.2025.0633 kjronline.org\n93.5% (58 of 62 participants) and specificity of 66.2% (45 \nof 68 participants) (Table 2).\nIn distinguishing between nonaggressive and aggressive \nEC, the classification analysis revealed that, among all \nmicrostructural features, the cellularity index achieved \nthe highest performance, with an AUC of 0.88 (95% CI: \n0.78, 0.95), sensitivity of 100.0% (12 of 12 participants) \nand specificity of 70.0% (35 of 50 participants), followed \nby diameter, with an AUC of 0.84 (95% CI: 0.70, 0.93), \nsensitivity of 91.7% (11 of 12 participants) and specificity \nof 66.0% (33 of 50 participants) (Table 3).\nCorrelation Between TDD-MRI-Derived Microstructural \nParameters and D\n0Hz\nFigure 6 shows the correlations between the fitted \nmicrostructural parameters and conventionally used D0Hz at \nthe participant level. D0Hz was positively correlated with Dex \n(P < 0.05) and negatively correlated with diameter (P < 0.05), \ncellularity index (P < 0.01) and vin (P < 0.001) in benign \nendometrial lesions. D0Hz was positively correlated with Dex \n(P < 0.001) and negatively correlated with vin (P < 0.001) \nin patients with EC. D0Hz was positively correlated with Dex \n(P < 0.05) and negatively correlated with vin (P < 0.001) in \nnonaggressive EC. D0Hz was positively correlated with Dex \n(P < 0.05) and negatively correlated with cellularity ( P < \n0.01) in aggressive EC. \nCorrelation and Simulation Validation With \nHistopathological Findings \nCell nuclei were automatically segmented from \nhematoxylin-eosin-stained whole-slide images (Fig. 7A-C). \nHistopathological correlation analysis was conducted in a \nrepresentative subset of 16 patients who underwent surgery, \nincluding 6 with benign endometrial lesions and 10 with \nEC. Surgical specimens were specifically chosen to ensure \nconsistent tissue sampling and high-quality hematoxylin-\neosin-stained slides, as they offer more standardized \nand comprehensive histological material, compared with \ncurettage samples. Among the patients, the EC in six was \nclassified as aggressive subtypes. (v\nin, r = 0.768; diameter, \nr = 0.768; cellularity, r = 0.832; all P < 0.001) (Fig. 7D-F).\nDISCUSSION\nThis study prospectively assessed patients with \nendometrial diseases to demonstrate the clinical application \nof microstructural parameters obtained using TDD-MRI for \ndifferentiating benign and malignant endometrial diseases \nand classifying aggressiveness of EC. These microstructural \nparameters demonstrated excellent predictive performance \nin differentiating endometrial diseases, with cellularity \nachieving the best performance, with an AUC of 0.86. For \ndistinguishing EC aggressiveness levels, the cellularity \nTable 2. Diagnostic performance of time-dependent diffusion MRI–derived microstructural parameters for distinguishing benign and \ncancerous endometrial lesions\nCharacteristic AUC Cutoff Sensitivity, % Specificity, % PPV, % NPV, %\nCellularity, cells x 10-3/μm2  0.86 [0.79, 0.93] 0.55 74.2 (46/62)\n[63.3, 85.1]\n91.2 (62/68)\n[84.4, 97.9]\n88.5 (46/52)\n[79.8, 97.1]\n79.5 (62/78)\n[70.5, 88.4]\nVin 0.83 [0.75, 0.90] 0.38 93.5 (58/62) \n[89.8, 100]\n66.2 (45/68)\n[54.9, 77.4]\n71.6 (58/81)\n[62.2, 81.7]\n91.8 (45/49)\n[86.9, 100]\nDiameter, μm 0.81 [0.74, 0.88] 0.31 96.8 (60/62)\n[92.4, 100]\n58.8 (40/68) \n[47.1, 70.5]\n68.2 (60/88)\n[58.5, 77.9]\n95.2 (40/42) \n[88.8, 100]\nDex, μm2/msec 0.74 [0.65, 0.82] 0.42 85.5 (53/62)\n[76.7, 94.3]\n60.3 (41/68)\n[48.7, 71.9]\n66.2 (53/80)\n[55.9, 76.6]\n82.0 (41/50)\n[71.4, 92.6]\nDiffusivity at 0 Hz, μm2/msec 0.71 [0.62, 0.80] 0.48 71.0 (44/62)\n[59.7, 82.3]\n69.1 (47/68)\n[58.1, 80.1]\n67.7 (44/65)\n[56.3, 79.1]\n72.3 (47/65)\n[61.4, 83.2]\nCellularity index, μm-1 0.70 [0.60, 0.79] 0.43 95.2 (59/62) \n[89.8, 100]\n57.4 (39/68)\n[45.6, 69.1]\n67.0 (59/88)\n[57.2, 76.9]\n92.9 (39/42)\n[85.1, 100]\nDiffusivity at 17 Hz, μm2/msec 0.67 [0.58, 0.77] 0.48 67.7 (42/62)\n[56.1, 79.4]\n69.1 (47/68)\n[58.1, 80.1]\n66.7 (42/63)\n[55.0, 78.3]\n70.1 (47/67)\n[59.2, 81.1]\nDiffusivity at 33 Hz, μm2/msec 0.62 [0.53, 0.72] 0.47 61.3 (38/62)\n[49.2, 73.4]\n61.8 (42/68)\n[50.2, 73.3]\n59.4 (38/64)\n[47.3, 71.4]\n63.6 (42/66)\n[52.0, 75.2]\nThe values in parentheses are numerators and denominators, and the values in brackets are 95% confidence intervals.\nAUC = area under the receiver operating characteristic curve, PPV = positive predictive value, NPV = negative predictive value, vin = \nintracellular volume fraction, Dex = extracellular diffusivity\n\n981\nTime-Dependent Diffusion MRI of Endometrial Cancer\nhttps://doi.org/10.3348/kjr.2025.0633kjronline.org\nindex achieved the best performance, with an AUC of \n0.88. The microstructural parameters derived from TDD-\nMRI showed strong correlation with pathology-determined \nmicrostructural properties (r = 0.77–0.83; P < 0.001).\nAccurate, noninvasive preoperative diagnosis of \nendometrial lesions and assessment of aggressiveness \nof EC are crucial for determining appropriate clinical \nmanagement. Although benign and malignant endometrial \nconditions often have similar symptoms, their treatment \nstrategies differ significantly. Benign lesions may not \nrequire surgical intervention, whereas malignant lesions \ntypically necessitate hysterectomy with bilateral salpingo-\noophorectomy and lymph node assessment. Although \nendometrial sampling via hysteroscopy or curettage remains \nthe standard method for histopathological confirmation, its \ndiagnostic accuracy can be limited by the lesion location, \nsampling limitations, and/or operator experience. Moreover, \nin patients with cervical stenosis, poor surgical tolerance, or \nTable 3. Diagnostic performance of time-dependent diffusion MRI–derived microstructural parameters in distinguishing non-aggressive and \naggressive endometrial cancers\nCharacteristic AUC (95% CI) Cutoff (95% CI) Sensitivity, % Specificity, % PPV, % NPV, %\nCellularity index, μm-1 0.88 [0.78, 0.95] 0.17 100 (12/12) \n[76.9, 100]\n70.0 (35/50)\n[58.0, 96.2]\n44.4 (12/27)\n[28.6, 85.8]\n100 (35/35)\n[94.0, 100]\nDiameter, μm 0.84 [0.70, 0.93] 0.11 91.7 (11/12)\n[60.0, 100]\n66.0 (33/50)\n[44.6, 96.2]\n39.3 (11/28)\n[23.1, 84.6]\n97.1 (33/34)\n[89.1, 100]\nDiffusivity at 33 Hz, μm2/msec 0.79 [0.65, 0.90] 0.11 100 (12/12)\n[54.5, 100]\n50.0 (25/50)\n[40.9, 98.1]\n32.4 (12/37)\n[21.4, 87.6]\n100 (25/25)\n[88.0, 100]\nVin 0.79 [0.62, 0.93] 0.16 83.3 (10/12)\n[57.1, 100]\n82.0 (41/50)\n[74.4, 94.4]\n52.6 (10/19)\n[33.3, 75.0]\n95.3 (41/43)\n[88.6, 100]\nDiffusivity at 0 Hz, μm2/msec 0.76 [0.56, 0.92] 0.18 83.3 (10/12)\n[38.4, 100]\n62.0 (31/50)\n[45.3, 100]\n34.5 (10/29)\n[21.6, 100]\n93.9 (31/33)\n[83.0, 100]\nDex, μm2/msec 0.74 [0.50, 0.92] 0.15 83.3 (10/12)\n[44.4, 100]\n56.0 (28/50)\n[28.5, 100]\n31.2 (10/32)\n[19.0, 100]\n93.3 (28/30)\n[83.0, 100]\nDiffusivity at 17 Hz, μm2/msec 0.66 [0.40, 0.85] 0.16 75.0 (9/12) \n[30.7, 100]\n54.0 (27/50)\n[24.4, 100]\n28.1 (9/32)\n[17.6, 100]\n90.0 (27/30)\n[81.0, 100]\nCellularity, cells x 10-3/μm2  0.61 [0.46, 0.86] 0.36 [0.09, 0.53] 50.0 (6/12)\n[36.3, 83.4]\n100 (50/50)\n[65.8, 100]\n100 (6/6)\n[20.0, 100]\n89.3 (50/56)\n[81.1, 96.4]\nAll diagnostic performance metrics (AUC, sensitivity, specificity, PPV, and NPV) are estimated using posterior mean from Bayesian logistic \nregression model with weakly informative prior N (0, 2.52). 95% CIs were calculated using Bootstrapping with 200 resampled datasets. \nThe values in parentheses are numerators and denominators, and the values in brackets are 95% CIs.\nAUC = area under the receiver operating characteristic curve, PPV = positive predictive value, NPV = negative predictive value, CI = \nconfidence interval, v\nin = intracellular volume fraction, Dex = extracellular diffusivity\nFig. 6. Correlations between the fitted microstructural parameters and D0Hz at the participant level (A-E) among benign lesions and EC \nand (F-J) among nonaggressive and aggressive EC. *P < 0.05, †P < 0.01, ‡P < 0.001. EC = endometrial cancer\nA\nF\nB\nG\nC\nH\nD\nI\nE\nJ\n‡\n†\n†\n‡\n‡\n†\n\n982\nYue et al.\nhttps://doi.org/10.3348/kjr.2025.0633 kjronline.org\nother anatomical constraints, these invasive procedures may \nbe contraindicated or pose significant risk [25]. By enhancing \ndiagnostic accuracy in challenging or indeterminate cases, \nMRI reduces the need for unnecessary biopsies or repeat \nsampling. A more accurate, noninvasive assessment of \nlesion type and tumor aggressiveness could inform clinical \ndecisions regarding whether to proceed with surgery, consider \na more conservative approach, or explore fertility-preserving \nstrategies for younger patients. Ultimately, the integration of \nMRI into the diagnostic workflow may improve preoperative \nrisk stratification, support personalized treatment planning, \navoid overtreatment, and reduce patient burden.\nMulti-parametric MRI, especially DWI and ADC, is \nindispensable in oncology. However, ADC only measures \nwater diffusivity, which is determined by various \nmicrostructural features [26-28]. ADC accuracy is affected \nby intracellular and extracellular diffusion coefficients, \nintracellular volume fraction, capillary perfusion, cell \nmembrane permeability, and cell size [29]. As changes in \ntumor cell size reflect both disease status and therapeutic \nresponse, the ability to noninvasively assess tissue \nmicrostructure is essential for monitoring tumor behavior \nand treatment outcomes. Previous studies have explored \nthe ability of functional molecular MRI and advanced \ndiffusion MRI techniques to differentiate the histological \ninformation of endometrial diseases and EC [30-32]. \nHowever, some parameters in the aforementioned studies \nexhibited relatively limited predictive performance. TDD-\nMRI, particularly with OGSE acquisitions to broaden the \ndiffusion time range, provides a unique means to gain \nsensitivity to different length scales [10], enabling the \ncharacterization of different microstructures in tumors \nand clinical feasibility to differentiate tumor malignancy \nand grades [33-35]. There is increasing interest in TDD-\nMRI, which uses multi-b values, multi-diffusion times, and \nmulti-compartmental biophysical models to quantitatively \ncharacterize microstructural information at the cellular \nlevel. Therefore, IMPULSED has been rapidly implemented \nin several clinical cancers, and the derived microstructural \nfeatures have promising potential for clinical applications \nsuch as differentiating between clinically significant and \ninsignificant prostate cancers [12,13,15] and predicting \nthe treatment response in breast cancer [14]. Compared \nwith previous studies, our study implemented TDD-MRI \nto differentiate endometrial diseases and EC and further \nclassify aggressiveness of EC. Additionally, we conducted \na consistency test on the microstructural parameters from \nthe ROI measurements taken by two radiologists, and the \nICC was >0.90, which demonstrated good stability of the \nmeasurements. \nA\nD\nB\nE\nC\nF\nFig. 7. Correlations between time-dependent diffusion MRI-derived microstructural parameters and pathology-based microstructural \nfeatures (n = 16). A: Hematoxylin-eosin-stained image (x40) shows pathological specimen from one participant with benign lesion. \nB: Hematoxylin-eosin-stained image (x40) illustrates nuclei segmented by a pretrained conditional generative adversarial network. \nC: Automated quantification of the pathological microstructural features. D-F: The graph depicts the correlations between time-\ndependent diffusion MRI-derived (D) diameter, (E) cellularity, (F) v\nin, and the pathology-based microstructural features. fnuclei = nuclei \nfraction, vin = intracellular volume fraction\n\n983\nTime-Dependent Diffusion MRI of Endometrial Cancer\nhttps://doi.org/10.3348/kjr.2025.0633kjronline.org\nAs shown above, compared with benign endometrial \ndiseases, EC exhibits higher cell diameters and vin and lower \ncellularity, cellularity index, and Dex. Additionally, the ADC \nvalues across the three oscillation frequencies were lower \nin EC, reflecting the restricted diffusion typically observed \nin dense tissues, including tumors, lymph nodes, or fibrotic \nregions [36]. Malignant tumor cells usually demonstrate \nlarger nuclei, greater pleomorphism, and higher proliferative \nactivity, compared with normal cells [4], which may lead \nto reduced cellularity per unit area and a corresponding \ndecrease in the cellularity index. These findings were \nsupported by our pathological analysis. Moreover, \naggressive EC exhibits higher cell diameters and v\nin values, \ncompared with nonaggressive EC, while displaying lower \ncellularity, cellularity index, and D\nex. Furthermore, the \nADC values of aggressive EC were lower than those of \nnonaggressive EC. These features are consistent with a more \ndisorganized and heterogeneous tumor microenvironment, \nwhere rapid proliferation may outpace neovascularization, \nresulting in relatively fewer viable tumor cells per unit \narea and more necrotic zones [37]. The difference in the \ncellularity index between aggressive and nonaggressive EC is \nmore pronounced. From the perspective of calculation, the \ncellularity index tends to show less fluctuation and is thus \nless susceptible to noise; yet its overall trend is consistent \nwith that of cellularity, both being lower in aggressive EC. \nImportantly, the pathological features of EC (e.g., cellular \natypia and necrosis) may influence the TDD-MRI parameters. \nIncreased atypia enhances intracellular complexity and \nreduces the extracellular space, boosting restricted diffusion \nand decreasing the ADC or D\nex values. Conversely, necrosis \nmay increase diffusivity due to the loss of structural barriers \nand increase free water content, potentially contributing \nto localized heterogeneity in DWI-derived parameters. v\nin \nis significantly underestimated owing to transcytolemmal \nwater exchange [38] but strongly correlates with pathology-\nderived values [19], indicating its potential utility in \nreflecting microstructural changes. In patients with EC, \nD\nex and vin values obtained via TDD-MRI were correlated \nwith D0Hz. Further, Dex and vin were correlated with D0Hz in \nnonaggressive EC. Dex and cellularity were correlated with \nD0Hz in nonaggressive EC, corroborating previous studies [14].\nOur study has several limitations. The sample size was \nrelatively limited, particularly in certain subgroups, such \nas aggressive EC, which accounts for only one-sixth of all \nEC cases, according to epidemiological data. Although our \ncohort reflects the disease distribution, the small number \nof aggressive cases may limit the statistical power for \nsubgroup comparisons. To address this limitation, we plan \nto expand patient recruitment in future studies, focusing \non the underrepresented subtypes. A larger cohort with \nimproved control of confounding factors will enable a more \ncomprehensive molecular-level evaluation of EC and more \nrobust statistical analyses. Second, the microstructural \nparameters were based on the ROI averages. Future \nwork will explore advanced postprocessing to visualize \nMRI microstructural parameters across lesions and the \nsurrounding environment. Furthermore, combining TDD-\nMRI with other novel MRI technologies to develop models \nincorporating multiple imaging modalities and parameters \ncan enhance our ability to noninvasively predict molecular-\nlevel information related to endometrial diseases and EC.\nIn summary, this study demonstrated the feasibility \nand high diagnostic performance of TDD-MRI-derived \nmicrostructural parameters in non-invasively differentiating \nbenign and malignant endometrial diseases, as well as in \nidentifying the aggressiveness of EC. Further validation \nis warranted, with a larger sample size, multicenter \ncollaboration, and incorporation of various pathological and \nmolecular information.\nAvailability of Data and Material\nThe datasets generated or analyzed during the study are \navailable from the corresponding author on reasonable \nrequest.\nConflicts of Interest\nThe authors have no potential conflicts of interest to \ndisclose.\nAuthor Contributions\nConceptualization: Qi Yang, Hua Li, Junzhong Xu. Data \ncuration: Wenyi Yue, Ruxue Han, Junzhong Xu. Formal \nanalysis: Wenyi Yue, Ruxue Han, Ning Xu. Funding \nacquisition: Qi Yang, Hua Li, Dan Zhao. Investigation: \nWenyi Yue, Ruxue Han. Methodology: Dandan Zheng, Jing \nPeng. Project administration: Qi Yang, Hua Li, Dan Zhao. \nResources: Chaoyang Jin, Xiaoyu Jiang. Software: Junzhong \nXu, Chaoyang Jin, Xiaoyu Jiang. Supervision: Qiming Liu, \nNing Xu. Validation: Jun Lu, Chaoyang Jin, Xiaoyu Jiang. \nVisualization: Wenyi Yue, Ruxue Han, Junzhong Xu, Dandan \nZheng. Writing—original draft: Wenyi Yue, Ruxue Han, \nJunzhong Xu. Writing—review & editing: Qi Yang, Hua Li.\n\n984\nYue et al.\nhttps://doi.org/10.3348/kjr.2025.0633 kjronline.org\nORCID IDs\nWenyi Yue\nhttps://orcid.org/0009-0005-1104-5153\nRuxue Han\nhttps://orcid.org/0009-0005-0577-7286\nJunzhong Xu\nhttps://orcid.org/0000-0001-7895-4232\nChaoyang Jin\nhttps://orcid.org/0000-0002-9749-8962\nXiaoyu Jiang\nhttps://orcid.org/0000-0002-0369-6301\nDandan Zheng\nhttps://orcid.org/0000-0001-8526-0386\nJing Peng\nhttps://orcid.org/0000-0002-8835-4545\nJun Lu\nhttps://orcid.org/0000-0003-0363-7288\nQiming Liu\nhttps://orcid.org/0000-0002-7928-4557\nNing Xu\nhttps://orcid.org/0000-0002-7770-8731\nDan Zhao\nhttps://orcid.org/0000-0002-3231-267X\nHua Li\nhttps://orcid.org/0000-0002-1172-6750\nQi Yang\nhttps://orcid.org/0000-0002-5773-0456\nFunding Statement\nThis work was supported by the Beijing Hospitals Authority’s \nAscent Plan (DFL20220303), Beijing Key Specialists in \nMajor Epidemic Prevention and Control, Beijing Science \nand Technology Cross-disciplinary New Star Project, \nNational Natural Science Foundation of China (62476180), \nNational Natural Science Foundation of China (62176267), \nCAMS Innovation Fund for Medical Sciences (2024-I2M-\nC&T-A-003), Qinghai Provincial Science and Technology Plan \nSpecial Aid Project for Qinghai (2025-QY-220).\nREFERENCES\n1. Lefebvre TL, Ueno Y, Dohan A, Chatterjee A, Vallières M, \nWinter-Reinhold E, et al. Development and validation of \nmultiparametric MRI-based radiomics models for preoperative \nrisk stratification of endometrial cancer. Radiology \n2022;305:375-386\n2. Clarke MA, Long BJ, Del Mar Morillo A, Arbyn M, Bakkum-\nGamez JN, Wentzensen N. Association of endometrial cancer \nrisk with postmenopausal bleeding in women: a systematic \nreview and meta-analysis. JAMA Intern Med 2018;178:1210-\n1222\n3. Berek JS, Matias-Guiu X, Creutzberg C, Fotopoulou C, Gaffney \nD, Kehoe S, et al. FIGO staging of endometrial cancer: 2023. J \nGynecol Oncol 2023;34:e85\n4. Yu Z, Zhang J, Zhang Q, Wei S, Shi R, Zhao R, et al. Single-\ncell sequencing reveals the heterogeneity and intratumoral \ncrosstalk in human endometrial cancer. Cell Prolif \n2022;55:e13249\n5. Ren F, Wang L, Wang Y, Wang J, Wang Y, Song X, et al. Single-\ncell transcriptome profiles the heterogeneity of tumor cells \nand microenvironments for different pathological endometrial \ncancer and identifies specific sensitive drugs. Cell Death Dis \n2024;15:571\n6. Ren X, Liang J, Zhang Y, Jiang N, Xu Y, Qiu M, et al. Single-\ncell transcriptomic analysis highlights origin and pathological \nprocess of human endometrioid endometrial carcinoma. Nat \nCommun 2022;13:6300\n7. Horn LC, Meinel A, Handzel R, Einenkel J. Histopathology \nof endometrial hyperplasia and endometrial carcinoma: an \nupdate. Ann Diagn Pathol 2007;11:297-311\n8. Subramaniam KS, Tham ST, Mohamed Z, Woo YL, Mat \nAdenan NA, Chung I. Cancer-associated fibroblasts \npromote proliferation of endometrial cancer cells. PLoS One \n2013;8:e68923\n9. Fujii S, Gonda T, Yunaga H. Clinical utility of diffusion-weighted \nimaging in gynecological imaging: revisited. Invest Radiol \n2024;59:78-91\n10. Gore JC, Xu J, Colvin DC, Yankeelov TE, Parsons EC, Does MD. \nCharacterization of tissue structure at varying length scales using \ntemporal diffusion spectroscopy. NMR Biomed 2010;23:745-756\n11. Xu J, Jiang X, Li H, Arlinghaus LR, McKinley ET, Devan SP, et al. \nMagnetic resonance imaging of mean cell size in human breast \ntumors. Magn Reson Med 2020;83:2002-2014\n12. Wu D, Jiang K, Li H, Zhang Z, Ba R, Zhang Y, et al. Time-\ndependent diffusion MRI for quantitative microstructural \nmapping of prostate cancer. Radiology 2022;303:578-587\n13. Ba R, Wang X, Zhang Z, Li Q, Sun Y, Zhang J, et al. Diffusion-\ntime dependent diffusion MRI: effect of diffusion-time on \nmicrostructural mapping and prediction of prognostic features \nin breast cancer. Eur Radiol 2023;33:6226-6237\n14. Wang X, Ba R, Huang Y, Cao Y, Chen H, Xu H, et al. Time-\ndependent diffusion MRI helps predict molecular subtypes and \ntreatment response to neoadjuvant chemotherapy in breast \ncancer. Radiology 2024;313:e240288\n15. Zhang H, Liu K, Ba R, Zhang Z, Zhang Y, Chen Y, et al. \nHistological and molecular classifications of pediatric glioma \nwith time-dependent diffusion MRI-based microstructural \nmapping. Neuro Oncol 2023;25:1146-1156\n16. Cao Y, Lu Y, Shao W, Zhai W, Song J, Zhang A, et al. Time-\ndependent diffusion MRI-based microstructural mapping for \ndifferentiating high-grade serous ovarian cancer from serous \n\n985\nTime-Dependent Diffusion MRI of Endometrial Cancer\nhttps://doi.org/10.3348/kjr.2025.0633kjronline.org\nborderline ovarian tumor. Eur J Radiol 2024;178:111622\n17. Solomon E, Lemberskiy G, Baete S, Hu K, Malyarenko D, \nSwanson S, et al. Time-dependent diffusivity and kurtosis \nin phantoms and patients with head and neck cancer. Magn \nReson Med 2023;89:522-535\n18. Jiang X, Li H, Xie J, Zhao P, Gore JC, Xu J. Quantification of \ncell size using temporal diffusion spectroscopy. Magn Reson \nMed 2016;75:1076-1085\n19. Jiang X, Li H, Xie J, McKinley ET, Zhao P, Gore JC, et al. In \nvivo imaging of cancer cell size and cellularity using temporal \ndiffusion spectroscopy. Magn Reson Med 2017;78:156-164\n20. Reynaud O, Winters KV, Hoang DM, Wadghiri YZ, Novikov DS, \nKim SG. Pulsed and oscillating gradient MRI for assessment of \ncell size and extracellular space (POMACE) in mouse gliomas. \nNMR Biomed 2016;29:1350-1363\n21. Manjón JV, Coupé P, Concha L, Buades A, Collins DL, Robles M. \nDiffusion weighted image denoising using overcomplete local \nPCA. PLoS One 2013;8:e73021\n22. Xu J, Devan SP, Shi D, Pamulaparthi A, Yan N, Zu Z, et al. \nMATI: a GPU-accelerated toolbox for microstructural diffusion \nMRI simulation and data fitting with a graphical user \ninterface. Magn Reson Imaging 2025;122:110428\n23. Stringer C, Wang T, Michaelos M, Pachitariu M. Cellpose: a \ngeneralist algorithm for cellular segmentation. Nat Methods \n2021;18:100-106\n24. Gelman A, Jakulin A, Pittau MG, Su YS. A weakly informative \ndefault prior distribution for logistic and other regression \nmodels. Ann Appl Stat 2008;2:1360-1383\n25. Helpman L, Kupets R, Covens A, Saad RS, Khalifa MA, Ismiil N, \net al. Assessment of endometrial sampling as a predictor of \nfinal surgical pathology in endometrial cancer. Br J Cancer \n2014;110:609-615\n26. Chatterjee A, Watson G, Myint E, Sved P, McEntee M, Bourne R. \nChanges in epithelium, stroma, and lumen space correlate more \nstrongly with Gleason pattern and are stronger predictors \nof prostate ADC changes than cellularity metrics. Radiology \n2015;277:751-762\n27. Meyer HJ, Wienke A, Surov A. ADC values of benign and high \ngrade meningiomas and associations with tumor cellularity \nand proliferation - A systematic review and meta-analysis. J \nNeurol Sci 2020;415:116975\n28. Wu X, Pertovaara H, Dastidar P, Vornanen M, Paavolainen L, \nMarjomäki V, et al. ADC measurements in diffuse large B-cell \nlymphoma and follicular lymphoma: a DWI and cellularity \nstudy. Eur J Radiol 2013;82:e158-e164\n29. Novikov DS. The present and the future of microstructure MRI: \nfrom a paradigm shift to normal science. J Neurosci Methods \n2021;351:108947\n30. Tian S, Chen A, Li Y, Wang N, Ma C, Lin L, et al. The combined \napplication of amide proton transfer imaging and diffusion \nkurtosis imaging for differentiating stage Ia endometrial \ncarcinoma and endometrial polyps. Magn Reson Imaging \n2023;99:67-72\n31. Ma C, Tian S, Song Q, Chen L, Meng X, Wang N, et al. Amide \nproton transfer-weighted imaging combined with intravoxel \nincoherent motion for evaluating microsatellite instability in \nendometrial cancer. J Magn Reson Imaging 2023;57:493-505\n32. Yamada I, Wakana K, Kobayashi D, Miyasaka N, Oshima N, \nWakabayashi A, et al. Endometrial carcinoma: evaluation using \ndiffusion-tensor imaging and its correlation with histopathologic \nfindings. J Magn Reson Imaging 2019;50:250-260\n33. Xu J, Li K, Smith RA, Waterton JC, Zhao P, Chen H, et al. \nCharacterizing tumor response to chemotherapy at various \nlength scales using temporal diffusion spectroscopy. PLoS One \n2012;7:e41714\n34. Iima M, Yamamoto A, Kataoka M, Yamada Y, Omori K, \nFeiweier T, et al. Time-dependent diffusion MRI to distinguish \nmalignant from benign head and neck tumors. J Magn Reson \nImaging 2019;50:88-95\n35. Maekawa T, Hori M, Murata K, Feiweier T, Kamiya K, Andica C, \net al. Differentiation of high-grade and low-grade intra-axial \nbrain tumors by time-dependent diffusion MRI. Magn Reson \nImaging 2020;72:34-41\n36. Bollineni VR, Kramer G, Liu Y, Melidis C, deSouza NM. A literature \nreview of the association between diffusion-weighted MRI derived \napparent diffusion coefficient and tumour aggressiveness in \npelvic cancer. Cancer Treat Rev 2015;41:496-502\n37. Tamai K, Koyama T, Saga T, Umeoka S, Mikami Y, Fujii S, et al. \nDiffusion-weighted MR imaging of uterine endometrial cancer. \nJ Magn Reson Imaging 2007;26:682-687\n38. Li H, Jiang X, Xie J, Gore JC, Xu J. Impact of transcytolemmal \nwater exchange on estimates of tissue microstructural \nproperties derived from diffusion MRI. Magn Reson Med \n2017;77:2239-2249","source_license":"CC0","license_restricted":false}