The role of apparent diffusion coefficient histogram analysis in improving O-RADS magnetic resonance imaging classification of adnexal lesions: a retrospective diagnostic accuracy study

Introduction

Globally, ovarian cancer is the fourth leading cause of death among neoplastic diseases in women, with 324,398 new cases and 206,839 death cases in 2022 (1). The poor prognosis is attributed to the fact that 75% of cases are diagnosed at an advanced stage [The International Federation of Gynaecology and Obstetrics (FIGO) stage III and IV] (2). Therefore, correct preoperative classification of benign and malignant adnexal lesions is crucial for developing suitable treatment strategies and improving the prognosis. This classification can avoid unnecessary or overly invasive surgical interventions for patients with benign lesions (3) and avoid suboptimal initial cytoreductive surgery for patients with malignant lesions (4).

The Ovarian-Adnexal Reporting Data System (O-RADS) magnetic resonance imaging (MRI) scoring is a five-point risk stratification system that combines morphological features with diffusion and perfusion features, encompassing signal intensity on T2-weighted, T1-weighted, diffusion-weighted sequences, along with contrast-enhanced (CE) MRI (5). This classification enables allocation of malignancy probability, with scores 2 or 3 predicting benignity and scores 4 or 5 predicting malignancy. When an adnexal lesion is classified as O-RADS MRI scores 2 or 3, the negative predictive value for malignancy was 98%. The O-RADS MRI recommends dynamic contrast-enhanced (DCE) MRI with time-intensity curve (TIC) analysis for the evaluation of the solid component of tumors, which allows categorization of adnexal masses as 3, 4, or 5. A non-DCE scan can serve as an alternative when DCE MRI is inaccessible (6). Subjectively comparing the enhancement of solid tissue between the lesion and the myometrium is often used to classify a lesion as O‑RADS 4 or 5. However, this subjective approach has inherent limitations: it may theoretically misclassify some O‑RADS 3 lesions as O‑RADS 4, thereby leading to an overestimation of malignancy. Wengert et al. indicated that the TIC analysis was more accurate than visual assessment (7). However, in clinical practice, including DCE MRI in magnetic resonance (MR) protocol can be challenging for certain patients and clinical centers, primarily due to extended acquisition time and technical limitations (lack of perfusion curve analysis software). In an ESUR/ESR survey, Nougaret et al. found that the full DCE protocol was only used by 243/840 (28.93%) radiologists (8).

Diffusion weighted imaging (DWI) is a noninvasive technique used to evaluate the microscopic mobility of water molecules in tissues. The qualitative evaluation of DWI signal intensity is incorporated into the O-RADS MRI score; while quantitative analysis of DWI [measurements of apparent diffusion coefficient (ADC) values] is not included. However, DWI and the ADC are crucial for differentiating benign from malignant lesions in clinical practice, and they play a significant role in other scoring systems, such as PIRADS for prostate assessment (9,10). Therefore, for patients without DCE imaging, combining DWI/ADC may help reclassify lesions initially assigned an O‑RADS score. Previous studies measured ADC mean value (ADC_Mean) on the largest sections of adnexal lesions and verified its additional value to the O-RADS MRI based classification (11-14). Among these studies, Hottat et al. (11) and Chen et al. (12) specifically indicated that incorporating DWI/ADC with non-DCE MRI reclassifying lesions presented with O-RADS MRI scores-4, which suggested that addition of DWI/ADC to a non-DCE protocol can help with diagnosis in scenarios without DCE.

Voxel-based whole-lesion volume histogram analysis of the ADC map can quantitatively assess the distribution of voxel ADC values and thus can comprehensively reflect histologic heterogeneity of tumors (15,16). This method has recently been used as a potential tool for improving identification of malignancy, grading and prediction of treatment response across a variety of tumor types (17-19). The purpose of the present study is to evaluate the value of ADC histogram analysis to improve the diagnostic performance of the O-RADS MRI score in differential diagnosis of adnexal lesions. We present this article in accordance with the STARD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2877/rc).

Methods

Patients

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by ethics committee of Beijing Tsinghua Changgung Hospital (No. 24684-4-01) and individual consent for this retrospective analysis was waived. From May 2017 to April 2024, 429 individuals diagnosed with indetermined adnexal lesions on ultrasound examination underwent pelvic MRI examination at Beijing Tsinghua Changgung Hospital. There were 233 patients excluded due to the following reasons: (I) absence of histopathological results and clinical information (n=130); (II) history of prior treatment for adnexal lesions (n=26); (III) poor image quality (The image quality was assessed visually and poor image quality included the following situations: presence of obvious artifacts and inadequate clarity of the boundaries of the adnexal lesions) (n=10); (IV) patients without ADC maps or CE MR images (n=67). The final cohort included 196 patients with 235 adnexal lesions (Figure 1). The levels of serum carbohydrate antigen 125 (CA-125) were measured within 1 week before the operation.

Figure 1 Flowchart of the study population. ADC, apparent diffusion coefficient; MR, magnetic resonance.

MRI acquisition

All MRI examinations were performed on a 3.0 T MR scanner (Ingenia 3.0 CX; Philips Healthcare, Best, The Netherlands) with a 16-channel Torso coil. The MRI acquisition protocol included the following sequences: T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), DWI, and CE MRI sequences. The detailed imaging parameters are shown in Table S1. Gadobutrol injection was given at a dose of 0.1 mL/kg at a rate of 2 mL/s. A pre-contrast T1WI was performed before the injection, and a 3-phase (arterial phase at 30 s, venous phase at 65 s, and delayed phase at 180 s after the end of the injection) post-contrast acquisition was employed.

Image analysis

Two radiologists (R1 and R2 with 4 and 10 years of experience in female pelvic MRI, respectively) independently classified all the adnexal lesions according to the O-RADS MRI scoring system (6), blinded to the clinical and histopathological data. For non- DCE MRI scans, any lesion with solid tissues enhancing less than or equal to the myometrium at 30 seconds (arterial phase) was classified as score 4, while those with greater enhancement than the myometrium were classified as score 5. Adnexal masses with scores 1–3 were considered benign, while those with scores 4–5 were considered malignant. Lesions with disagreement in the O-RADS classification were reviewed by a senior radiologist (R3) and a final classification was used for statistical analysis.

The detailed three-dimensional (3D) volume of interest (VOI) constructing process is shown in Figure 2: (I) all images were transferred to FireVoxel software (https://firevoxel.org/download); (II) region of interests (ROIs) were drawn inside the outer margin of each adnexal lesion on every fourth slice of the lesion avoiding obvious artifacts (e.g., the lesion could be seen on 9 slices, the ROIs were drawn on the 1st, 5th, and 9th slices); (III) the tool “Fill 2D Contours and Morph Convex” was then used to extend the ROIs over the whole lesion, and a 3D VOI was constructed; (IV) an ADC histogram including the ADC values of voxels of the 3D VOI was generated, which included the following parameters: ADC minimum value (ADC_Min), ADC maximum value (ADC_Max), ADC_Mean, ADC 10th, 25th, 50th, 75th, 90th percentiles, standard deviation (ADC_SD), coefficient of variation (ADC_CV), skewness, kurtosis, and entropy. Quantitative analysis of all the lesions was performed by a radiologist (R1 with 4 years of experience in female pelvic MRI), blinded to clinical and histological data. 47 lesions (20% of the sample size) were randomly selected, and quantitative analysis was performed independently by another radiologist (R2 with 10 years of experience in female pelvic MRI). T2-weighted and CE-MRI images were used as anatomical reference.

Figure 2 The process of ROI construction. 2D, two-dimensional; 3D, three-dimensional; ADC, apparent diffusion coefficient; VOI, volume of interest.

Standard reference

The type of adnexal surgery for each patient was determined by a multidisciplinary team in accordance with established clinical protocols. Histopathological diagnoses were made by gynecologic pathologists who were blinded to the quantitative ADC histogram results and the specific O‑RADS MRI score, although they were aware of the gross description of the specimen according to standard clinical practice. Histological types were determined based on surgically excised specimens, according to the criteria of World Health Organization (WHO) classification of female genital tumors (5^th edition) (20). Borderline lesions were considered malignant for analytical purpose.

Statistical analysis

SPSS software (Version 20.0, IBM, Armonk, NY, USA) and MedCalc version 15.6.1 for Windows (MedCalc software, Mariakerke, Belgium) were used for statistical analysis. To assess the stability of the multivariable logistic regression model, we applied the widely accepted 10 events per variable (EPV) rule (21). Kolmogorov-Smirnov test was used to test whether the distribution of the measured data varied significantly from normal. Independent samples t-test or Mann-Whitney U-test were employed to compare the differences of all parameters between benign and malignant tumors. A P value <0.05 was considered statistically significant. Multicollinearity analysis and univariate logistic regression analysis were performed to screen independent variables for predicting malignancy and invasiveness [variance inflation factor (VIF) larger than 10 represented the existence of collinearity]. Candidate variables with a P value <0.2 on univariate analysis were included in the multivariable model. Backward stepwise multivariate logistic regression analysis was then performed to determine the optimum model. The diagnostic performance of ADC histogram parameters and the O-RADS MRI score were assessed in terms of the area under the curve (AUC), the sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and accuracy. The AUC was compared using the Delong method (20). The intraclass correlation coefficient (ICC) was employed for inter-observer agreements of ADC values and the quadratic kappa coefficient was used for inter-observer agreement of the O-RADS MRI score. An ICC greater than 0.90 and a quadratic kappa coefficient greater than 0.8 are indicative of excellent reliability (22).

Results

Patient characteristics

A total of 196 patients with 235 ovarian lesions were involved in the final analysis. There were 108 benign, 17 borderline, and 110 malignant lesions. The number of benign and malignancy events in our cohort satisfied the 10 EPV rule (21). The mean patient age was 53.49 years (range:18.70–91.90 years). Patients in the malignancy group had a significantly higher mean age (55.60±13.59 years) and mean CA‑125 level (643.19±1,339.98 U/mL) compared with those in the benign group (51.00±16.24 years, P=0.02; and 32.42±49.17 U/mL, P<0.001, respectively). Regarding menopausal status, 99 patients (42.13%) were pre‑menopausal and 136 (57.87%) were post‑menopausal. The distribution differed significantly between the two groups (χ²=7.75, P=0.005). In the benign group, 56 (51.85%) were pre‑menopausal and 52 (48.15%) were post‑menopausal, whereas in the malignancy group, 43 (33.86%) were pre‑menopausal and 84 (66.14%) were post‑menopausal. The final diagnoses of all lesions according to the histopathology results are shown in Table 1. The mean time between MRI examination and gynecological operation was 52.8 days (range: 2–107 days). No adverse effects were observed during or after MRI examination.

Inter-rater reproducibility

The ADC histogram parameters used for univariate and multivariate analysis exhibited good to excellent inter-observer consistency (the ICCs ranged from 0.841 to 0.948) (Table S2). The inter-observer agreement for the O-RADS MRI score was excellent [quadratic kappa coefficient: 0.931, 95% confidence interval (CI): 0.910 to 0.953], with different classifications in 37 (15.74%) lesions.

Comparison of ADC histogram parameters between benign and malignant adnexal lesions

As shown in Table 2, there were significant differences in 12 ADC histogram parameters (ADC_Min, ADC_Mean, ADC_10th, ADC_25th, ADC_50th, ADC_75th, ADC_90th, ADC_SD, ADC_CV, ADC_Skewness, ADC_Kurtosis, ADC_Entropy; P<0.001) for discriminating between benign and malignant adnexal lesions (Borderline lesions were considered malignant). No significant difference was found in ADC_Max between benign and malignant lesions (P=0.97). Since the distribution of SD is typically skewed, the one-term transformation Lg ADC_SD is used for statistical analysis. There was a significant difference in O-RADS MRI scores of benign and malignant adnexal lesions (P<0.001) (Table 2).

Independent predictors of malignancy

Based on multicollinearity diagnosis and Pearson correlation analysis, the ADC_Min, ADC_Mean, ADC_10th, ADC_25th, ADC_50th, ADC_75th, ADC_90th, and ADC_CV were removed, which reduced the multicollinearity with all VIFs less than 5 (Table 3). The univariate and multivariate analysis of ADC histogram parameters and CA-125 levels are shown in Table 3. Since the distribution of ADC_SD is typically skewed, the one-term transformation Lg ADC_SD is used for statistical analysis (23). Among the histogram parameters, the ADC_Mean, the ADC_Kurtosis and the CA-125 level were independent predictors for identifying malignant adnexal lesions. There was an improvement in the overall performance of the O-RADS MRI score after the integration of the histogram parameters (Figure 3). For identifying malignancy of adnexal masses, the O-RADS MRI score alone achieved an AUC of 0.915 (95% CI: 0.872 to 0.948) (using a cut‑off value of 3), with a sensitivity of 84.25%, specificity of 90.74%, PLR of 9.10, NLR of 0.17, and accuracy of 87.23%. The ADC-CA-125 model (based on ADC histogram parameters and serum CA-125 levels) yielded an AUC of 0.920 (95% CI: 0.862 to 0.941) (using a cut‑off value of 0.58), with a sensitivity of 83.46%, specificity of 92.59%, PLR of 11.27, NLR of 0.18, and accuracy of 85.96%. The combination of the ADC-CA-125 model and the O‑RADS MRI score exhibited higher AUC (0.956, 95% CI: 0.921 to 0.978, using a cut-off value of 0.41), with a sensitivity of 92.91%, specificity of 89.81%, PLR of 9.12, NLR of 0.08, and accuracy of 90.64% (Table 4, Figure 4). There is a significant difference between the AUCs of the combination and the O-RADS MRI score (P<0.001). Among the 57 lesions characterized as O-RADS 4, 7 lesions were benign. After adding ADC histogram parameters and CA-125 levels, 5 lesions were reclassified as benign lesions (Figure 3).

Figure 3 A 54-year-old woman with pathologically confirmed serous cystadenoma and classified as O-RADS MRI 4. (A) Axial T2-weighted image shows a unilocular cystic mass (asterisk) of the right ovary, with a papillary projection (arrow). (B) Axial T1-weighted contrast-enhanced MRI showed the papillary projection (arrow) enhanced < myometrium at 30 s. (C) Axial DWI (b-value, 800 s/mm²) showed a predominantly low-signal mass (asterisk) with a papillary projection exhibiting high signal intensity (arrow). (D,E) ADC map (D) and histogram (E) revealed a whole-lesion ADC_Mean of 2.67×10⁻³ mm²/s, an ADC_Kurtosis of 8.57 (high peak, a sharp shape of the histogram), and a CA-125 level of 23.2 U/mL; the ADC map (D) showed a cystic mass (asterisk) with predominantly high ADC value a papillary projection (arrow) with low ADC value. The lesion was characterized as O-RADS 4 with intermediate risk of malignancy and reclassified as benign lesion after adding ADC histogram analysis. ADC, apparent diffusion coefficient; ADC_Kurtosis, ADC kurtosis; ADC_Mean, ADC mean value; CA-125, cancer antigen 125; DWI, diffusion weighted imaging; MRI, magnetic resonance imaging; O-RADS, Ovarian-Adnexal Reporting Data System.

Figure 4 Comparison of the ROC curves. ROC analysis for differentiating benign from malignant adnexal masses. The O-RADS MRI score showed an AUC of 0.915. The ADC-CA-125 model showed an AUC of 0.920. The combination of the ADC-CA-125 model and the O‑RADS MRI score exhibited higher AUC of 0.956. ADC, apparent diffusion coefficient; AUC, area under the curve; CA-125, cancer antigen 125; MRI, magnetic resonance imaging; O-RADS, Ovarian-Adnexal Reporting Data System; ROC, receiver operating characteristic.

Discussion

Summary of main results

Our study demonstrated that the O-RADS MRI score combined with ADC histogram parameters (the ADC_Mean and the ADC_kurtosis) and the serum CA-125 levels could improve the identification of malignant adnexal lesions, with a good to excellent inter-observer consistency. The main findings support our hypothesis about the value of ADC histogram analysis to improve the diagnostic performance of the O-RADS MRI score in adnexal lesions.

Results in the context of published literature

The role of quantitative analysis of DWI in the characterization of adnexal lesions has still not been verified. Previous studies have varied conclusions on the diagnostic efficacy of ADC values in evaluating the malignancy of ovarian lesions. The mean ADC cutoff values discriminating malignant from benign adnexal masses ranged from 1.15 to 1.361×10⁻³ mm²/s, with a large overlap between malignant and benign lesions (22). This overlap in mean ADC values might be related to the heterogeneity of malignant lesions, due to the existence of desmoplastic stroma, interstitial edema and necrosis or cystic areas intervening papillary components (23,24). Therefore, the mean ADC along might not be enough for malignancy identification. Several studies have investigated the value of DWI quantitative analysis to improve diagnostic performance of the O-RADS MRI score (11-14). Hottat et al. first demonstrated that incorporating ADC thresholds improves diagnostic performance in O-RADS MRI score 4 lesions, with both single-slice and whole-lesion histogram ADC means yielding similar improvements (11). Similarly, Manganaro et al. reported that ADC values aided in reclassifying O-RADS 3–5 lesions (13). Most of these studies delineated ROIs on a single axial slice, typically the largest cross section of a lesion, and only included the solid components. There are some potential problems of the quantitative analysis method. Firstly, benign adnexal lesions (including benign ovarian tumors and non-neoplastic lesions) usually appear as unilocular or multilocular cysts with thin walls, minimal septations, and absence of papillary projections (25), and thus it is difficult to draw ROIs on solid components of these lesions. ADC measurements based on small ROIs drawn on artificially selected lesion regions (usually the darkest regions on the ADC maps) may lead to wide variability in ADC values, and the interobserver variability in ROI positioning could result in lack of objectivity and reproducibility (26).

In the present study, we performed whole-lesion histogram analysis of ADC maps. Theoretically, whole-lesion analysis provides a more comprehensive evaluation of lesions than single-slice ROI by incorporating all voxels—both cystic and solid components, and demonstrates superior interobserver agreement due to larger number of pixels sampled (27). Histogram analysis of the whole lesion provides detailed information about the examined tissue such as percentiles, skewness, kurtosis, and inhomogeneity parameters including SD and entropy. These parameters were reported to be reliable quantitative markers of tumor heterogeneity with a more direct correlation to the underlying structural, physiological, molecular as well as metabolic changes inside tumors (28). Exploring tumor heterogeneity is important in cancer research, as it is related to tumor prognosis and will point the way to more rational anti-cancer therapy (29).

In our study, multivariate analysis demonstrated that in addition to the ADC_Mean, the ADC_Kurtosis were also independent predictors of malignancy. Kurtosis reflects the peakedness of the distribution and is a measure of the shape of the probability distribution, and usually related with tumor progression (28). In our study, the malignant group exhibited a lower ADC_Kurtosis compared with the benign group. Previous studies have shown conflict results. Zhu et al. found that ADC_Kurtosis of lung cancer was lower than that of benign pulmonary lesions (16); while Suo et al. and Bougias et al. found that ADC_Kurtosis of malignant breast lesions was higher than benign lesions without significant statistical significance (30,31). The role of ADC_Kurtosis in identification of malignant tumors merits further investigation.

CA125 has been used as a tumor marker for the diagnosis and monitoring of ovarian cancer for over 30 years, with a widely adopted serum threshold of <35 U/mL. Its clinical value is well established in the diagnosis, treatment response evaluation, and recurrence monitoring of ovarian cancer patients (32). The present study further showed that CA‑125 is an independent risk factor for malignant adnexal masses, and its integration into the O‑RADS system may improve diagnostic performance. This is consistent with prior evidence that CA‑125 improves the diagnostic efficacy of ORADS US (33).

In our study, the sensitivity and specificity of the O-RADS MRI score were 84.25% and 90.74%, respectively. The validation study by Tomassin-Naggara et al. implied a sensitivity of 93% and a specificity of 91% (5). A meta-analysis including 13 study parts from 12 studies found a 92% summary sensitivity (ranging from 81% to 98%) and a 91% summary specificity (ranging from 83% to 98%) (34). The differences behind these findings may be due to patient selection bias and variability in the protocols of CE imaging. The O-RADS MRI recommends DCE-MRI with TIC analysis for the evaluation of the solid component of tumors, which allows categorization of adnexal masses as 3, 4 or 5. The American College of Radiology (ACR) acknowledges non-DCE scan as an acceptable alternative (6), by comparing the enhancement of solid tissue of the lesion and the myometrium, the lesion can be determined as O-RADS 4 or 5. This alternative method may result in misclassification of some O-RADS 3 lesions as O-RADS 4, potentially leading to an over-estimation of malignancy. In the present study, participants underwent non-DCE MRI examination, and the incorporation of ADC histogram analysis and CA-125 facilitates reclassification of O-RADS 4 lesions into benign category (Figure 3). Including DCE MRI in MR protocol can be challenging for certain patients and clinical centers, primarily due to extended acquisition time and technical limitations (lack of perfusion curve analysis software). For a scanning protocol with non-DCE MRI, the addition of ADC histogram analysis and CA-125 levels could serve as a supplementary tool to improve the accuracy of classification based on the O-RADS MRI score.

Strengths and limitations

The main strength of this study lies in its integration of ADC histogram analysis and CA-125 with O-RADS MRI score, which may enhance the objectivity and reproducibility of adnexal lesion evaluation. The strong inter-observer agreement demonstrated in this study supports its clinical applicability.

This study has several limitations. Its retrospective, single-center nature, lack of a priori sample size estimation and the higher-than-average incidence of malignant lesions may restrict the generalizability of our results. Nevertheless, the 10 EPV rule supported the stability of the multivariable model. The increased incidence of malignancy may be attributed to the clinical utility of the O-RADS MRI score, which is designed for sonographically indeterminate adnexal masses (5). Consequently, patients with benign-appearing lesions on ultrasound may be more likely to undergo ultrasound monitoring rather than MRI, and less likely to pursue surgery or long-term follow-up. Additionally, the absence of external validation for the multivariate model is a significant limitation of our study. Therefore, external validation in a multi‑center prospective cohort is mandatory before the proposed combined model can be considered for clinical implementation.

Implications for practice and future research

ADC histogram analysis and serum CA-125 levels may serve as a supplementary tool to improve diagnostic confidence in clinical settings when DCE-MRI is inaccessible or radiologists face diagnostic uncertainty based on the O-RADS MRI score. Future studies with multi-center, prospective cohorts are needed to further validate generalizability of the combined diagnostic method.

Conclusions

In conclusion, incorporating ADC histogram analysis shows potential to improve the diagnostic performance of the O-RADS MRI score for adnexal lesions. In clinical scenarios where DCE-MRI is inaccessible or O-RADS MRI-based diagnosis is equivocal, ADC histogram analysis and serum CA-125 levels—owing to their objectivity and reproducibility—may provide valuable supplemental information to strengthen diagnostic confidence. This study should be regarded as hypothesis generating and requires multi-center prospective validation.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2877/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2877/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2877/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2877/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by ethics committee of Beijing Tsinghua Changgung Hospital (No. 24684-4-01) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Torre LA, Trabert B, DeSantis CE, et al. Ovarian cancer statistics, 2018. CA Cancer J Clin 2018;68:284-96. [Crossref] [PubMed]
3. Sideris M, Menon U, Manchanda R. Screening and prevention of ovarian cancer. Med J Aust 2024;220:264-74. [Crossref] [PubMed]
4. Rizzo S, Avesani G, Panico C, et al. Ovarian cancer staging and follow-up: updated guidelines from the European Society of Urogenital Radiology female pelvic imaging working group. Eur Radiol 2025;35:4029-39. [Crossref] [PubMed]
5. Thomassin-Naggara I, Poncelet E, Jalaguier-Coudray A, et al. Ovarian-Adnexal Reporting Data System Magnetic Resonance Imaging (O-RADS MRI) Score for Risk Stratification of Sonographically Indeterminate Adnexal Masses. JAMA Netw Open 2020;3:e1919896. [Crossref] [PubMed]
6. Sadowski EA, Thomassin-Naggara I, Rockall A, et al. Erratum for: O-RADS MRI Risk Stratification System: Guide for Assessing Adnexal Lesions from the ACR O-RADS Committee. Radiology 2023;308:e239017. [Crossref] [PubMed]
7. Wengert GJ, Dabi Y, Kermarrec E, et al. O-RADS MRI Classification of Indeterminate Adnexal Lesions: Time-Intensity Curve Analysis Is Better Than Visual Assessment. Radiology 2022;303:566-75. [Crossref] [PubMed]
8. Nougaret S, Lakhman Y, Bahadir S, et al. Ovarian-Adnexal Reporting and Data System for Magnetic Resonance Imaging (O-RADS MRI): Genesis and Future Directions. Can Assoc Radiol J 2023;74:370-81. [Crossref] [PubMed]
9. Liu X, Xiong Q, Zeng W, et al. Comparison of the Utility of PI-RADS 2.1, ADC Values, and Combined Use of Both, for the Diagnosis of Transition Zone Prostate Cancers. J Comput Assist Tomogr 2024;48:206-11. [Crossref] [PubMed]
10. Huang H, Liu Z, Ma Y, et al. Based on PI-RADS v2.1 combining PHI and ADC values to guide prostate biopsy in patients with PSA 4-20 ng/mL. Prostate 2024;84:376-88. [Crossref] [PubMed]
11. Hottat NA, Badr DA, Van Pachterbeke C, et al. Added Value of Quantitative Analysis of Diffusion-Weighted Imaging in Ovarian-Adnexal Reporting and Data System Magnetic Resonance Imaging. J Magn Reson Imaging 2022;56:158-70. [Crossref] [PubMed]
12. Chen T, Qian X, Zhu Z, et al. Assessment of the O-RADS scoring system for the differentiation of different types of ovarian neoplasms: A modified approach with non-DCE-MRI. Clin Imaging 2024;116:110285. [Crossref] [PubMed]
13. Manganaro L, Ciulla S, Celli V, et al. Impact of DWI and ADC values in Ovarian-Adnexal Reporting and Data System (O-RADS) MRI score. Radiol Med 2023;128:565-77. [Crossref] [PubMed]
14. Elshetry ASF, Hamed EM, Frere RAF, et al. Impact of Adding Mean Apparent Diffusion Coefficient (ADCmean) Measurements to O-RADS MRI Scoring For Adnexal Lesions Characterization: A Combined O-RADS MRI/ADCmean Approach. Acad Radiol 2023;30:300-11. [Crossref] [PubMed]
15. Liu X, Wang T, Wang Y, et al. Histogram Analysis of Apparent Diffusion Coefficient on Diffusion Weighted Magnetic Resonance Imaging in Differentiation between Low and High Grade Serous Ovarian Cancer. Curr Med Imaging 2023;19:167-74. [Crossref] [PubMed]
16. Zhu Q, Ren C, Xu JJ, et al. Whole-lesion histogram analysis of mono-exponential and bi-exponential diffusion-weighted imaging in differentiating lung cancer from benign pulmonary lesions using 3 T MRI. Clin Radiol 2021;76:846-53. [Crossref] [PubMed]
17. Palmér M, Åkesson Å, Ljungberg M, et al. Preoperative risk assessment of endometrial cancer using histogram analysis of weighted and quantitative MRI images. Abdom Radiol (NY) 2026;51:346-56. [Crossref] [PubMed]
18. Qin Y, Wu F, Hu Q, et al. Histogram analysis of multi-model high-resolution diffusion-weighted MRI in breast cancer: correlations with molecular prognostic factors and subtypes. Front Oncol 2023;13:1139189. [Crossref] [PubMed]
19. Gündoğdu E, Aşılıoğlu BK, Yazıcı C. Whole-lesion CT histogram analysis as an advanced technique in the portal venous phase: differentiating lipid poor adrenal adenomas from pheochromocytomas. Abdom Radiol (NY) 2025;50:1219-27. [Crossref] [PubMed]
20. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837-45. [Crossref] [PubMed]
21. Peduzzi P, Concato J, Kemper E, et al. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373-9. [Crossref] [PubMed]
22. Yuan X, Guo L, Du W, et al. Diagnostic accuracy of DWI in patients with ovarian cancer: A meta-analysis. Medicine (Baltimore) 2017;96:e6659. [Crossref] [PubMed]
23. Fujii S, Kakite S, Nishihara K, et al. Diagnostic accuracy of diffusion-weighted imaging in differentiating benign from malignant ovarian lesions. J Magn Reson Imaging 2008;28:1149-56. [Crossref] [PubMed]
24. Bakir B, Bakan S, Tunaci M, et al. Diffusion-weighted imaging of solid or predominantly solid gynaecological adnexial masses: is it useful in the differential diagnosis? Br J Radiol 2011;84:600-11. [Crossref] [PubMed]
25. Ohya A, Fujinaga Y. Magnetic resonance imaging findings of cystic ovarian tumors: major differential diagnoses in five types frequently encountered in daily clinical practice. Jpn J Radiol 2022;40:1213-34. [Crossref] [PubMed]
26. Yuan Y, Huang M, Peng J, et al. Whole-lesion histogram analysis of multi-model diffusion-weighted imaging for characterization and molecular classification of breast lesions. Radiol Med 2026;131:395-405. [Crossref] [PubMed]
27. Lambregts DM, Beets GL, Maas M, et al. Tumour ADC measurements in rectal cancer: effect of ROI methods on ADC values and interobserver variability. Eur Radiol 2011;21:2567-74. [Crossref] [PubMed]
28. Just N. Improving tumour heterogeneity MRI assessment with histograms. Br J Cancer 2014;111:2205-13. [Crossref] [PubMed]
29. Patel AS, Yanai I. A developmental constraint model of cancer cell states and tumor heterogeneity. Cell 2024;187:2907-18. [Crossref] [PubMed]
30. Bougias H, Ghiatas A, Priovolos D, et al. Whole-lesion histogram analysis metrics of the apparent diffusion coefficient as a marker of breast lesions characterization at 1.5 T. Radiography (Lond) 2017;23:e41-6. [Crossref] [PubMed]
31. Suo S, Zhang K, Cao M, et al. Characterization of breast masses as benign or malignant at 3.0T MRI with whole-lesion histogram analysis of the apparent diffusion coefficient. J Magn Reson Imaging 2016;43:894-902. [Crossref] [PubMed]
32. Ghose A, McCann L, Makker S, et al. Diagnostic biomarkers in ovarian cancer: advances beyond CA125 and HE4. Ther Adv Med Oncol 2024;16:17588359241233225. [Crossref] [PubMed]
33. Pan RK, Zhang SQ, Zhang XY, et al. Clinical value of ACR O-RADS combined with CA125 in the risk stratification of adnexal masses. Front Oncol 2024;14:1369900. [Crossref] [PubMed]
34. Rizzo S, Cozzi A, Dolciami M, et al. O-RADS MRI: A Systematic Review and Meta-Analysis of Diagnostic Performance and Category-wise Malignancy Rates. Radiology 2023;307:e220795. [Crossref] [PubMed]