Integrating O-RADS US v2022, CEUS, and CA125 to enhance the diagnostic differentiation of ovarian masses: development of the OCC-US model.

This retrospective diagnostic study, conducted at Sichuan Cancer Hospital from May 2022 to March 2025, included patients who underwent ultrasound examinations for suspected adnexal masses. The study was reviewed and approved by the Ethics Committee of Sichuan Cancer Hospital (Ethics No.: SCCHEC022019025), with informed consent waived. Initially, 1319 consecutively registered patients with adnexal masses were identified. The selection criteria were as follows: (a) patients with complete clinical data; (b) patients who underwent transvaginal or abdominal ultrasound, or both; (c) patients with pathologic diagnosis or sufficient follow-up. The exclusion criteria were as follows: (a) previous treatment before the ultrasound examination; (b) history of adnexal malignancy; (c) pregnant or lactating women; (d) images with poor quality or incomplete evaluation (O-RADS 0); (e) patients with normal ovaries (O-RADS 1); (f) lacking CEUS records. The patient selection process is illustrated in Fig.  1 . Sufficient follow-up was defined as regular ultrasound examinations conducted every 3 months for a duration exceeding 2 years after the initial assessment, without evidence of lesion progression or malignant transformation [ 4 ]. Demographic data, including age, menopausal status, clinical examination findings, tumor marker levels, surgical findings, pathological diagnosis, and follow-up information, were collected by reviewing hospital electronic medical records and case notes. CA125 levels were required to be measured within two weeks before surgery or follow-up. For patients with bilateral adnexal masses, the mass with the highest O-RADS US v2022 classification and CEUS score was selected. Finally, 609 patients enrolled between May 2022 and May 2024 were included in the development cohort of the OCC-US model. To evaluate the generalizability of the model, a temporally independent validation cohort was retrospectively constructed at the same center, including 300 patients who underwent ultrasound examinations between June 2024 and March 2025, using the same inclusion and exclusion criteria as the development cohort. Among the 909 patients, 773 (85.0%) had histopathological diagnoses, and 136 (15.0%) were diagnosed as benign based on sufficient follow-up. Fig. 1 Flow chart of patient enrollment Flow chart of patient enrollment Transvaginal ultrasound was the preferred examination method. When transvaginal ultrasound could not simultaneously display the mass and the uterine muscle layer, the mass was too large to display, the patient was unmarried, or could not tolerate the transvaginal route, transabdominal ultrasound was used. A total of 71 patients (7.8%) underwent transabdominal ultrasound only. This study primarily used the following ultrasound equipment: Samsung RS80A ultrasound system (Samsung Medison Co., Seoul, Korea), GE LOGIQ E10 (GE Healthcare, FetalHQ ® , Chicago, IL, USA), Philips EPIQ7 (Philips Healthcare, Bothell, WA, USA), and Aplio i800 ultrasound system (Canon Medical Systems, Tokyo, Japan). Transvaginal scans were conducted using endovaginal probes with frequencies of 5–9 MHz, while transabdominal scans used convex probes with frequencies of 1–5 MHz. CEUS was carried out in dual-window mode with a low mechanical index (MI) (0.06 ~ 0.08). The contrast agent SonoVue (Bracco Imaging, Milan, Italy) was suspended in 5 mL of saline solution and gently agitated to obtain a homogeneous suspension, in accordance with the manufacturer’s instructions. A dosage of 2.4 mL was administered for each CEUS examination. B-mode ultrasound, color Doppler ultrasound, and CEUS examinations were performed by experienced sonologists with over 10 years of diagnostic experience in gynecological ultrasound. The onographic morphological features of each lesion were recorded. All images were stored in the Picture Archiving and Communication System (PACS) of the participating hospitals. Before the study began, all participating sonologists received theoretical training, which included explanations of the O-RADS US v2022 ultrasound lexicon, risk stratification management system, and CEUS evaluation standards by senior experts. All ultrasound images were anonymized and randomized, and were independently reviewed and categorized by two sonologists with 5 and 10 years of experience, respectively, in accordance with the O-RADS US v2022 system [ 2 , 14 ]. For retrospective analysis of CEUS examination recordings, the ultrasound examiner subjectively assessed the contrast signal peak intensity (PI) in solid tumors using a contrast score. The scoring was based on comparing the tumor’s vascularity with that of the myometrium, and classified as: 1 = low enhancement, 2 = homogeneous enhancement, 3 = high enhancement. The time when the tumor began to wash-in and reached its PI, followed by wash-out, was recorded in comparison to the myometrium and scored as: 1 = slow wash-in and wash-out, 2 = fast wash-in and slow wash-out, 3 = fast wash-in and wash-out. The reviewers were blinded to the pathological and other imaging results. In case of discrepancies between the evaluations of the two physicians, a third senior physician was involved in the evaluation to reach a consensus. To assess the consistency of the OCC-US scoring system between a junior sonologist (3 years of experience) and a senior expert (11 years of experience), 50 randomly selected adnexal masses were independently analyzed, with relevant features recorded and corresponding scores assigned. To evaluate the OCC-US scoring system, surgical histology and follow-up results served as the reference standard. Clinical and pathological data were collected from electronic records, with histological diagnoses confirmed by surgery or biopsy. All patients included in the study measured serum CA-125 within 1 to 14 days prior to surgery. No patients were excluded due to missing CA-125 values. For multiple similar adnexal masses, the largest or most complex lesion was analyzed, while significantly different bilateral masses were described separately. patients were presumed to be postmenopausal if they were over 50 years of age, had undergone hysterectomy with unknown ovarian status, or had missing menstrual history records. The ultrasound images of patients in the independent validation cohort were independently evaluated by a separate sonologist with 8 years of experience who was blinded to the histopathological and follow-up outcomes. The OCC-US model was applied to all cases using the same scoring criteria as the development cohort. Statistical analysis of the above results was performed using the SPSS 26.0 software package (SPSS, Chicago, IL, USA). For comparisons of categorical variables, the chi-square test or Fisher’s exact test was used. Continuous variables were expressed as mean ± standard deviation, and comparisons were made using the independent samples t-test. Multivariate logistic regression analysis was conducted to identify risk factors and develop a scoring system. Receiver operating characteristic (ROC) curves were used to calculate the area under the curve (AUC) and determine the optimal cutoff value. All tests were two-tailed, and a P-value of less than 0.05 was considered statistically significant. Kappa (κ) statistics were used to assess the inter-observer agreement of the OCC-US scoring classification results. Calibration curves and decision curve analysis (DCA) were used to evaluate the calibration performance and clinical net benefit of the model. The Hosmer-Lemeshow test was conducted to assess the agreement between the predicted probabilities and the actual outcomes.

The development cohort included 609 patients (mean age 42.9 ± 13.9 years, range 17–76 years) with 609 adnexal masses (Fig.  1 ). Among these, 388 (63.7%) had confirmed pathological diagnoses, while the remaining 221 (36.3%) resolved spontaneously or improved with conservative medical treatment during follow-up and were considered benign. These benign lesions included 80 hemorrhagic cysts, 41 corpus luteum cysts, 36 follicular cysts, 13 cases of endometriosis, 18 dermoid cysts, 15 paraovarian cysts, 9 peritoneal inclusion cysts, and 9 cases of hydrosalpinx or pyosalpinx. The pathological findings, long-term follow-up results, and O-RADS US v2022 classifications of the included lesions are summarized in Table S1 . Of the 609 patients, 405 (66.5%) were premenopausal, while 204 (33.5%) were postmenopausal. Malignant adnexal masses were more frequently observed in postmenopausal women (68.5%). Among all adnexal masses, 390 (64.0%) were benign. The most common benign lesion was a hemorrhagic cyst (31.2%), while the most prevalent malignant adnexal mass was serous cystadenocarcinoma (31.5%). Univariate analysis of the O-RADS US v2022 classification, CEUS score, and CA125 levels revealed statistically significant differences (all P  < 0.001). To distinguish malignant ovarian tumors from adnexal masses, ROC curves were generated for individual predictive factors, including O-RADS US v2022, CEUS score, and CA125 level. The optimal cut-off values determined from the ROC analysis were O-RADS ≥ 4, CEUS score ≥ 4, and CA125 ≥ 37.815 U/ml. Multivariate logistic regression analysis was performed to identify independent risk factors for malignant ovarian tumors. The results indicated that O-RADS ≥ 4 (OR 7.813, 95% CI 4.571–13.356, P  < 0.001), CEUS score ≥ 4 (OR 7.928, 95% CI 4.711–13.340, P  < 0.001), and CA125 ≥ 37.815 U/ml (OR 4.264, 95% CI 2.544–7.147, P  < 0.001) were significant independent predictors of malignancy. Based on these predictive factors and their OR values, a diagnostic scoring system for malignant ovarian tumors, termed OCC-US, was developed. In this system, an adnexal mass classified as O-RADS ≥ 4 or a CEUS score ≥ 4 was assigned 2 points each, while a CA125 level ≥ 37.815 U/ml was assigned 1 point, resulting in a total possible score of 5. Each patient was subsequently evaluated using the OCC-US scoring system. The OCC-US scoring criteria for adnexal masses, along with the sensitivity and specificity at different thresholds, are summarized in Table  1 . According to the ROC curve analysis of the OCC-US scoring system, a threshold score of ≥ 3 was determined as the optimal cutoff for diagnosing malignant ovarian tumors. To further evaluate the diagnostic performance of individual and combined parameters, ROC curves were plotted for O-RADS US v2022, CEUS score, CA125 level, OC-US (a combined ultrasound-based scoring system incorporating O-RADS US v2022 and CEUS scores), and OCC-US (Fig. S1 and Table  2 ). Compared to O-RADS US v2022 alone, the OCC-US system significantly improved specificity ( P  < 0.001) and overall diagnostic accuracy ( P  < 0.001) while maintaining a comparable sensitivity. Additionally, the false-positive rate was significantly reduced from 23.1 to 6.2%, indicating that the OCC-US scoring system provides a more reliable and clinically applicable method for distinguishing benign from malignant adnexal masses. Therefore, these results indicate that OCC-US outperformed other scoring systems in terms of overall diagnostic accuracy, while maintaining a balance between sensitivity and specificity. Table 1 Score of OCC-US, and sensitivity, specificity and accuracy of different scores Score ≥ 0 ≥ 1 ≥ 2 ≥ 3 ≥ 4 ≥ 5 Benign tumor 390 162 132 57 36 15 Malignant tumor 219 213 210 186 162 135 Total 609 375 342 243 198 150 Sensitivity (%) 100 97.3 95.9 84.9 74.0 61.6 Specificity (%) 0 58.5 66.2 85.4 90.8 96.2 AUC 0.359 0.568 0.614 0.765 0.818 0.900 OCC-US: O-RADS US v2022, CEUS and CA125, AUC: area under the curve Score of OCC-US, and sensitivity, specificity and accuracy of different scores OCC-US: O-RADS US v2022, CEUS and CA125, AUC: area under the curve Table 2 Diagnostic efficacy of each evaluation method O-RADS US v2022 CEUS score CA125 OC-US OCC-US Sensitivity (%) 86.3 83.6 74.0 74.0 84.9 Specificity (%) 71.5 85.4 83.1 90.8 85.4 AUC 0.871 0.877 0.784 0.891 0.916 O-RADS: ovarian-adnexal reporting and data system, CEUS: contrast-enhanced ultrasound, OC-US: O-RADS US v2022 and CEUS, OCC-US: O-RADS US v2022, CEUS and CA125, AUC: area under the curve Diagnostic efficacy of each evaluation method O-RADS: ovarian-adnexal reporting and data system, CEUS: contrast-enhanced ultrasound, OC-US: O-RADS US v2022 and CEUS, OCC-US: O-RADS US v2022, CEUS and CA125, AUC: area under the curve The OCC-US scores of the study subjects, along with histological diagnoses and long-term follow-up results, are summarized in Table  3 . Among the 234 patients with an OCC-US score of 0, 114 (48.7%) were classified as O-RADS 2 and 120 (51.3%) as O-RADS 3. Among the 33 patients with an OCC-US score of 1, 15 (45.5%) were classified as O-RADS 2 and 18 (54.5%) as O-RADS 3. According to the O-RADS US v2022 guidelines, O-RADS 2 lesions and certain O-RADS 3 lesions with benign imaging features, particularly in premenopausal women, are considered appropriate for follow-up rather than immediate surgical intervention. In our study, patients with OCC-US scores of 0 or 1, regardless of whether they were classified as O-RADS 2 or O-RADS 3, met the criteria for non-operative management. These findings highlight the potential of the OCC-US model to accurately identify patients suitable for surveillance, thereby helping to avoid unnecessary surgery. Notably, Fig.  2 presents a representative case of score upgrading, in which a lesion initially classified as O-RADS 2 was reassessed as OCC-US score 3 due to a high CEUS score and elevated CA125 level. This reclassification led to timely surgical intervention, and the final pathology confirmed a borderline mucinous tumor, highlighting the added diagnostic value of the OCC-US scoring system. Table 3 OCC-US score and histological and long-term follow-up diagnostic results of ovarian adnexal masses Pathological type OCC-US 0 OCC-US 1 OCC-US 2 OCC-US 3 OCC-US 4 OCC-US 5 Total Benign adnexal mass 228 30 75 21 21 15 390 Hemorrhagic cyst 82 0 23 0 0 0 105 Luteal cyst 48 0 0 0 0 0 48 Follicular cyst 36 0 0 0 0 0 36 Endometriosis caused by endometriosis 0 14 14 2 0 0 30 Dermoid cyst 13 4 15 1 0 0 33 Periovarian cyst 13 2 0 0 0 0 15 Peritoneal inclusion cyst 10 0 2 0 0 0 12 Hydrosalpinx and pus accumulation 5 4 4 2 0 0 15 Serous cystadenoma 0 1 3 8 2 7 21 Mucinous cystadenoma 0 2 4 4 2 3 15 Fibroma/Fibrothecoma 0 0 6 3 9 3 21 Exogenous uterine fibroids 0 3 2 1 1 2 9 Mature teratoma 21 0 2 0 7 0 30 Borderline adnexal mass 3 3 8 10 8 4 36 borderline mucinous cystadenoma 1 2 3 4 3 2 15 Borderline serous cystadenoma 2 1 5 6 5 2 21 Malignant adnexal mass 3 0 16 14 19 131 183 Serous cystadenocarcinoma 1 0 4 4 5 55 69 Mucinous cystadenocarcinoma 1 0 6 4 6 31 48 Undifferentiated carcinoma 0 0 3 1 1 1 6 Malignant stromal tumor 0 0 0 1 1 13 15 Granulosa cell tumor 1 0 1 1 2 7 12 Malignant mesodermal mixed tumor 0 0 0 0 2 10 12 Immature teratoma 0 0 1 2 1 5 9 Metastatic cancer 0 0 1 1 1 9 12 O-RADS: ovarian-adnexal reporting and data system; OCC-US: O-RADS US v2022, CEUS and CA125 OCC-US score and histological and long-term follow-up diagnostic results of ovarian adnexal masses O-RADS: ovarian-adnexal reporting and data system; OCC-US: O-RADS US v2022, CEUS and CA125 According to O-RADS US v2022 guidelines, cases classified as O-RADS 4 and O-RADS 5 are recommended for surgery. In our study, within the OCC-US scoring system, we found that for O-RADS ≥ 4 cases, if both CEUS and CA125 scores were 0, the OCC-US score was 2; however, if at least one of CEUS or CA125 scores was > 0, the OCC-US score was ≥ 3. Therefore, we propose that if a lesion is classified as O-RADS ≥ 4 but has CEUS and CA125 scores of 0, the risk level may be downgraded, and the patient may be advised to undergo follow-up observation or elective surgery (Fig.  3 ). However, if the lesion is classified as O-RADS ≥ 4 and at least one of CEUS or CA125 scores is > 0 (OCC-US score ≥ 3), the original O-RADS US v2022 recommendation for surgery should be maintained. Our findings suggest that OCC-US < 3 may indicate a lower malignancy risk, for which continued observation or elective surgery could be appropriate. In contrast, lesions with OCC-US ≥ 3 are more likely to be malignant and should be considered for surgical resection. Additionally, we determined that the optimal cutoff value for OCC-US is ≥ 3, further validating the scientific and clinical effectiveness of this OCC-US-based recommendation. Among the 609 lesions analyzed, the proportion of cases recommended for surgery based on O-RADS US v2022 and OCC-US was 300 and 243, respectively (49.3% vs. 39.9%, P  < 0.001). Fig. 2 Transvaginal ultrasound in a 51-year-old woman revealed a right adnexal mass, initially classified as O-RADS 2 and mistakenly diagnosed as a mature teratoma. However, based on CEUS findings (score = 6) and an elevated CA125 level (57.81 U/ml), the lesion was reassessed with an OCC-US score of 3, leading to its upgrading. Surgical pathology confirmed a borderline mucinous tumor. (A) Gray-scale ultrasound: A 6.5 × 5.4 × 9.8 cm predominantly cystic, mixed echogenic mass was detected, containing a 5.4 cm spherical hyperechoic structure. The mass exhibited a well-defined boundary and a relatively regular shape, with poor acoustic transmission in the cystic component. (B) Color doppler flow imaging (CDFI): The solid portion displayed minimal detectable blood flow signals (color score 1). (C) CEUS (Arterial phase, 20s): The mass exhibited high enhancement with fast wash-in. (D) CEUS (Portal Phase, 42s): The mass demonstrated fast wash-out Transvaginal ultrasound in a 51-year-old woman revealed a right adnexal mass, initially classified as O-RADS 2 and mistakenly diagnosed as a mature teratoma. However, based on CEUS findings (score = 6) and an elevated CA125 level (57.81 U/ml), the lesion was reassessed with an OCC-US score of 3, leading to its upgrading. Surgical pathology confirmed a borderline mucinous tumor. (A) Gray-scale ultrasound: A 6.5 × 5.4 × 9.8 cm predominantly cystic, mixed echogenic mass was detected, containing a 5.4 cm spherical hyperechoic structure. The mass exhibited a well-defined boundary and a relatively regular shape, with poor acoustic transmission in the cystic component. (B) Color doppler flow imaging (CDFI): The solid portion displayed minimal detectable blood flow signals (color score 1). (C) CEUS (Arterial phase, 20s): The mass exhibited high enhancement with fast wash-in. (D) CEUS (Portal Phase, 42s): The mass demonstrated fast wash-out Fig. 3 Transvaginal ultrasound in a 32-year-old woman revealed a right adnexal mass, initially classified as O-RADS 4. However, based on CEUS findings (score = 3) and a normal CA125 level (7.83 U/ml), the lesion received an OCC-US score of 2 and was subsequently downgraded. Surgical pathology confirmed an endometriotic cyst. ( A ) Gray-scale ultrasound: A slightly hyperechoic mass measuring approximately 3.5 × 2.5 × 2.9 cm was detected. The lesion exhibited a well-defined boundary and a relatively regular shape. ( B ) Color doppler flow imaging (CDFI): A few dot-like and strip-like peripheral blood flow signals were observed, corresponding to a color score of 2. ( C ) CEUS (Arterial phase, 30 s): The lesion margin showed slow, isoenhancement, while the interior demonstrated no contrast perfusion. ( D ) CEUS (Portal phase, 54 s): The lesion margin maintained isoenhancement with mild wash-out (low wash-out) Transvaginal ultrasound in a 32-year-old woman revealed a right adnexal mass, initially classified as O-RADS 4. However, based on CEUS findings (score = 3) and a normal CA125 level (7.83 U/ml), the lesion received an OCC-US score of 2 and was subsequently downgraded. Surgical pathology confirmed an endometriotic cyst. ( A ) Gray-scale ultrasound: A slightly hyperechoic mass measuring approximately 3.5 × 2.5 × 2.9 cm was detected. The lesion exhibited a well-defined boundary and a relatively regular shape. ( B ) Color doppler flow imaging (CDFI): A few dot-like and strip-like peripheral blood flow signals were observed, corresponding to a color score of 2. ( C ) CEUS (Arterial phase, 30 s): The lesion margin showed slow, isoenhancement, while the interior demonstrated no contrast perfusion. ( D ) CEUS (Portal phase, 54 s): The lesion margin maintained isoenhancement with mild wash-out (low wash-out) The inter-observer agreement analysis based on lesions was conducted between a junior sonologist and a senior expert. A total of 50 ovarian adnexal masses were randomly selected for OCC-US scoring and classification. The inter-observer agreement of the OCC-US scoring system (< 3 vs. ≥3) was excellent (κ = 0.840, P  < 0.001) (Table S3). An independent validation cohort consisting of 300 patients was retrospectively collected. In both the development and validation cohorts, the optimal cutoff score for the OCC-US model was 3, demonstrating favorable diagnostic performance. As shown in Fig.  4 A, the AUC in the validation cohort was 86.7%, with a sensitivity of 86.9% and a specificity of 74.8%. The OCC-US model also exhibited good calibration in both cohorts. In the validation cohort, the predicted probabilities showed good agreement with actual outcomes, as indicated by the Hosmer-Lemeshow test ( P  = 0.071) (Fig.  4 B). Furthermore, DCA demonstrated that the net clinical benefit of the OCC-US model in both the development and validation cohorts was markedly superior to the “treat-all” or “treat-none” strategies, suggesting strong potential for clinical decision-making (Fig.  4 C). Fig. 4 Performance evaluation of the OCC-US model in the development and validation cohorts. (A) Receiver operating characteristic (ROC) curves. (B) Calibration curves. (C) Decision curve analysis (DCA) Performance evaluation of the OCC-US model in the development and validation cohorts. (A) Receiver operating characteristic (ROC) curves. (B) Calibration curves. (C) Decision curve analysis (DCA)

In conclusion, the OCC-US model demonstrates excellent diagnostic accuracy (AUC = 0.916) and clinical reliability by integrating O-RADS US v2022, CEUS, and CA125. It enhances sensitivity (84.9%) and specificity (85.4%), significantly reduces false positives (from 23.1 to 6.2%), and may help avoid unnecessary surgeries. These findings support OCC-US as a promising tri-modal approach for preoperative risk stratification of ovarian adnexal masses. Therefore, multicenter prospective studies are needed to further validate the OCC-US model and confirm its generalizability in diverse clinical settings.

Although the O-RADS US v2022 risk stratification system has demonstrated clinical utility and has been externally validated, its specificity remains limited when used alone, which may lead to unnecessary clinical interventions [ 15 , 16 ]. To address this limitation, we developed the OCC-US model by integrating CEUS and CA125, two commonly used diagnostic tools, to enhance accuracy and guide clinical decisions. In this retrospective study, we evaluated the diagnostic accuracy, clinical utility, and inter-observer consistency among sonologists with varying levels of experience. Our findings demonstrate that the OCC-US model outperforms O-RADS US v2022 alone by optimizing risk stratification, reducing false-positive results, improving specificity, and enhancing reliability in identifying ovarian malignancies among practitioners. Additionally, the OCC-US model exhibited robust diagnostic performance in an independent external validation cohort. Given the acknowledged limitations of O-RADS US v2022, particularly its relatively low specificity, researchers have increasingly focused on strategies to improve its diagnostic precision. Integrating additional diagnostic tools such as CEUS and CA125 offers a promising solution. Studies have shown that CEUS achieves a sensitivity and specificity of 90–95% in diagnosing ovarian adnexal masses [ 8 , 11 , 17 – 19 ], outperforming conventional ultrasound [ 18 ]. One study found that O-RADS US v2022 combined with CEUS had significantly higher sensitivity, specificity, and accuracy (96.6%, 91.5%, and 93.0%) compared to O-RADS US v2022 alone (96.6%, 66.2%, and 75.0%), with specificity, accuracy, PPV, and AUC significantly improved ( P  < 0.01) [ 8 ]. Another study confirmed similar findings, showing an AUC of 93.0% for O-RADS US v2022 combined with CEUS, significantly higher than O-RADS US v2022 alone (AUC = 71.0%, P  < 0.001) [ 10 ]. However, CEUS is not incorporated into the O-RADS US v2022 system. CA125 is the first widely recognized serum biomarker. Despite some limitations, it has been the most widely used biomarker for ovarian malignancies, especially epithelial ovarian cancer, over the past few decades [ 20 , 21 ]. In our cohort, we observed a rising trend in CA125 levels among ovarian malignancy patients, which aligns with findings from previous studies. Therefore, we believe that CEUS and CA125 are both important indicators for diagnosing ovarian malignancy and can be combined with the O-RADS US v2022 system as powerful complementary tools. Our results show that the OCC-US scoring system constructed from O-RADS US V2022, CEUS, and CA125 is superior to the individual use of these methods, with a sensitivity of 84.9%, specificity of 85.6%, and an AUC of 91.6%. This approach improves diagnostic specificity while maintaining high diagnostic sensitivity, thereby enhancing the diagnostic accuracy of ovarian adnexal masses. Ultrasound has high sensitivity in detecting ovarian cancer, and the presence of benign imaging features can reliably rule out malignancy [ 1 , 5 , 12 , 22 – 24 ]. Specifically, ultrasound helps differentiate simple cysts, hemorrhagic cysts, endometriomas, dermoids, and other benign tumors from potential malignancies, allowing for appropriate referrals and counseling before any potential surgery [ 25 – 29 ]. If cancer appears unlikely on imaging, the patient can be offered conservative treatment or expectant management [ 25 , 30 ]. Recent studies have shown that before the introduction of the O-RADS US v2022, 42% of ovarian and adnexal lesion cases that underwent surgery without acute symptoms met the criteria for O-RADS 2 classification. This indicates that these patients could have been managed with imaging follow-up or conservative treatment instead of undergoing immediate surgery [ 31 ]. This study underscores the value of the O-RADS US v2022 in reducing unnecessary surgeries. In our study, O-RADS US v2022 was integrated into the OCC-US scoring system, with recommendations aligned with O-RADS US v2022. Among 609 lesions, surgery was recommended for 300 cases (49.3%) based on O-RADS US v2022 and 243 cases (39.9%) based on OCC-US ( P  < 0.001), demonstrating that the OCC-US model significantly reduced unnecessary surgical interventions, contributing to more efficient use of medical resources and reduced psychological burden on patients. Notably, the malignancy rate among patients with an OCC-US score of 2 was 24.2%, which aligns with the malignancy risk defined for O-RADS 4 lesions (10–50%) in the O-RADS US v2022 system. Given the potential for misdiagnosis at this level, further evaluation by expert sonologists or additional imaging modalities such as MRI should be considered to refine diagnosis and guide management decisions more accurately. In addition to the development cohort, we performed external validation using a temporally independent patient cohort from the same institution. Although derived from a single center, the validation cohort represented a distinct patient population assessed by independent reviewers, thereby reducing the risk of information leakage. The OCC-US model maintained consistent diagnostic performance, with an AUC of 0.867 and high sensitivity and specificity, comparable to those observed in the development cohort. Calibration analysis confirmed the model’s reliability in estimating malignancy risk, while the decision curve analysis demonstrated a favorable net clinical benefit across a range of threshold probabilities. Collectively, these results highlight the robust generalizability and potential clinical utility of the OCC-US model within real-world diagnostic workflows. Our study still has some limitations. First, this retrospective study was conducted at a single tertiary center using specific imaging protocols and experienced operators, which may limit the generalizability of the findings. The reliance on existing medical records and imaging data may introduce selection and information bias. Furthermore, although an external validation cohort was included, it was derived from the same institution as the development cohort, potentially leading to center-related bias. Second, the ultrasound images from the included cases were not obtained from the same device, and potential differences in parameter settings may have introduced bias into the results. Third, the ultrasound contrast parameters were interpreted semi-quantitatively, without using quantitative analysis software.

Ovarian masses are histopathologically classified as benign, borderline, or malignant. Due to the insidious onset of ovarian malignancies, early detection is challenging, often delaying diagnosis and treatment. Diagnosis combines history, physical exam, imaging, tumor markers, and pathology, but selecting a reliable risk assessment model remains crucial. In recent years, various domestic and international models have been developed, improving early detection and malignancy evaluation. Ultrasound is a key tool for diagnosing ovarian masses but is highly influenced by subjective factors, with diagnostic standards varying across hospitals. To improve diagnostic consistency, the American College of Radiology (ACR) introduced the Ovarian-Adnexal Reporting and Data System (O-RADS) in 2018, incorporating standardized terminology based on the International Ovarian Tumor Analysis (IOTA) group [ 1 ]. The O-RADS Ultrasound Version 2022 (O-RADS US v2022) represents a further refinement of the original system, providing an essential reference for standardized imaging assessment of ovarian masses [ 2 ]. Although the O-RADS system has demonstrated superior sensitivity compared to other models (O-RADS > IOTA ADNEX > IOTA SR > RMI > GI-RADS) [ 3 – 7 ], there remains room for improvement in its specificity and overall accuracy, particularly in evaluating benign masses such as fibromas, fibrothecoma and teratoma, where a higher false-positive rate may occur [ 8 ]. Early-stage ovarian cancers and borderline tumors often present with ultrasound features resembling benign lesions, which may lead to underestimation of malignancy risk when relying solely on O-RADS US v2022. Therefore, an accurate diagnosis requires a comprehensive approach that integrates clinical characteristics and tumor markers such as CA125 to improve diagnostic sensitivity. Contrast-enhanced ultrasound (CEUS), as a novel non-invasive imaging technique, uses contrast agents to significantly enhance the differences in microvascular perfusion between different lesions. Compared to conventional color Doppler ultrasound, CEUS offers distinct advantages in visualizing microcirculatory blood flow and internal tumor vascularization, thereby improving diagnostic accuracy. Suspected malignant vascular features on CEUS include large, twisted, or irregular vessels, early blood flow perfusion, and uneven or excessive internal enhancement [ 8 ]– [ 9 ]. Studies have shown that combining CEUS with the O-RADS US v2022 system can significantly improve the specificity and overall diagnostic performance in evaluating adnexal masses [ 8 ]. CEUS provides real-time perfusion information that complements the morphological classification of O-RADS, particularly for lesions with indeterminate gray-scale features. Recent evidence indicates that the integration of CEUS can increase the AUC and the differentiation between benign and malignant tumors [ 10 , 11 ]. These findings support CEUS as a valuable complementary tool for refining risk stratification beyond conventional ultrasound assessment. In addition to its diagnostic performance, CEUS has demonstrated a favorable safety profile. It is non-nephrotoxic, free of ionizing radiation, and associated with a very low incidence of adverse reactions in large adult populations [ 12 ]. Moreover, a recent meta-analysis focusing on adnexal masses confirmed both the diagnostic value and safety of CEUS in the evaluation of ovarian lesions [ 13 ]. Therefore, this study aims to integrate O-RADS US v2022, CEUS, and CA125 to develop the OCC-US model, with the goal of improving the diagnostic accuracy in differentiating benign and malignant ovarian adnexal masses. This model is expected to further optimize diagnostic strategies in clinical practice, reduce false-positive rates, and enhance specificity, providing more precise guidance for the early diagnosis and treatment of patients with ovarian masses.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1