Utility of DWI with quantitative ADC values in ovarian tumors: a meta-analysis of diagnostic test performance

Apparent diffusion coefficient analysis of solid tissue helps distinguish borderline from invasive malignant adnexal masses rated O-RADS MRI 4

PURPOSE

The purpose of this study was to evaluate the contribution of apparent diffusion coefficient (ADC) analysis of the solid tissue of adnexal masses to optimize tumor characterization and possibly refine the risk stratification of the O-RADS MRI 4 category.

MATERIALS AND METHODS

The EURAD cohort was retrospectively analyzed to select all patients with an adnexal mass with solid tissue and feasible ADC measurements. Two radiologists independently measured the ADC values of solid tissue, excluding necrotic areas, surrounding structures, and magnetic susceptibility artifacts. Significant differences in diffusion quantitative parameters in the overall population and according to the morphological aspect of solid tissue were analyzed to identify its impact on ADC reliability. Receiver operating characteristics curve (ROC) was used to determine the optimum cutoff of the ADC for distinguishing invasive from non-invasive tumors in the O-RADS MRI score 4 population.

RESULTS

The final study population included 180 women with a mean age of 57 ± 15.5 (standard deviation) years; age range: 19-95 years) with 93 benign, 23 borderline, and 137 malignant masses. The median ADC values of solid tissue was greater in borderline masses (1.310 × 10-3 mm2/s (Q1, Q3: 1.152, 1.560 × 10-3 mm2/s) than in benign masses (1.035 × 10-3 mm2/s; Q1, Q3: 0.900, 1.560 × 10-3 mm2/s) (P= 0.002) and in benign tumors compared by comparison with invasive masses (0.850 × 10-3 mm2/s; Q1, Q3: 0.750, 0.990 × 10-3 mm2/s) (P < 0.001). Solid tissue corresponded to irregular septa or papillary projection in 18.6% (47/253), to a mural nodule or a mixed mass in 46.2% (117/253), and to a purely solid mass in 35.2% (89/253) of adnexal masses. In mixed masses or masses with mural nodule subgroup, invasive masses had a significantly lower ADC (0.830 × 10-3 mm2/s (Q1, Q3: 0.738, 0.960) than borderline (1.385; Q1, Q3: 1.300, 1.930) (P= 0.0012) and benign masses (P= 0.04). An ADC cutoff of 1.08 × 10-3 mm2/s yielded 71.4% sensitivity and 100% specificity for identifying invasive lesions in the mixed or mural nodule subgroup with an AUC of 0.92 (95% confidence interval: 0.76-0.99).

CONCLUSION

ADC analysis of solid tissue of adnexal masses could help distinguish invasive masses within the O-RADS MRI 4 category, especially in mixed masses or those with mural nodule.

Diagnostic performance of a modified O-RADS classification system for adnexal lesions incorporating clinical features

PURPOSE

To compare the diagnostic efficacy of the Ovarian-Adnexal Reporting and Data System (O-RADS) MRI score with that of the modified O-RADS score on the basis of a simplified contrast-enhanced (CE) MRI protocol in characterizing adnexal masses with solid tissue. The added value of clinical features was evaluated to improve the ability of the scoring system to classify adnexal masses.

METHODS

A total of 124 patients with 124 adnexal lesions containing solid tissue were included in this two-center retrospective study. Among them, there were 40 benign lesions (40/124, 32.3%) and 84 were malignant lesions (84/124, 67.7%). Three radiologists independently reviewed the images and assigned the O-RADS MRI score and the modified O-RADS score for each adnexal mass. Histopathology was used as the reference standard. The diagnostic efficacy of the two scoring methods was compared. Univariate and multivariate logistic regression were performed to evaluate the value of significant features in the prediction of malignant tumors.

RESULTS

The O-RADS MRI score and modified O-RADS score showed sensitivity at 100.0% (95% CI, 95.7-100.0%) and 71.4% (95% CI, 60.5-80.8%), specificity at 12.5% (95% CI, 4.2-26.8%) and 75.0% (95% CI, 58.8-87.3%), respectively. The area under the curve of the modified O-RADS score was higher than the O-RADS score (0.732 [95% CI, 0.645-0.808] vs 0.575 [95% CI, 0.483-0.663]; p < 0.001). Multivariate analysis showed that the modified O-RADS score 4b or 5 combined with patient age > 38.5 years, nullipara, maximum diameter > 40.5 mm and HE4 > 78.9 pmol/L significantly improved the diagnostic efficacy up to 0.954 (95% CI, 0.901-0.984) (p < 0.001).

CONCLUSION

A modified O-RADS score combined with certain clinical features can significantly improve the diagnostic efficacy in predicting malignant tumors.

Role of magnetic resonance imaging to differentiate between borderline and malignant serous epithelial ovarian tumors

PURPOSE

We aimed to differentiate serous borderline ovarian tumors (SBOT) from serous epithelial ovarian carcinomas (SEOC) using morphological and functional MRI findings, to improve the patient management.

METHOD

We retrospectively investigated 24 ovarian lesions diagnosed with SBOT and 64 ovarian lesions diagnosed with SEOC. Additional to the demographic and morphological findings T2W signal intensity ratio, mean apparent diffusion coefficient (ADCmean) and total apparent diffusion coefficient (ADCtotal) values were analyzed and compared between two groups.

RESULTS

Bilaterality, pelvic free fluid presence, serum CA-125 level (U/mL), presence of pelvic peritoneal implant were in favor of SEOC. Lower maximum size of solid component and solid size to maximum size ratio, dominantly cystic and solid-cystic appearance, exophytic growth pattern, presence of papiller projection and papillary architecture and internal branching pattern, higher T2W signal intensity ratio, ADCmean and ADCtotal values were in favor of SBOT.

CONCLUSION

Our study revealed that morphological and functional imaging findings were valuable in differentiating BSOT from SEOC.

Clinical value of minimum apparent diffusion coefficient for prediction of clinically significant prostate cancer in the transition zone

BACKGROUND

This study investigated the association between apparent diffusion coefficients in Prostate Imaging Reporting and Data System 4/5 lesions and clinically significant prostate cancer in the transition zone.

METHODS

We included 102 patients who underwent transperineal cognitive fusion targeted biopsy for Prostate Imaging Reporting and Data System 4/5 lesions in the transition zone between 2016 and 2020. The association between apparent diffusion coefficients and prostate cancers in the transition zone was analyzed.

RESULTS

The detection rate of prostate cancer was 49% (50/102), including clinically significant prostate cancer in 37.3% (38/102) of patients. The minimum apparent diffusion coefficients in patients with clinically significant prostate cancer were 494.5 ± 133.6 µm2/s, which was significantly lower than 653.8 ± 172.5 µm2/s in patients with benign histology or clinically insignificant prostate cancer. Age, prostate volume, transition zone volume, and mean and minimum apparent diffusion coefficients were associated with clinically significant prostate cancer. Multivariate analysis demonstrated that only the minimum apparent diffusion coefficient value (odds ratio: 0.994; p < 0.001) was an independent predictor of clinically significant prostate cancer. When the cutoff value of the minimum apparent diffusion coefficient was less than 595 µm2/s, indicating the presence of prostate cancer in the transition zone, the detection rate increased to 59.2% (29/49) in this cohort.

CONCLUSION

The minimum apparent diffusion coefficient provided additional value to indicate the presence of clinically significant prostate cancer in the transition zone. It may help consider the need for subsequent biopsies in patients with Prostate Imaging Reporting and Data System 4/5 lesions and an initial negative targeted biopsy.

Neoadjuvant chemotherapy induces an elevation of tumour apparent diffusion coefficient values in patients with ovarian cancer

OBJECTIVES

Multiparametric magnetic resonance imaging (mMRI) is the modality of choice in the imaging of ovarian cancer (OC). We aimed to investigate the feasibility of different types of regions of interest (ROIs) in the measurement of apparent diffusion coefficient (ADC) values of diffusion-weighted imaging in OC patients treated with neoadjuvant chemotherapy (NACT).

METHODS

We retrospectively enrolled 23 consecutive patients with advanced OC who had undergone NACT and mMRI. Seventeen of them had been imaged before and after NACT. Two observers independently measured the ADC values in both ovaries and in the metastatic mass by drawing on a single slice of (1) freehand large ROIs (L-ROIs) covering the solid parts of the whole tumour and (2) three small round ROIs (S-ROIs). The side of the primary ovarian tumour was defined. We evaluated the interobserver reproducibility and statistical significance of the change in tumoural pre- and post-NACT ADC values. Each patient's disease was defined as platinum-sensitive, semi-sensitive, or resistant. The patients were deemed either responders or non-responders.

RESULTS

The interobserver reproducibility of the L-ROI and S-ROI measurements ranged from good to excellent (ICC range: 0.71-0.99). The mean ADC values were significantly higher after NACT in the primary tumour (L-ROI p  < 0.001, S-ROIs p < 0.01), and the increase after NACT was associated with sensitivity to platinum-based chemotherapy. The changes in the ADC values of the omental mass were associated with a response to NACT.

CONCLUSION

The mean ADC values of the primary tumour increased significantly after NACT in the OC patients, and the amount of increase in omental mass was associated with the response to platinum-based NACT. Our study indicates that quantitative analysis of ADC values with a single slice and a whole tumour ROI placement is a reproducible method that has a potential role in the evaluation of NACT response in patients with OC.

TRIAL REGISTRATION

Retrospectively registered (institutional permission code: 5302501; date of the permission: 31.7.2020).

Diffusion-Weighted Magnetic Resonance Imaging in Ovarian Cancer: Exploiting Strengths and Understanding Limitations

Detection, characterization, staging, and response assessment are key steps in the imaging pathway of ovarian cancer. The most common type, high grade serous ovarian cancer, often presents late, so that accurate disease staging and response assessment are required through imaging in order to improve patient management. Currently, computerized tomography (CT) is the most common method for these tasks, but due to its poor soft-tissue contrast, it is unable to quantify early response within lesions before shrinkage is observed by size criteria. Therefore, quantifiable techniques, such as diffusion-weighted magnetic resonance imaging (DW-MRI), which generates high contrast between tumor and healthy tissue, are increasingly being explored. This article discusses the basis of diffusion-weighted contrast and the technical issues that must be addressed in order to achieve optimal implementation and robust quantifiable diffusion-weighted metrics in the abdomen and pelvis. The role of DW-MRI in characterizing adnexal masses in order to distinguish benign from malignant disease, and to differentiate borderline from frankly invasive malignancy is discussed, emphasizing the importance of morphological imaging over diffusion-weighted metrics in this regard. Its key role in disease staging and predicting resectability in comparison to CT is addressed, including its valuable use as a biomarker for following response within individual lesions, where early changes in the apparent diffusion coefficient in peritoneal metastases may be detected. Finally, the task of implementing DW-MRI into clinical trials in order to validate this biomarker for clinical use are discussed, along with the trials that include it within their protocols.

Diagnosing Ovarian Cancer on MRI: A Preliminary Study Comparing Deep Learning and Radiologist Assessments

Simple Summary

As a preliminary experiment to explore the possibility of clinical application as a future reading assist, we present CNNs for the diagnosis of ovarian carcinomas and borderline tumors on MRI, including T2WI, DWI, ADC map, and CE-T1WI, and compare their diagnostic performance with interpretations by experienced radiologists. CNNs were trained using 1798 images from 146 patients and 1865 images from 219 patients with malignant tumors, including borderline tumors, and non-malignant lesions, respectively, for each MRI sequence and tested with 48 and 52 images of patients with malignant and non-malignant lesions. The CNN of each sequence had a sensitivity of 0.77–0.85, specificity of 0.77–0.92, accuracy of 0.81–0.87, and an AUC of 0.83–0.89, demonstrating diagnostic performances that were non-inferior to those of experienced radiologists, and the CNN showed the highest diagnostic performance on the ADC map for each sequence (specificity = 0.85; sensitivity = 0.77; accuracy = 0.81; AUC = 0.89).

Abstract

Background: This study aimed to compare deep learning with radiologists’ assessments for diagnosing ovarian carcinoma using MRI. Methods: This retrospective study included 194 patients with pathologically confirmed ovarian carcinomas or borderline tumors and 271 patients with non-malignant lesions who underwent MRI between January 2015 and December 2020. T2WI, DWI, ADC map, and fat-saturated contrast-enhanced T1WI were used for the analysis. A deep learning model based on a convolutional neural network (CNN) was trained using 1798 images from 146 patients with malignant tumors and 1865 images from 219 patients with non-malignant lesions for each sequence, and we tested with 48 and 52 images of patients with malignant and non-malignant lesions, respectively. The sensitivity, specificity, accuracy, and AUC were compared between the CNN and interpretations of three experienced radiologists. Results: The CNN of each sequence had a sensitivity of 0.77–0.85, specificity of 0.77–0.92, accuracy of 0.81–0.87, and an AUC of 0.83–0.89, and it achieved a diagnostic performance equivalent to the radiologists. The CNN showed the highest diagnostic performance on the ADC map among all sequences (specificity = 0.85; sensitivity = 0.77; accuracy = 0.81; AUC = 0.89). Conclusion: The CNNs provided a diagnostic performance that was non-inferior to the radiologists for diagnosing ovarian carcinomas on MRI.

The role of "penumbra sign" and diffusion-weighted imaging in adnexal masses: do they provide a clue in differentiating tubo-ovarian abscess from ovarian malignancy?

Purpose

To evaluate the role of “penumbra sign”, diffusion-weighted imaging (DWI), and the apparent diffusion coefficient (ADC) value in differentiating tubo-ovarian abscess (TOA) from ovarian malignancy.

Material and methods

Thirty-six patients with 50 adnexal masses (tubo-ovarian abscess, n = 24; ovarian malignancy, n = 26), who underwent magnetic resonance imaging (MRI) with DWI, were retrospectively evaluated. “Penumbra sign” (hyperintense rim on T1W images), diffusion restriction, and mean apparent diffusion coefficient (ADC) values from cystic (c-ADC) and solid (s-ADC) components were evaluated for all the masses.

Results

“Penumbra sign” on T1W images was significantly more common in the TOA group (n = 21, 87.5%) than in the ovarian malignancy group (n = 2, 7.7%) (p < 0.001). Similarly, diffusion restriction in the cystic component was more frequent in the TOA group (n = 24, 100% vs. n = 2, 10.5%; p < 0.001). In contrast, diffusion restriction in the solid component was more common in the ovarian malignancy group (n = 5, 20.8% vs. n = 26, 100%; p < 0.001). The mean c-ADC value was significantly lower in TOAs (p < 0.001). A c-ADC value of 1.31 × 10^-3 mm²/s may be an optimal cut-off in distinguishing TOAs from ovarian malignancies. Conversely, the mean s-ADC value was significantly lower in the ovarian malignancy group (p < 0.001). An s-ADC value of 0.869 × 10^-3 mm²/s may be an optimal cut-off in differentiating ovarian malignancies from TOAs (p < 0.001). ROC curve analysis showed that c-ADC values had a higher diagnostic accuracy than s-ADC values.

Conclusions

“Penumbra sign” on T1W images, diffusion characteristics, and ADC values provide important clues in addition to conventional MR imaging features in differentiating TOA from ovarian malignancy.

An Application of Machine Learning That Uses the Magnetic Resonance Imaging Metric, Mean Apparent Diffusion Coefficient, to Differentiate between the Histological Types of Ovarian Cancer

This retrospective single-center study included patients diagnosed with epithelial ovarian cancer (EOC) using preoperative pelvic magnetic resonance imaging (MRI). The apparent diffusion coefficient (ADC) of the axial MRI maps that included the largest solid portion of the ovarian mass was analysed. The mean ADC values (ADC_mean) were derived from the regions of interest (ROIs) of each largest solid portion. Logistic regression and three types of machine learning (ML) applications were used to analyse the ADCs and clinical factors. Of the 200 patients, 103 had high-grade serous ovarian cancer (HGSOC), and 97 had non-HGSOC (endometrioid carcinoma, clear cell carcinoma, mucinous carcinoma, and low-grade serous ovarian cancer). The median ADC_mean of patients with HGSOC was significantly lower than that of patients without HGSOCs. Low ADC_mean and CA 19-9 levels were independent predictors for HGSOC over non-HGSOC. Compared to stage I disease, stage III disease was associated with HGSOC. Gradient boosting machine and extreme gradient boosting machine showed the highest accuracy in distinguishing between the histological findings of HGSOC versus non-HGSOC and between the five histological types of EOC. In conclusion, ADC_mean, disease stage at diagnosis, and CA 19-9 level were significant factors for differentiating between EOC histological types.

MRI-Based Multiple Instance Convolutional Neural Network for Increased Accuracy in the Differentiation of Borderline and Malignant Epithelial Ovarian Tumors

BACKGROUND

Preoperative differentiation of borderline from malignant epithelial ovarian tumors (BEOT vs. MEOT) is challenging and can significantly impact surgical management.

PURPOSE

To develop a multiple instance convolutional neural network (MICNN) that can differentiate BEOT from MEOT, and to compare its diagnostic performance with that of radiologists.

STUDY TYPE

Retrospective study of eight clinical centers.

SUBJECTS

Between January 2010 and June 2018, a total of 501 women (mean age, 48.93 ± 14.05 years) with histopathologically confirmed BEOT (N = 165) or MEOT (N = 336) were divided into the training (N = 342) and validation cohorts (N = 159).

FIELD STRENGTH/SEQUENCE

Three axial sequences from 1.5 or 3 T scanner were used: fast spin echo T2-weighted imaging with fat saturation (T2WI FS), echo planar diffusion-weighted imaging, and 2D volumetric interpolated breath-hold examination of contrast-enhanced T1-weighted imaging (CE-T1WI) with FS.

ASSESSMENT

Three monoparametric MICNN models were built based on T2WI FS, apparent diffusion coefficient map, and CE-T1WI. Based on these monoparametric models, we constructed an early multiparametric (EMP) model and a late multiparametric (LMP) model using early and late information fusion methods, respectively. The diagnostic performance of the models was evaluated using the receiver operating characteristic (ROC) curve and compared to the performance of six radiologists with varying levels of experience.

STATISTICAL TESTS

We used DeLong test, chi-square test, Mann-Whitney U-test, and t-test, with significance level of 0.05.

RESULTS

Both EMP and LMP models differentiated BEOT from MEOT, with an area under the ROC curve (AUC) of 0.855 (95% CI, 0.795-0.915) and 0.884 (95% CI, 0.831-0.938), respectively. The AUC of the LMP model was significantly higher than the radiologists' pooled AUC (0.884 vs. 0.797).

DATA CONCLUSION

The developed MICNN models can effectively differentiate BEOT from MEOT and the diagnostic performances (AUCs) were more superior than that of the radiologists' assessments.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY STAGE: 2.

Diagnostic performance of apparent diffusion coefficient (ADC) for differentiating endometrial carcinoma from benign lesions: a systematic review and meta-analysis

To determine the diagnostic performance of mean ADC values in the characterization of endometrial carcinoma (EC) from benign lesions by systematic review of the literature and performing meta-analysis. A systematic search of major electronic bibliographic databases was performed to find studies that used ADC values for differentiating EC from benign lesions. Two reviewers independently screened the titles and abstracts of the search results and then by reading the full texts selected the pertinent studies for final analyses. A bivariate random-effects model with pooled sensitivity and specificity values with 95% CI (confidence interval) was used. Summary receiver operating characteristic (SROC) curve and area under curve (AUC) were created. Between-study heterogeneity was measured using I squared (I2) index. Eleven studies including 269 ECs and 208 benign lesions were analyzed. Pooled average (95% CI) ADC in EC and benign lesions groups were, respectively, 0.82 (0.77-0.87) × 10-3 mm2/s and 1.41 (1.29-1.52) × 10-3 mm2/s. The combined (95% CI) sensitivity and specificity of mean ADC values for differentiating EC from benign lesions were 93% (87-96%; I2 = 41.19%) and 94% (88-97%; I2 = 46.91%), respectively. The AUC (95% CI) of the SROC curve was 98% (96-99%). ADC values had good diagnostic accuracy for differentiating EC from benign lesions. In order to recommend ADC measurement for detecting endometrial lesions in routine clinical practice, more primary studies, especially trials and comparative studies including hysteroscopically-guided biopsy method, with larger sample sizes are still required.

Prediction of Platinum-based Chemotherapy Response in Advanced High-grade Serous Ovarian Cancer: ADC Histogram Analysis of Primary Tumors

RATIONALE AND OBJECTIVES

To investigate the feasibility of apparent diffusion coefficient (ADC) histogram analysis of primary advanced high-grade serous ovarian cancer (HGSOC) to predict patient response to platinum-based chemotherapy.

MATERIALS AND METHODS

A total of 70 patients with 102 advanced stage HGSOCs (International Federation of Gynecology and Obstetrics (FIGO) stages III-IV) who received standard treatment of primary debulking surgery followed by the first line of platinum-based chemotherapy were retrospectively enrolled. Patients were grouped as platinum-resistant and platinum-sensitive according to whether relapse occurred within 6 months. Clinical characteristics, including age, pretherapy CA125 level, International Federation of Gynecology and Obstetrics stage, residual tumor, and histogram parameters derived from whole tumor and solid component such as ADC_mean; 10th, 20th, 25th, 30th, 40th, 50th, 60th, 70th, 75th, 80th, 90th percentiles; skewness and kurtosis, were compared between platinum-resistant and platinum-sensitive groups.

RESULTS

No significantly different clinical characteristics were observed between platinum-sensitive and platinum-resistant patients. There were no significant differences in any whole-tumor histogram-derived parameters between the two groups. Significantly higher ADC_mean and percentiles and significantly lower skewness and kurtosis from the solid-component histogram parameters were observed in the platinum-sensitive group when compared with the platinum-resistant group. ADC_mean, skewness and kurtosis showed moderate prediction performances, with areas under the curve of 0.667, 0.733 and 0.616, respectively. Skewness was an independent risk factor for platinum resistance.

CONCLUSION

Pretreatment ADC histogram analysis of primary tumors has the potential to allow prediction of response to platinum-based chemotherapy in patients with advanced HGSOC.

Evaluation of a convolutional neural network for ovarian tumor differentiation based on magnetic resonance imaging

OBJECTIVES

There currently lacks a noninvasive and accurate method to distinguish benign and malignant ovarian lesion prior to treatment. This study developed a deep learning algorithm that distinguishes benign from malignant ovarian lesion by applying a convolutional neural network on routine MR imaging.

METHODS

Five hundred forty-five lesions (379 benign and 166 malignant) from 451 patients from a single institution were divided into training, validation, and testing set in a 7:2:1 ratio. Model performance was compared with four junior and three senior radiologists on the test set.

RESULTS

Compared with junior radiologists averaged, the final ensemble model combining MR imaging and clinical variables had a higher test accuracy (0.87 vs 0.64, p < 0.001) and specificity (0.92 vs 0.64, p < 0.001) with comparable sensitivity (0.75 vs 0.63, p = 0.407). Against the senior radiologists averaged, the final ensemble model also had a higher test accuracy (0.87 vs 0.74, p = 0.033) and specificity (0.92 vs 0.70, p < 0.001) with comparable sensitivity (0.75 vs 0.83, p = 0.557). Assisted by the model's probabilities, the junior radiologists achieved a higher average test accuracy (0.77 vs 0.64, Δ = 0.13, p < 0.001) and specificity (0.81 vs 0.64, Δ = 0.17, p < 0.001) with unchanged sensitivity (0.69 vs 0.63, Δ = 0.06, p = 0.302). With the AI probabilities, the junior radiologists had higher specificity (0.81 vs 0.70, Δ = 0.11, p = 0.005) but similar accuracy (0.77 vs 0.74, Δ = 0.03, p = 0.409) and sensitivity (0.69 vs 0.83, Δ = -0.146, p = 0.097) when compared with the senior radiologists.

CONCLUSIONS

These results demonstrate that artificial intelligence based on deep learning can assist radiologists in assessing the nature of ovarian lesions and improve their performance.

KEY POINTS

• Artificial Intelligence based on deep learning can assess the nature of ovarian lesions on routine MRI with higher accuracy and specificity than radiologists. • Assisted by the deep learning model's probabilities, junior radiologists achieved better performance that matched those of senior radiologists.

Ultra-High-b-Value Kurtosis Imaging for Noninvasive Tissue Characterization of Ovarian Lesions

Background MRI with contrast material enhancement is the imaging modality of choice to evaluate sonographically indeterminate adnexal masses. The role of diffusion-weighted MRI, however, remains controversial. Purpose To evaluate the diagnostic performance of ultra-high-b-value diffusion kurtosis MRI in discriminating benign and malignant ovarian lesions. Materials and Methods This prospective cohort study evaluated consecutive women with sonographically indeterminate adnexal masses between November 2016 and December 2018. MRI at 3.0 T was performed, including diffusion-weighted MRI (b values of 0-2000 sec/mm²). Lesions were segmented on b of 1500 sec/mm² by two readers in consensus and an additional independent reader by using full-lesion segmentations on a single transversal slice. Apparent diffusion coefficient (ADC) calculation and kurtosis fitting were performed. Differences in ADC, kurtosis-derived ADC (D_app), and apparent kurtosis coefficient (K_app) between malignant and benign lesions were assessed by using a logistic mixed model. Area under the receiver operating characteristic curve (AUC) for ADC, D_app, and K_app to discriminate malignant from benign lesions was calculated, as was specificity at a sensitivity level of 100%. Results from two independent reads were compared. Histopathologic analysis served as the reference standard. Results A total of 79 ovarian lesions in 58 women (mean age ± standard deviation, 48 years ± 14) were evaluated. Sixty-two (78%) lesions showed benign and 17 (22%) lesions showed malignant histologic findings. ADC and D_app were lower and K_app was higher in malignant lesions: median ADC, D_app, and K_app were 0.74 µm²/msec (range, 0.52-1.44 µm²/msec), 0.98 µm²/msec (range, 0.63-2.12 µm²/msec), and 1.01 (range, 0.69-1.30) for malignant lesions, and 1.13 µm²/msec (range, 0.35-2.63 µm²/msec), 1.45 µm²/msec (range, 0.44-3.34 µm²/msec), and 0.65 (range, 0.44-1.43) for benign lesions (P values of .01, .02, < .001, respectively). AUC for K_app of 0.85 (95% confidence interval: 0.77, 0.94) was higher than was AUC from ADC of 0.78 (95% confidence interval: 0.67, 0.89; P = .047). Conclusion Diffusion-weighted MRI by using quantitative kurtosis variables is superior to apparent diffusion coefficient values in discriminating benign and malignant ovarian lesions and might be of future help in clinical practice, especially in patients with contraindication to contrast media application. © RSNA, 2020 Online supplemental material is available for this article.

MRI-Based Machine Learning for Differentiating Borderline From Malignant Epithelial Ovarian Tumors: A Multicenter Study

BACKGROUND

Preoperative differentiation of borderline from malignant epithelial ovarian tumors (BEOT from MEOT) can impact surgical management. MRI has improved this assessment but subjective interpretation by radiologists may lead to inconsistent results.

PURPOSE

To develop and validate an objective MRI-based machine-learning (ML) assessment model for differentiating BEOT from MEOT, and compare the performance against radiologists' interpretation.

STUDY TYPE

Retrospective study of eight clinical centers.

POPULATION

In all, 501 women with histopathologically-confirmed BEOT (n = 165) or MEOT (n = 336) from 2010 to 2018 were enrolled. Three cohorts were constructed: a training cohort (n = 250), an internal validation cohort (n = 92), and an external validation cohort (n = 159).

FIELD STRENGTH/SEQUENCE

Preoperative MRI within 2 weeks of surgery. Single- and multiparameter (MP) machine-learning assessment models were built utilizing the following four MRI sequences: T₂ -weighted imaging (T₂ WI), fat saturation (FS), diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC), and contrast-enhanced (CE)-T₁ WI.

ASSESSMENT

Diagnostic performance of the models was assessed for both whole tumor (WT) and solid tumor (ST) components. Assessment of the performance of the model in discriminating BEOT vs. early-stage MEOT was made. Six radiologists of varying experience also interpreted the MR images.

STATISTICAL TESTS

Mann-Whitney U-test: significance of the clinical characteristics; chi-square test: difference of label; DeLong test: difference of receiver operating characteristic (ROC).

RESULTS

The MP-ST model performed better than the MP-WT model for both the internal validation cohort (area under the curve [AUC] = 0.932 vs. 0.917) and external validation cohort (AUC = 0.902 vs. 0.767). The model showed capability in discriminating BEOT vs. early-stage MEOT, with AUCs of 0.909 and 0.920, respectively. Radiologist performance was considerably poorer than both the internal (mean AUC = 0.792; range, 0.679-0.924) and external (mean AUC = 0.797; range, 0.744-0.867) validation cohorts.

DATA CONCLUSION

Performance of the MRI-based ML model was robust and superior to subjective assessment of radiologists. If our approach can be implemented in clinical practice, improved preoperative prediction could potentially lead to preserved ovarian function and fertility for some women.

LEVEL OF EVIDENCE

Level 4.

TECHNICAL EFFICACY

Stage 2. J. Magn. Reson. Imaging 2020;52:897-904.

Value of normalized apparent diffusion coefficients in differentiating between borderline and malignant epithelial ovarian tumors

PURPOSE

To compare the diagnostic performance of normalized apparent diffusion coefficients (nADCs) of different references with that of ADCs at differentb factors in differentiating borderline epithelial ovarian tumors (BEOTs) from malignant epithelial ovarian tumors (MEOTs).

METHOD

This retrospective study included 53 BEOTs and 148 MEOTs. Conventional magnetic resonance and diffusion-weighted imaging withb factors of 800 and 1000s/mm² were performed. ADC was measured three times at solid components of tumors, gluteus maximus muscle (GMM), iliopsoas muscle (IM) and urine and then averaged. ADC_tumor, nADCs were then obtained. Differences and the diagnostic performance of ADC_tumor and nADCs between BEOTs and MEOTs with different b factors were compared.

RESULTS

ADC_tumor, nADCs regardless of b factors were significantly higher in BEOTs than MEOTs. The diagnostic performance of nADC_urine regardless of b factors was significantly larger than that of nADC_GMM and nADC_IM. There was no significant difference in the diagnostic performance between ADC_tumor and nADC_urine regardless of b factors. A significantly lower ADC_tumor and a larger diagnostic performance for ADC_tumor was found with a b factor of 1000s/mm² than 800 s/mm². There were no significant differences in nADC_urine of BEOTs or MEOTs or in the diagnostic performance of nADC_urine with b factors between 800 and 1000s/mm².

CONCLUSIONS

ADC_tumor and nADCs were both capable of differentiating BEOTs from MEOTs. nADC_urine was the best of all nADCs and was superior to ADC_tumor because of its stable performance in differentiating BEOTs from MEOTs, regardless of b factors.