AI-based automated segmentation for ovarian/adnexal masses and their internal components on ultrasound imaging

Abstract

Purpose: Segmentation of ovarian/adnexal masses from surrounding tissue on ultrasound images is a challenging task. The separation of masses into different components may also be important for radiomic feature extraction. Our study aimed to develop an artificial intelligence-based automatic segmentation method for transvaginal ultrasound images that (1) outlines the exterior boundary of adnexal masses and (2) separates internal components.

Approach: A retrospective ultrasound imaging database of adnexal masses was reviewed for exclusion criteria at the patient, mass, and image levels, with one image per mass. The resulting 54 adnexal masses (36 benign/18 malignant) from 53 patients were separated by patient into training (26 benign/12 malignant) and independent test (10 benign/6 malignant) sets. U-net segmentation performance on test images compared to expert detailed outlines was measured using the Dice similarity coefficient (DSC) and the ratio of the Hausdorff distance to the effective diameter of the outline ( $R_{\mathrm{HD}\text{-}D}$ ) for each mass. Subsequently, in discovery mode, a two-level fuzzy c-means (FCM) unsupervised clustering approach was used to separate the pixels within masses belonging to hypoechoic or hyperechoic components.

Results: The DSC (median [95% confidence interval]) was 0.91 [0.78, 0.96], and $R_{\mathrm{HD}\text{-}D}$ was 0.04 [0.01, 0.12], indicating strong agreement with expert outlines. Clinical review of the internal separation of masses into echogenic components demonstrated a strong association with mass characteristics.

Conclusion: A combined U-net and FCM algorithm for automatic segmentation of adnexal masses and their internal components achieved excellent results compared with expert outlines and review, supporting future radiomic feature-based classification of the masses by components.

Keywords: adnexal diseases; deep learning; machine learning; ovarian cancer; segmentation; ultrasound.

Fig. 1

Consort diagram reporting the selection of adnexal masses used in the study. The final cohort for the U-net segmentation study was 54 unique masses (from 53 patients).

Fig. 2

Workflow for supervised U-net segmentation of the adnexal mass.

Fig. 3

Workflow for segmentation of masses into internal components using an unsupervised FCM algorithm.

Fig. 4

U-net segmentation performance in the test set in the task of segmenting the entire adnexal mass from the surrounding tissue, compared to expert outlines. ($R_{\mathrm{HD}\text{-}\mathrm{D}}$: ratio of the average Hausdorff distance to the effective diameter of the mass.) Images of the four masses with the best performance (highest Dice coefficient and lowest $R_{\mathrm{HD}\text{-}\mathrm{D}}$) and lowest performance (lowest Dice coefficient and highest $R_{\mathrm{HD}\text{-}\mathrm{D}}$) are shown. Clockwise from top left with pathology (with patient diagnosis) details: borderline ovarian mass (borderline serous tumor), epithelial ovarian mass (benign serous cystadenoma), epithelial ovarian cancer (high-grade serous ovarian cancer), and metastasis to the ovaries (cancer of gastro-intestinal primary origin). B, benign and M, malignant.

Fig. 5

(a)–(j) Example results of internal component segmentation using an unsupervised FCM algorithm. Mass pathology subtypes are as given in Table 1 (i.e., the pathology subtypes used for dataset splitting). Specific patient diagnoses for these masses are described further in the text. B, benign and M, malignant.