A machine learning model for separating epithelial and stromal regions in oral cavity squamous cell carcinomas using H&E-stained histology images: A multi-center, retrospective study. (August 2022)

• Record Type: Journal Article
• Authors: Wu, Yuxin; Koyuncu, Can F.; Toro, Paula; Corredor, German; Feng, Qianyu; Buzzy, Christina; Old, Matthew; Teknos, Theodoros; Connelly, Stephen Thaddeus; Jordan, Richard C.; Lang Kuhs, Krystle A.; Lu, Cheng; Lewis, James S.; Madabhushi, Anant
• Is Part Of: Oral oncology. Volume 131(2022)
• Publication Date: 2022-08
• Subjects: Mouth -- Cancer -- Periodicals; Mouth -- Tumors -- Periodicals; Mouth Diseases -- Periodicals; Mouth Neoplasms -- Periodicals; Bouche -- Cancer -- Périodiques; Bouche -- Tumeurs -- Périodiques; Tumeurs -- Périodiques; Electronic journals; 616.9943105
• DOI: 10.1016/j.oraloncology.2022.105942
• Languages: English
• ISSNs: 1368-8375

Abstract

Graphical abstract: Highlights: Deep learning model achieved a consistently accurate performance for epithelium segmentation The model trained on 10x magnification images achieved the best performance. Morphologic features extracted from human- and AI-annotated epithelial regions are equivalent. Abstract: Objective: Tissue slides from Oral cavity squamous cell carcinoma (OC-SCC), particularly the epithelial regions, hold morphologic features that are both diagnostic and prognostic. Yet, previously developed approaches for automated epithelium segmentation in OC-SCC have not been independently tested in a multi-center setting. In this study, we aimed to investigate the effectiveness and applicability of a convolutional neural network (CNN) model to perform epithelial segmentation using digitized H&E-stained diagnostic slides from OC-SCC patients in a multi-center setting. Methods: A CNN model was developed to segment the epithelial regions of digitized slides (n = 810), retrospectively collected from five different centers. Deep learning models were trained and validated using well-annotated tissue microarray (TMA) images (n = 212) at various magnifications. The best performing model was locked down and used for independent testing with a total of 478 whole-slide images (WSIs). Manually annotated epithelial regions were used as the reference standard for evaluation. We also compared the model generated results with IHC-stained epithelium (n = 120) as the reference. Results: The locked-down CNN model trained on the TMA image training cohorts with 10x magnification achieved the best segmentation performance. The locked-down model performed consistently and yielded Pixel Accuracy, Recall Rate, Precision Rate, and Dice Coefficient that ranged from 95.8% to 96.6%, 79.1% to 93.8%, 85.7% to 89.3%, and 82.3% to 89.0%, respectively for the three independent testing WSI cohorts. Conclusion: The automated model achieved a consistently accurate performance for automated epithelial region segmentation compared to manual annotations. This model could be integrated into a computer-aided diagnosis or prognosis system.