Abstract: Gleason grading of histological images is important in risk assessment and treatment planning for prostate cancer patients. Much research has been done in classifying small homogeneous cancer regions within histological images. However, semi-supervised methods published to date depend on pre-selected regions and cannot be easily extended to an image of heterogeneous tissue composition. In this paper, we propose a multi-scale U-Net model to classify images at the pixel-level using 224 histological image tiles from radical prostatectomies of 20 patients. Our model was evaluated by a patient-based 10-fold cross validation, and achieved a mean Jaccard index of 65.8% across 4 classes (stroma, Gleason 3, Gleason 4 and benign glands), and 75.5% for 3 classes (stroma, benign glands, prostate cancer), outperforming other methods.
Learning Objective 1: To develop a deep learning approach for semantic image segmentation (4 classes: stroma, Gleason 3, Gleason 4 and benign glands) of prostate cancer histological images.
Jiayun Li (Presenter)
University of California, Los Angeles
Karthik Sarma, University of California, Los Angeles
King Chung Ho, University of California, Los Angeles
Arkadiusz Gertych, Cedars-Sinai Medical Center
Beatrice Knudsen, Cedars-Sinai Medical Center
Corey Arnold, University of California, Los Angeles