Fovea-UNet: Detection and Segmentation of Lymph Node Metastases in Colorectal Cancers with Deep Learning

Abstract

Abstract Objective: Colorectal cancer is one of the most serious malignant tumors, and lymph node metastasis (LNM) from colorectal cancer is a major factor for patient management and prognosis. Accurate image detection of LNM is an important task to help pathologists diagnose cancer. However, effective image detection with the whole slide image (WSI) can only be performed by patch-based classification method, which are far from enough for cancer region segmentation and location due to a small patch image has less non-local contextual information. Recently, the U-Net architecture has been widely used to segment image to accomplish more precise cancer diagnosis. In this work, we aggregate the detailed and non-local contextual information into a U-Net baseline to segment the important region with high diagnostic value. Method: Inspired by the working principle of Fovea in visual neuroscience, a novel network framework based on U-Net for cancer segmentation named Fovea-UNet is proposed to adaptively adjust the resolution according to the importance-aware of information and selectively focuses on the region most relevant to colorectal LNM. Specifically, we design an effective adaptively optimized pooling operation called Fovea Pooling (FP), which dynamically aggregate the detailed and non-local contextual information according to pixel-level feature importance. In addition, the improved lightweight backbone network based on GhostNet is adopted to reduce the computational cost caused by FP pooling. Results & Conclusions: Experimental results show that our proposed framework can achieve higher performance than other state-of-the-art segmentation networks with 92.82% sensitivity and 88.51% F1 score on the LNM dataset. Clinical impact: The proposed framework can provide a valid tool for cancer diagnosis, especially for LNM of colorectal cancer.