A Loopback Network for Explainable Microvascular Invasion Classification

Microvascular invasion (MVI) is a critical factor for prognosis evaluation and cancer treatment. The current diagnosis of MVI relies on pathologists to manually find out cancerous cells from hundreds of blood vessels, which is timeconsuming, tedious, and subjective. Recently, deep learning has achieved promising results in medical image analysis tasks. However, the unexplainability of black box models and the requirement of massive annotated samples limit the clinical application of deep learning based diagnostic methods. * Corresponding author vessels healthy/MVI vessels 117140×273140 px (a) (b) (c) healthy/cancerous cells Figure 1. Examples of MVI and healthy vessels extracted from a pathological image of liver cancer. (a) The super large sample contains numerous blood vessels of varied sizes. (b) The healthy vessels are composed of a variety of cells with similar appearances. (c) The cancerous cells have varied types and similar appearances to parts of healthy cells.