Instance-Based Vision Transformer for Subtyping of Papillary Renal Cell Carcinoma in Histopathological Image

Histological subtype of papillary (p) renal cellular and cell-layer level patterns almost cannot be captured by existing CNN-based models in large-size histopathological images, which brings obstacles to directly applying these such a fine-grained classification task. This paper proposes novel instance-based Vision Transformer (i-ViT) learn robust representations images for the pRCC subtyping task extracting finer features from instance patches (by cropping around segmented nuclei assigning predicted grades). The proposed i-ViT takes top-K instances as inputs aggregates them capturing both position-embedding layer, grade-embedding multi-head multi-layer self-attention module. To evaluate performance framework, experienced pathologists select 1162 regions interest 171 whole slide type 1 2 pRCC. Experimental results show that method achieves better than with significant margin.