Benchmarking Hierarchical Image Pyramid Transformer for the classification of colon biopsies and polyps in histopathology images

TL;DR

This paper evaluates the Hierarchical Image Pyramid Transformer (HIPT) model for classifying colorectal biopsies and polyps in histopathology images, exploring pretraining strategies to improve performance without extensive annotations.

Contribution

It introduces and compares two pretraining strategies for HIPT on colorectal biopsy classification, demonstrating their effectiveness in histopathology image analysis.

Findings

Pretraining HIPT with colon biopsy data improves classification accuracy.

Fine-tuning HIPT from TCGA weights outperforms random initialization.

HIPT achieves competitive results on binary and multiclass tasks.

Abstract

Training neural networks with high-quality pixel-level annotation in histopathology whole-slide images (WSI) is an expensive process due to gigapixel resolution of WSIs. However, recent advances in self-supervised learning have shown that highly descriptive image representations can be learned without the need for annotations. We investigate the application of the recent Hierarchical Image Pyramid Transformer (HIPT) model for the specific task of classification of colorectal biopsies and polyps. After evaluating the effectiveness of TCGA-learned features in the original HIPT model, we incorporate colon biopsy image information into HIPT's pretraining using two distinct strategies: (1) fine-tuning HIPT from the existing TCGA weights and (2) pretraining HIPT from random weight initialization. We compare the performance of these pretraining regimes on two colorectal biopsy classification tasks: binary and multiclass classification.