Segmentation by Factorization: Unsupervised Semantic Segmentation for Pathology by Factorizing Foundation Model Features

TL;DR

The results show that F-SEG provides robust unsupervised segmentation capabilities for H&E pathology images, and that the segmentation quality is greatly improved by utilizing pathology foundation models.

Abstract

We introduce Segmentation by Factorization (F-SEG), an unsupervised segmentation method for pathology that generates segmentation masks from pre-trained deep learning models. F-SEG allows the use of pre-trained deep neural networks, including recently developed pathology foundation models, for semantic segmentation. It achieves this without requiring additional training or finetuning, by factorizing the spatial features extracted by the models into segmentation masks and their associated concept features. We create generic tissue phenotypes for H&E images by training clustering models for multiple numbers of clusters on features extracted from several deep learning models on The Cancer Genome Atlas Program (TCGA), and then show how the clusters can be used for factorizing corresponding segmentation masks using off-the-shelf deep learning models. Our results show that F-SEG provides robust unsupervised segmentation capabilities for H&E pathology images, and that the segmentation quality is greatly improved by utilizing pathology foundation models. We discuss and propose methods for evaluating the performance of unsupervised segmentation in pathology.

Figures (8)

• Figure 1: An example of F-SEG semantic segmentation with the UNI foundation model and k=64 TCGA clusters
• Figure 2: Overview of F-SEG when classifying factorized concepts into the most similar cluster
• Figure 3: Overview of F-SEG when fixing the concept matrix to be cluster centers
• Figure 4: Evaluating Segmentation by Factorization on the Breast Cancer Semantic Segmentation Dataset
• Figure 5: Evaluating Segmentation by Factorization on the WSSS4LUAD Dataset