Microvasculature Segmentation in Human BioMolecular Atlas Program (HuBMAP)

TL;DR

This paper explores advanced segmentation techniques for microvascular structures in human kidney histology images, benchmarking various deep learning architectures to improve accuracy in biomedical image analysis.

Contribution

It introduces a comprehensive evaluation of different backbone architectures and models for microvasculature segmentation in histology images, expanding beyond the baseline U-Net approach.

Findings

Deeper models and Feature Pyramid Networks improve segmentation accuracy.

Benchmarking reveals the most effective architectures for microvasculature segmentation.

The study provides insights into the performance trade-offs of various deep learning models.

Abstract

Image segmentation serves as a critical tool across a range of applications, encompassing autonomous driving's pedestrian detection and pre-operative tumor delineation in the medical sector. Among these applications, we focus on the National Institutes of Health's (NIH) Human BioMolecular Atlas Program (HuBMAP), a significant initiative aimed at creating detailed cellular maps of the human body. In this study, we concentrate on segmenting various microvascular structures in human kidneys, utilizing 2D Periodic Acid-Schiff (PAS)-stained histology images. Our methodology begins with a foundational FastAI U-Net model, upon which we investigate alternative backbone architectures, delve into deeper models, and experiment with Feature Pyramid Networks. We rigorously evaluate these varied approaches by benchmarking their performance against our baseline U-Net model. This study thus offers a comprehensive exploration of cutting-edge segmentation techniques, providing valuable insights for future research in the field.