PathML

About

PathML provides a comprehensive toolkit for computational pathology workflows, enabling researchers and developers to process, analyze, and model complex histopathology data. It facilitates the loading of over 160 whole-slide image formats, offers modular preprocessing pipelines for stain normalization and tissue detection, and supports advanced spatial analysis through graph construction. With built-in support for deep learning models like HoVer-Net and multiparametric data from platforms like CODEX and Vectra, it streamlines the path from raw imaging to quantifiable insights and machine learning inference.

Key Features

• Spatial graph construction for modeling cellular and tissue-level relationships
• Unified loading for 160+ proprietary whole-slide image formats including SVS, NDPI, and DICOM
• Advanced analysis for multiparametric imaging data including CODEX and spatial proteomics
• Modular preprocessing pipelines for H&E stain normalization and tissue detection
• 16 GitHub stars
• Deep learning integration for nucleus segmentation and classification using models like HoVer-Net

Use Cases

• Training and deploying machine learning models for automated nucleus detection and classification
• Processing and normalizing H&E stained tissue images for large-scale pathology studies
• Performing cell segmentation and marker quantification in multiplex immunofluorescence data