Unify computational pathology workflows for whole-slide images and related data.
npx skills add https://github.com/Hung-3008/agusta --skill pathml-hung-3008
Please help me install this Agent Skill. Skill: pathml Source: https://github.com/Hung-3008/agusta/tree/main/.agents/skills/pathml Command: npx skills add https://github.com/Hung-3008/agusta --skill pathml-hung-3008
PathML addresses the complexity of computational pathology workflows by providing an end-to-end toolkit that unifies loading, preprocessing, graph construction, ML model training, and multiparametric imaging analysis for whole-slide images and related data.
Install PathML, load a slide with SlideData, generate tiles, apply a preprocessing pipeline, and save the results to an HDF5 file.
High-intent search queries and answers about installing and using this skill.
Preprocess whole-slide images by loading slides into a SlideData class, generating tiles, and applying modular transforms for stain normalization, tissue detection, and QC before saving to HDF5.
Build spatial graphs from whole-slide images by detecting nuclei and constructing cellular relationship models, enabling graph neural network training with built-in architectures like HACTNet.
Computational pathology workflows support multiplex imaging formats including CODEX, Vectra, and MERFISH, enabling per-cell marker quantification alongside H&E whole-slide image analysis.
Manage large pathology datasets using HDF5 storage for tile management and distributed processing, ensuring scalable data handling across high-resolution whole-slide images and multiplex data.
Train deep learning models for nucleus detection using built-in architectures like HoVer-Net, supported by comprehensive reference documentation and scalable HDF5 dataset management for evaluation and ONNX export.
Load over 160 proprietary whole-slide image formats using unified slide classes and interfaces, standardizing diverse inputs into a consistent computational pathology workflow.
LeonChaoXLoad and preprocess whole-slide pathology images into ML-ready datasets.
jackspaceAnalyze whole-slide pathology images and multiparametric imaging data for computational pathology workflows.
LynxLabVNUnify distributed training for PyTorch models across DDP, DeepSpeed, FSDP, and Megatron.
jimmc414Extract tissue tiles from gigapixel whole slide images for ML datasets.
