Unify image loading, preprocessing, graph construction, and ML analysis for computational pathology.
npx skills add https://github.com/OwnLabAI/ownlab --skill pathml-ownlabai
Please help me install this Agent Skill. Skill: pathml Source: https://github.com/OwnLabAI/ownlab/tree/main/mart/skills/scientific-skills/pathml Command: npx skills add https://github.com/OwnLabAI/ownlab --skill pathml-ownlabai
PathML unifies loading, preprocessing, graph construction, and ML-driven analysis to automate complex computational pathology workflows, reducing manual steps and error-prone processes.
Install PathML, load a sample slide, and run a minimal preprocessing plus nucleus analysis pipeline.
High-intent search queries and answers about installing and using this skill.
Computational pathology workflows can be automated by unifying image loading, preprocessing, graph construction, and ML-based analysis into a single pipeline. This supports whole-slide images across H&E, CODEX, Vectra, and MERFISH datasets from raw image to spatial analytics.
Preprocessing multiparametric imaging data for machine learning is best handled through modular pipelines offering stain normalization, tissue and nucleus detection, and QC transforms. This ensures reproducible data preparation before model training.
Yes, you can use PyTorch models for nucleus detection in whole-slide images. Pre-built nuclei models like HoVer-Net and HACTNet are supported with end-to-end training workflows for analyzing tissue and cellular structures.
Spatial graph construction for cell interaction analysis is fully supported. The pipeline builds spatial graphs from preprocessed images to analyze cell and tissue interactions, enabling advanced spatial analytics across multiple imaging modalities.
Managing large pathology datasets for distributed batch processing is achieved using HDF5 storage, dataset versioning, and Dask distributed processing. This combination delivers reproducible and scalable pipelines for high-volume image analysis.
Computational pathology image loading supports over 160 slide formats and modalities. This includes standard H&E datasets as well as multiparametric imaging formats like CODEX, Vectra, and MERFISH for comprehensive workflow integration.
dralkhConvert whole-slide and multiparametric pathology images into normalized tiles, masks, and model-ready datasets.
omicverseAlign single-cell references with spatial transcriptomics to reconstruct tissue architecture.
LeonChaoXLoad and preprocess whole-slide pathology images into ML-ready datasets.
sanand0Processes whole-slide pathology images with machine learning and spatial analysis.
