HistoAnalyzer

Overview

HistoAnalyzer is a cross-platform desktop application for reproducible H-DAB histology analysis. It combines QuPath-exported pixel classifiers, InstanSeg nuclei segmentation, regional machine learning and compartment-specific DAB quantification in a unified batch-processing interface.

Main workflow

1. Tissue/background classification from a QuPath ANN JSON.
2. Anthracosis/ink detection from a QuPath RTrees JSON.
3. Dilation and subtraction to obtain CleanTissue.
4. InstanSeg brightfield nuclei segmentation.
5. H-channel nuclear, neighborhood and regional feature extraction.
6. Random Forest classification of Tumor, Stroma and Other.
7. DAB quantification independently inside Tumor and Stroma.
8. GeoJSON, CSV, mask TIFF and quality-control PNG export.

Desktop features

• Single-image and batch processing
• File, folder and drag-and-drop input
• Baseline, training, prediction and train-then-predict workflows
• Live logs, progress, cancellation and per-image status
• Stage, nuclei, compartment and DAB preview tabs
• Persistent classifier and output settings
• InstanSeg with watershed fallback
• CPU, CUDA and Apple MPS execution options
• TIFF / OME-TIFF and common OpenSlide formats

Typical use cases

• Reproducible H-DAB tissue quantification
• Tumor/stroma compartment analysis
• Nuclei segmentation-assisted histology analysis
• Batch export for downstream quantitative workflows

Outputs

• Tissue and anthracosis GeoJSON masks
• Positive and negative DAB GeoJSON outputs
• Tumor, Stroma and Other compartment GeoJSONs
• Compartment-specific positive / negative overlays
• CSV measurement tables
• Mask TIFF and QC montage / overlay PNG files

Scientific use note

HistoAnalyzer is research-use software. Tissue masks, artifact detection, nuclei segmentation, compartment predictions, DAB thresholds and final measurements should be validated on representative data before quantitative interpretation or scientific reporting.