TUM.ai + Helmholtz

This plugin provides one-click color normalization, denoising, and Cellpose-based nuclear segmentation.

Key Features

Quick Start

Installation

pip install neurogenesis-napari

Or install through napari:

1. Open napari
2. Go to Plugins → Install/Uninstall Plugins
3. Search for "TumAI Histology Toolkit"
4. Click Install

Basic Usage

1. Load your images into napari
2. Select the appropriate widget from the Plugins menu
3. Choose your image layers from the dropdown menus
4. Click the action button to process

The plugin will automatically download required AI models on first use.

Widget Documentation

Normalize + Denoise

Purpose: Standardizes color variations and reduces noise in bright-field images.

Usage

1. Load a bright-field image into napari
2. Open Plugins → Normalize and Denoise
3. Select your bright-field image from the BF dropdown
4. Click "Normalize + Denoise"

What it does

• Color Normalization: Adjusts colors against an internal reference to standardize appearance across different images/scanners
• Denoising: Removes noise while preserving important cellular structures
• Output: Creates a new layer named {original_name}_denoised

Segment

Purpose: Detects and segments individual cell nuclei using Cellpose.

Usage

1. Load a nuclear staining image (DAPI) into napari
2. Open Plugins → Segment
3. Select your nuclear image from the DAPI dropdown
4. Optionally adjust:
   • GPU: Enable for faster processing
   • Model: Choose Cellpose model (cyto3 default)
5. Click "Segment Nuclei"

What it does

• Segmentation: Uses Cellpose to identify individual nuclei
• Creates 3 new layers:
  • {name}_masks: Segmentation masks
  • {name}_centroids: Center points of each detected cell
  • {name}_bboxes: Bounding boxes around each cell

Segment + Classify

Purpose: Complete pipeline that segments nuclei AND classifies cell types in multi-channel images.

Usage

1. Load a 4-channel image into napari as separate layers:
   • DAPI: Nuclear staining
   • Tuj1: β-III-tubulin
   • RFP: Red fluorescent protein marker
   • BF: Bright-field
2. Open Plugins → Segment and Classify
3. Select each channel from the respective dropdowns
4. Choose Reuse cached:
   • True: Reuse previous segmentation (faster) from the segment widget
   • False: Perform fresh segmentation
5. Click "Segment + Classify"

What it does

1. Segmentation: Does segmentation same as the segment widget above
2. Feature extraction: Uses a Variational Autoencoder (VAE) to extract features
3. Classification: Nearest-centroid classifier assigns cell types

Output

Creates colored bounding box layers for each detected cell type:

• 🟣 Astrocytes (magenta boxes)
• ⚫ Dead Cells (gray boxes)
• 🔵 Neurons (cyan boxes)
• 🟢 OPCs (lime boxes)

Layer names show counts: {count}_{cell_type}s (e.g., 23_Neurons)

Supported Image Formats

• .czi (via napari-czifile2)
• .png, .jpg