ALL Detection System

Automated Acute Lymphoblastic Leukemia (ALL) detection from peripheral blood smear images, powered by deep learning and on-device AI.

How It Works

1. Segmentation — White blood cells are extracted from full blood smear images using K-Means clustering on the LAB color space
2. Classification — Each cell is analyzed by a MobileNetV3 model (TFLite) to determine if it is a healthy WBC or a blast cell (suspected ALL)
3. AI Summary — An on-device LLM (Phi-3 via Ollama) generates a clinical-style interpretation of the results
4. Visualization — Results are displayed in a real-time dashboard with cell cards, annotations, and confidence scores

Quick Start

Desktop (Windows/Linux/macOS)

pip install -r requirements.txt
cd src
python ui/app.py

Raspberry Pi 5

chmod +x setup_pi.sh
./setup_pi.sh    # Installs everything automatically
./run.sh         # Launch the app

The setup script handles: Python dependencies, Ollama installation, model download, desktop shortcut.

Project Structure

src/
├── ui/                  # Desktop GUI (CustomTkinter)
│   ├── app.py           # Main application
│   └── theme.py         # Design system
├── detection/           # Core detection pipeline
│   ├── blast_detector_v5.py  # Cell detection + classification
│   ├── llm_utils.py     # AI summary generation (Ollama)
│   └── stage1_screening.py   # TFLite screening model
├── segmentation/        # WBC segmentation (K-Means)
├── classifier/          # Dataset loader + model definitions
├── training/            # Model training scripts (reference)
├── api/                 # REST API for Android app (planned)
└── utils/               # Visualization, preprocessing utilities

android_app/             # Android companion app (Jetpack Compose)
models/                  # Trained model files (.tflite)
data/                    # Reference images, processed data
docs/                    # Research papers, technical docs

Datasets

Dataset	Purpose	Details
C-NMC 2019	Training	~10,600 pre-cropped cell images (ALL vs Healthy)
ALL-IDB1	Testing	Full blood smear images for end-to-end pipeline testing

See DATASETS.md for full details.

Hardware Targets

• Primary: Raspberry Pi 5 (8GB) + 1080p display
• Secondary: Android devices (via companion app)

Tech Stack

Component	Technology
Detection Model	MobileNetV3 (TFLite)
AI Summary	Phi-3 via Ollama
Desktop GUI	CustomTkinter (Python)
Android App	Jetpack Compose + Material 3
Segmentation	OpenCV, scikit-image

Requirements

See requirements.txt for Python dependencies.

System Requirements:

• Python 3.10+
• Ollama (for AI summaries)
• 4GB+ RAM (8GB recommended for LLM)