priyanshu200223/multi-organ-tumor-detection

OncoVision AI Pro

Advanced Multi-Organ Tumor Detection & Clinical Analysis Platform

OncoVision AI Pro is a diagnostic-assistance system built for research and educational use. It performs automated tumor detection on multiple organ systems using YOLO-based deep learning models, evaluates image quality, generates clinical-style reports, saves patient records, and provides analytics through a modern Gradio interface.

This project is designed as an end-to-end prototype for medical imaging AI workflows.

Features

1. Multi-Organ Tumor Detection

Supports:

• —Brain (MRI)
• —Lung (CT)
• —Liver (CT)
• —Kidney (CT)

Each organ uses a dedicated model (custom YOLO weights). If custom weights are missing, the system safely falls back to a general YOLOv8 model.

2. Ensemble Prediction Pipeline

The platform can optionally run predictions at multiple confidence levels and merge them. This improves detection sensitivity and robustness in challenging scans.

3. Clinical-Style Reporting

Every analysis generates:

• —Primary diagnosis and severity stage
• —Tumor size estimation in millimeters
• —Risk stratification (Low / Moderate / High / Critical)
• —Cost estimation (based on typical treatment protocols)
• —Recommended treatment pathway
• —Confidence metrics and thresholds

4. Image Quality Assessment

Automatic detection of:

• —Motion blur
• —Underexposure / overexposure
• —Low contrast

This provides quick feedback on whether a scan is suitable for AI-based analysis.

5. Patient Records Management

The platform maintains:

• —A complete scan history
• —Timestamped doctor notes
• —Filterable record views (by organ, risk level, or date range)

Records are saved to local CSV + JSON for immediate portability.

6. Batch Processing

Upload multiple CT/MRI images and run automated diagnostics in bulk. A summary and complete table of results are generated instantly.

7. Analytics Dashboard

Displays:

• —Organ-wise scan counts
• —Risk distribution
• —Recent activity
• —Dataset summary statistics

Useful for operational tracking and research analysis.

8. Professional UI (No Emojis)

The interface uses:

• —Gradio 4.x
• —Clean typography (Inter)
• —Soft clinical color palette
• —Optional simpleicons SVG icons for branding consistency

Installation

Requirements

Python 3.9+

Install dependencies

pip install -r requirements.txt

Or manually:

pip install gradio ultralytics opencv-python numpy pandas simpleicons

Add your models

Place model files in the project directory:

brain_yolo.pt
lung_yolo.pt
liver_yolo.pt
kidney_yolo.pt

If any of these are missing, the system will automatically fall back to a YOLOv8-Nano model.

Running the Application

python app.py

This launches a Gradio web interface at:

http://127.0.0.1:7860

Project Structure

.
├── app.py                     # Main application
├── brain_yolo.pt              # Custom YOLO model (optional)
├── lung_yolo.pt
├── liver_yolo.pt
├── kidney_yolo.pt
├── patient_records.csv        # Auto-generated
├── doctor_notes.json          # Auto-generated
├── scan_comparisons.json      # Reserved for future features
└── README.md

Disclaimers

This project is strictly intended for:

• —Research
• —Prototyping
• —Educational demonstrations
• —Non-clinical experimentation

It is not a medically approved diagnostic tool. All outputs must be reviewed by certified medical professionals before use in any real-world scenario.

Future Enhancements (Roadmap)

• —DICOM support with metadata extraction
• —3D CT/MRI volumetric processing
• —UNet-based segmentation models
• —PACS integration
• —Role-based authentication for hospital settings
• —Real-time comparative scan analysis