AI system for prostate cancer MRI analysis combining computer vision and LLMs.
- DICOM MRI preprocessing (T2W + ADC multiparametric sequences)
- PI-RADS scoring (1–5 scale) with lesion mask overlay
- LLM-generated clinical reports via Groq LLaMA 3.1
- Automated PDF export with ReportLab
- Gradio web interface
- PROSTATEx (TCIA) — 346 patients, 5.93GB
- 70 cancer positive, 130 benign cases
- T2W + ADC MRI sequences with PI-RADS annotations
| Component | Technology |
|---|---|
| MRI Loading | PyDICOM, SimpleITK |
| Visualization | Matplotlib |
| LLM Reports | Groq LLaMA 3.1 |
| UI | Gradio |
| PDF Export | ReportLab |
pip install pydicom simpleitk gradio groq reportlab matplotlib
# Add your GROQ_API_KEY
# Run in Google Colab with PROSTATEx dataset- Patient: ProstateX-0000 | PI-RADS: 4 | CANCER DETECTED
- Patient: ProstateX-0009 | PI-RADS: 2 | BENIGN
Built by Vibhavari Tummewar | MTech Advanced Computing, MANIT Bhopal