Medical AI Research
Explore Assem Sabry's research and projects in the medical field, focusing on AI-powered diagnostic tools and healthcare solutions that bridge the gap between artificial intelligence and medical practice.
Intelligent Brain Tumor Detector (IBTD)
Overview
The Intelligent Brain Tumor Detector (IBTD) is a deep learning-based classifier built to accurately detect and classify brain tumor types from MRI images. The model is trained using a fine-tuned ResNet50 architecture and leverages modern training techniques like data augmentation and MixUp regularization.
Model Evaluation Report
The model was evaluated on a balanced test set containing four brain tumor categories: GLIOMA, MENINGIOMA, NOTUMOR, and PITUITARY.
Classification Performance:
| Class | Precision | Recall | F1-score | Support |
|---|---|---|---|---|
| GLIOMA | 99.65% | 95.00% | 97.27% | 300 |
| MENINGIOMA | 93.87% | 95.10% | 94.48% | 306 |
| NOTUMOR | 99.26% | 99.51% | 99.38% | 405 |
| PITUITARY | 96.44% | 99.33% | 97.87% | 300 |
Overall Accuracy: 98.41%
Technical Implementation
Model Architecture
- • Fine-tuned ResNet50 architecture
- • Data augmentation techniques
- • MixUp regularization
- • Multi-class classification (4 classes)
Performance Metrics
- • Overall accuracy: 98.41%
- • Balanced test set evaluation
- • High precision across all classes
- • Robust recall performance
Histopathological AI Detection (HAID)
Overview
HAID (Histopathological AI Detection) is an AI-powered deep learning model developed to detect breast cancer in histopathological images. It's designed to assist medical professionals with faster, more consistent diagnoses, empowering hospitals and pathology labs with intelligent diagnostic tools.
Dataset
Dataset Specifications
- Images: Over 250,000 high-resolution histopathological samples
- Classes: Binary classification: Normal vs. Cancer
- Image Size: All images resized to 150x150 pixels
Model Performance
Validation Accuracy: ~80% (to be updated after final evaluation)
Classification Report:
| Class | Precision | Recall | F1-Score |
|---|---|---|---|
| Normal | 78.3% | 80.0% | 79.1% |
| Cancer | 81.2% | 79.5% | 80.3% |