A deep learning and machine learning project for predicting the age of Retinal Pigment Epithelium (RPE) cells using Convolutional Neural Networks (CNNs), Support Vector Machines (SVMs), and K-Nearest Neighbors (KNN).
This project leverages AI to aid medical research and improve understanding of retinal aging patterns.
- 🧬 AI-Powered RPE Classification – Predicts the age group of RPE cells from retinal images.
- 🧠 Multiple ML Models – CNN, SVM, and KNN implemented for comparative performance.
- 📈 High Accuracy – Achieved 90% accuracy with CNN.
- 🔍 Confusion Matrix for detailed performance analysis.
- 🐍 Python-Based and fully reproducible with minimal setup.
| Category | Technology |
|---|---|
| Language | Python |
| Deep Learning | PyTorch |
| Machine Learning | scikit-learn |
| Visualization | Matplotlib |
| Environment | Jupyter Notebook / Python Scripts |
- The dataset contains RPE cell images collected from Emory Eye Center provided by Dr.Yi Jiang
- The dataset is structured into age groups for classification.
git clone https://github.com/H1BOI/RPE-classification.git
cd RPE-classificationpython -m venv venv
# On Mac/Linux
source venv/bin/activate
# On Windows
venv\Scripts\activatepip install -r requirements.txtRun the models individually:
# Run CNN model
python cnn.py
# Run SVM model
python SVM.py
# Run KNN model
python KNN.pyMake sure your dataset paths (if required by the scripts) are correctly set inside each file.
| Model | Accuracy |
|---|---|
| CNN | 90% ✅ |
| SVM | ~78% |
| KNN | ~72% |
Special thanks to:
- Dr. Yi Jiang for guidance and expertise.
- Emory Eye Center for providing RPE cell images for research.
This project is licensed under the MIT License.
See the LICENSE file for more information.