CNN-Driven Classification of Alzheimer’s Severity from MRI

This project was inspired by the fact that Alzheimer’s disease is often diagnosed quite late, which can limit treatment options and negatively affect patients’ quality of life. We were interested in exploring whether machine learning techniques could be used to analyse MRI brain scans in a way that helps assess the severity of the disease more consistently.

Throughout the project, we learned how Convolutional Neural Networks work and how they can be applied to medical imaging problems. We also gained practical experience with data preprocessing, training neural networks, and evaluating model performance. Along the way, we became more aware of the challenges involved in healthcare-related machine learning, such as working with small datasets, dealing with class imbalance, and avoiding overfitting.

To build the project, we trained a CNN to classify Alzheimer’s disease severity using MRI scans. This involved preparing the dataset, selecting an appropriate model architecture, and experimenting with different settings to improve performance.

One of the main challenges we faced was the limited amount of available data, which made it difficult to achieve stable results. We also found it challenging to interpret the model’s predictions, which highlighted the importance of transparency and responsible use of AI in medical applications.

Built With

• colab
• matplotlib
• numpy
• pandas
• parquet
• pillow
• pyarrow
• python
• scikit-learn
• tensorflow