The question behind the work
Image classification is about more than a final label. This experiment explores how preprocessing, class balance and architecture choices affect a model’s behavior.
COMPUTER VISION
Look beyond the prediction.
An experimental MRI classification study using transfer learning, augmentation and model interpretation.
THE PROCESSINPUT → UNDERSTANDING
01Prepare
Explore and preprocess MRI image data.
02Balance
Apply augmentation and class balancing.
03Train
Experiment with transfer learning architectures.
04Inspect
Review classification reports, confusion matrices and attention maps.
THE PROJECT
Alzheimer’s Image Classification
Source and context from my portfolio
How it comes together
The study includes ResNet50 and InceptionV3, with classification reports and confusion matrices to inspect performance. TorchCam visualizations support a closer look at model attention.
What to keep in mind
This is an experimental machine learning project, not a clinically validated diagnostic tool. Model outputs should be interpreted within that research context.