Colorectal Histology Classification (Solo Deep Learning Project)
This project was part of my Master’s in Statistics and builds on a more rudimentary version I originally attempted during undergrad. It applies deep convolutional neural networks to classify colorectal tissue images into eight categories, including tumor, stroma, and lymphocytes. I experimented with 14 different model architectures using various combinations of layer depth, dropout, activation functions, and image augmentation. The final model achieved 91.5% overall accuracy, with 96% accuracy in tumor classification.
While this version is rough and still lacks some narrative polish, I’ve posted it here as a snapshot of my solo modeling work. A more detailed write-up and perhaps cleaned-up code will be added soon.
