Working with New Data
This week I received data sets from Dr. Zarella. The goal is to process a full-slide histological image into small tiles and then use the program to extract convolution features from these images. The extracted feature data will then be passed to a trained model that can predict if a tumor is benign or malignant. The result of this work should be numerous predictions on a large set of image data. After several marathon programming sessions, I was able to achieve these results and then created a separate program that would make a color coded grid aligning to areas predicted by the program that were either benign or malignant. This image when overlaid on the original slide, is helpful for visually assessing the program's accuracy.