Identifying Cancer Types in Microscope Images of Lung and Colon Cells

Lung adenocarcinoma cells make up large proportions of lung cancer cases and colon cancer is one of the most prevalent cancers in the United States. Although many different medicines have been recently developed to attack these cancers, the most effective way to stop it is early detection.

This program seeks to reach a high accuracy of detecting types of cancer in addition to whether or not cells are benign. The model categorizes the cell images into 5 classes: benign colon, benign lung, cancerous lung adenocarcinoma, cancerous colon adenocarcinoma, and cancerous lung squamous cell carcinoma. Two models are then trained with the images: a random forest classifier, and a deep neural network using convolutional layers. Finally a third model predicts which classifier is more likely to make the correct prediction for a given image, and that classifier’s prediction is used. This tactic allows the models to learn different patterns, but still be effective at predicting all classes of images.

Explore More!

Research Paper