https://github.com/charlesyuan02/medical_image_classifier

A CNN classifier trained on medical images from the Kvasir Dataset, along with a ResNet50 model trained using transfer learning for comparison.
https://github.com/charlesyuan02/medical_image_classifier

Last synced: 9 days ago
JSON representation

A CNN classifier trained on medical images from the Kvasir Dataset, along with a ResNet50 model trained using transfer learning for comparison.

• Host: GitHub
• URL: https://github.com/charlesyuan02/medical_image_classifier
• Owner: CharlesYuan02
• Created: 2020-11-21T05:31:38.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2020-11-21T06:16:24.000Z (almost 4 years ago)
• Last Synced: 2024-10-10T19:10:07.898Z (26 days ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 790 KB
• Stars: 5
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Medical_Image_Classifier

A CNN classifier trained on medical images from the Kvasir Dataset, along with a ResNet50 model trained using transfer learning for comparison. 

Automatic detection is an important application of deep learning, and has the potential to improve medical practices and refine health care systems. This project was made as a demonstration of this application. Utilizing the Kvasir dataset, which contains images from the gastrointestinal tract, the images are classified into three anatomical landmarks and three clinically significant findings, as well as two categories of images related to endoscopic polyp removal, for a total of eight classes comprised of 8000 images. The sorting and annotation of the dataset was performed by trained endoscopists.

In the end, the CNN classifier was shown to perform better than the ResNet50 model, with an accuracy of 65% compared to the ResNet's 51%.

The dataset can be found here (Kvasir Dataset v2):

https://datasets.simula.no//kvasir/#data-collection