Breast Cancer Detection on Histopathological Data

2022 Fourth International Conference on Emerging Research in Electronics, Computer Science and Technology (ICERECT)

Trained prominent CNN architectures ResNet50, AlexNet, and GoogleNet on BreakHis dataset from scratch
Training was done by utlizing various image processing techniques such as Contrast Limited Adaptive Histogram Equalization (CLAHE), Image sharpening and Edge enhancement
The optimum model amongst the proposed i.e. Resnet50 + Image Sharpening achieved an accuracy of 95%

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
README.md		README.md
after_transform.png		after_transform.png
before_transform.png		before_transform.png
breakhis-2.ipynb		breakhis-2.ipynb
dataset.py		dataset.py
image_preprocess.ipynb		image_preprocess.ipynb
preprocess.py		preprocess.py
run.py		run.py
utils.py		utils.py

Provide feedback