Overview / Usage

The goal of this project is to automate the process of detection of nucleus. This is helpful in the analysis and disease prediction. Deep Learning approach is used for this task.

Methodology / Approach

The Dataset has 3 types of images containing multiple nuclei, this image is taken as input and a grayscale image with the nuclei pixels as white and rest of the pixels black is the output. U-net architecture is used for this task. This project has a lot of applications in biomedical analysis and can be easily extended for more applications. Some of them are -

1. Counting the number of nuclei.
2. Segmentation and detection of different types of nuclei and counting them.
3. Segmentation of nuclei based on their sizes, shapes, length, etc.

Technologies Used

Python, PyTorch, OpenCV

Repository

https://github.com/alishdipani/U-net-Pytorch

Other links

• Dataset

• Project Report (https://github.com/alishdipani/U-net-Pytorch/blob/master/CS_F266.pdf)