Clustering for COVID-19 CT Scan Segmentation and Analysis

This repository contains a Python implementation of a method to identify ground-glass opacities in lung CT scans, which may indicate the presence of COVID-19. The method utilizes two clustering algorithms, K-means and DBSCAN, for image segmentation and identification of opacities. You can find the detailed information in Clustering for COVID-19 CT Scan Analysis.pdf.

Dataset

The dataset used for this project is available at Kaggle. It consists of CT scans from COVID-19 patients, with each patient having around 300 slices in the axial plane. Each slice is a grayscale image with a resolution of 512x512 pixels.

Package Installation

To install the required packages for this project, run the following command:

pip install pandas, numpy, matplotlib, scikit-learn, nibabel

Usage

1. Clone this repository.
2. Download the dataset from Kaggle mentioned in the references.
3. Run the code using the downloaded dataset in load folder and run python main.py in the directory of the repository.

References

1. COVID-19 CT Scans dataset: Kaggle
2. Kevin P. Murphy, Machine Learning: A Probabilistic Perspective, MIT Press, 2012
3. Martin Ester, Hans-Peter Kriegel, Jörg Sander, Xiaowei Xu, A density-based algorithm for discovering clusters in large spatial databases with noise, Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96), 1996
4. Zou, Kelly H., et al. Statistical Validation of Image Segmentation Quality Based on a Spatial Overlap Index1. Academic Radiology, no. 2, Elsevier BV, Feb. 2004, pp. 178–89. Crossref, doi:10.1016/s1076-6332(03)00671-8
5. Kwee, Thomas C., and Robert M. Kwee. Chest CT in COVID-19: What the Radiologist Needs to Know. RadioGraphics, no. 7, Radiological Society of North America (RSNA), Nov. 2020, pp. 1848–65. Crossref, doi:10.1148/rg.2020200159.