Project information
- ๐ฏ Objective: The project aims to utilize Python and deep learning models (UNet, AlexNet, and VGG) to analyze X-ray images of lung cancer. The goal is to detect and compare carcinogenic regions across different models, providing a comprehensive evaluation of their performance in cancer identification.
- date: Apr 2023 - Apr 2023
- Report of the Project: Report
- Github Link: Github Link
Description:
๐ก Features:
- ๐๐จ๐๐๐ฅ ๐๐๐ซ๐ข๐๐ญ๐ฒ: Employs three different deep learning modelsโUNet, AlexNet, and VGGโto identify carcinogenic foci in lung X-ray images, allowing for a broad comparison of their effectiveness.
- ๐๐๐ซ๐๐จ๐ซ๐ฆ๐๐ง๐๐ ๐๐ฏ๐๐ฅ๐ฎ๐๐ญ๐ข๐จ๐ง: Compares the models based on accuracy, errors, and training-testing times, with results visualized through matplotlib to highlight differences and insights.
- ๐๐๐ฏ๐๐ง๐๐๐ ๐๐ฆ๐๐ ๐ข๐ง๐ ๐๐๐๐ก๐ง๐ข๐ช๐ฎ๐๐ฌ: Focuses on advanced image processing tasks such as lung segmentation and normalization, crucial for improving model accuracy and reliability in medical imaging.
- ๐๐จ๐ฆ๐ฉ๐ซ๐๐ก๐๐ง๐ฌ๐ข๐ฏ๐ ๐๐๐ญ๐๐ฌ๐๐ญ ๐๐ฌ๐: Utilizes a dataset containing over 1000 CT scans from the LUNA16 challenge, which includes detailed annotations and metadata, enhancing the training and testing phases of model development.
- ๐๐ข๐ ๐ง๐ข๐๐ข๐๐๐ง๐ญ ๐
๐ข๐ง๐๐ข๐ง๐ ๐ฌ ๐๐ง๐ ๐๐๐ญ๐ซ๐ข๐๐ฌ: Reports metrics such as accuracy, precision, recall, and F1-score to quantify each model's performance, with UNet emerging as the most effective based on the obtained results.
- ๐๐ฐ๐๐ซ๐-๐๐ข๐ง๐ง๐ข๐ง๐ ๐๐๐๐จ๐ ๐ง๐ข๐ญ๐ข๐จ๐ง: This project won first place at the AI Night competition, highlighting its innovative approach and significant impact in the field of AI-driven medical diagnostics.
These features outline the projectโs capability to leverage complex machine learning algorithms to enhance the diagnostic processes of lung cancer through imaging technology.