End-to-End PCOS Detection Using GenAI and Ultrasound Imaging

Overview

This repository presents a comprehensive approach to detecting Polycystic Ovary Syndrome (PCOS) by integrating Generative AI techniques with ultrasound imaging analysis. The project encompasses multiple methodologies, each detailed in individual Jupyter notebooks.

Repository Structure

• PCOS Detection Based on Physical and Clinical Parameters.ipynb: This notebook explores the use of physical and clinical data for PCOS detection. It involves data preprocessing, feature selection, and the application of machine learning models to classify PCOS cases based on clinical parameters.

• deepseek-for-pcos-detection.ipynb: This notebook delves into the application of DeepSeek, a deep learning framework, for analyzing ultrasound images to detect PCOS. It covers image preprocessing, model architecture design, training procedures, and evaluation metrics to assess model performance.

• pcos-detection-from-usg-images.ipynb: Focusing on ultrasound (USG) image analysis, this notebook demonstrates the use of convolutional neural networks (CNNs) for feature extraction and classification. It includes steps for image augmentation, model training, validation, and testing to accurately identify PCOS from ultrasound imagery.

Data Availability

• One of the dataset (PCOS_data_without_infertility.xlsx) utilized in these notebooks are included in the repository.
• Another dataset consisting of USG images due to size contrainsts could not be uploaded but is available at: https://figshare.com/articles/dataset/PCOS_Dataset/27682557?file=50407062