ThoraxScanAI

ThoraxScanAI is an advanced deep learning project aimed at classifying lung diseases from X-ray images using TensorFlow and Keras. The project utilizes pre-trained models for transfer learning, augmented with custom layers to tailor the network for lung disease detection.

Project Structure

The project is structured into several key scripts, each fulfilling a specific role in the development and deployment of the machine learning model:

• split_dataset.py: Organizes the original dataset into training, validation, and test sets to ensure a proper distribution of data across different phases.
• custom_data_loader.py: A comprehensive script for loading and preprocessing data. It supports data augmentation for the training set to enhance model generalization.
• model.py: Constructs the model architecture, leveraging VGG16 for transfer learning and adding custom layers for the specific task of lung disease classification.
• train.py: Manages the model training process, incorporating callbacks for optimization and early stopping to improve learning efficiency.
• evaluate.py: Assesses the trained model on the test dataset, providing key performance metrics to gauge the model's effectiveness.
• utils.py: Contains utility functions for visualizing training progress and preprocessing single images for model predictions.
• predict_single_image.py: Demonstrates the application of the trained model to classify individual X-ray images, showcasing the model's practical utility.

Getting Started

Prerequisites

• Python 3.8 or later
• TensorFlow 2.x
• NumPy
• Matplotlib
• scikit-learn

Installation

Clone the repository to your local machine:

git clone https://github.com/darwin757/ThoraxScanAI.git
cd ThoraxScanAI

Install the required dependencies:

pip install tensorflow numpy matplotlib scikit-learn

Dataset

The original dataset, "Lung X-Ray Images", is processed and split into training, validation, and test sets using split_dataset.py. The script organizes the data into a structure suitable for training and evaluation.

Training the Model

Run train.py to start the training process. The script will use the processed data and apply transfer learning with a VGG16 base model:

python train.py

Evaluating the Model

After training, evaluate the model's performance on the test set with evaluate.py:

python evaluate.py

Single Image Prediction

The project now includes a script for predicting the class of a single X-ray image using the trained model. After training, you can predict new images with:

python predict_single_image.py path/to/image.jpg

License

This project is open source and available under the MIT License.

Acknowledgments

• Dataset sourced from Kaggle - Lung X-Ray Images by Md Alamin Talukder and Fatemeh Mehrparvar.
• ThoraxScanAI utilizes TensorFlow and the VGG16 model for its deep learning pipeline.

Contact

For any queries or further discussion, please contact project maintainers at [ala.korabi@gmail.com].