The main aim of this project is to develop a machine learning-based system that can automate the process of identifying diseases from medical images. Using Convolutional Neural Networks (CNNs) and transfer learning methods (e.g., ResNet, EfficientNet), the system is intended to support early diagnosis in medical settings especially where radiologists are scarce. This automation can expedite diagnosis, improve patient outcomes, and help medical professionals by providing a reliable second opinion. While the system will not replace doctors, it aims to serve as a diagnostic aid, significantly enhancing efficiency and accessibility in healthcare services.
The project follows a structured twelve-week plan. In the first week, students are introduced to the fundamentals of CNNs and the project’s architecture. In the following weeks, they create a basic framework, explore key machine learning libraries (like TensorFlow, Keras, or PyTorch), and gather datasets for various diseases. Once the dataset is prepared and cleaned, the CNN model will be trained and tested for accuracy. Students will experiment with model improvements, apply transfer learning if needed, and extract final results. The last phases focus on full project development, testing, documentation, and a team presentation. Tools like Anaconda Navigator or Google Colab are recommended for implementation, and students are expected to adhere to ethical coding and documentation practices throughout.
