Coronary Artery Disease (CAD) is the single leading cause of death globally and early detection methods are urgently needed.To assist the diagnostic process among physicians, it is desired to have an automatic labelling algorithm for coronary arteries in CT images.
This paper proposes a novel Convolutional Neural Network framework due to its ability to account for large variation in its input. Another objective of this research is to achieve an accuracy that is on par or superior than the contemporary literatuee works. The preliminary results of CNN demonstrated an overall accuracy and precision of 80.5% and 81.1% respectively.
Further action will be undertaken to augment the prediction accuracy of all branches in a coronary tree, especially the side branches. This includes
- Using a hierachial Tree-CNN or Tree-LSTM model
- spherical coordinate transform
- tuning the hyperparameters.