Oldřich Kodym

Enabling deep learning methods on small-scale computed tomography image datasets

Školitel: Adam Herout

Instituce: Brno University of Technology, Faculty of Information Technology

Obor: Computer Science and Engineering


O mém projektu

Applications of convolutional neural networks and other deep learning methods in medical data could have enormous impact in various areas of clinical diagnostics and treatment planning. However, their employment in this area is inhibited by scarcity of data available for training of neural models. This project proposes use of novel experimental methods to tackle this issue in various image analysis tasks in medical computed tomography data such as tissue segmentation. This could lead to faster and more reliable implementations of new deep learning methods into clinical practice.