Medical imaging data analysis using 3D deep learning models towards improving the individual treatment plans
K. Kalecińska, T. Fiutowski, P. Jurgielewicz, D. Kabat, B. Rachwał, Ł. Kapłon, M. Kopeć, S. Koperny, D. Kulig, J. Moroń, G. Moskal, A. Ruciński, P. Wiącek, B. Mindur, T. Szumlak
abstract
This work is a part of a research project aiming at delivering the next generation active medical phantom, Dose-3D, with high spatial granulation for quasi-real time measurement of the volumetric radiotherapeutic dose deposited during photon therapy. The preliminary results, discussed here, pertain to the intelligent medical data augmentation using Generative Adversarial Networks (GANs) technique implemented inside MONAI framework. However, in the scope of the project, we perform a broad search for the most efficient and advanced Deep Learning (DL) models to create tools for 3D Computed Tomography (CT) images segmentation and cancer diagnosis improvement that will be an integral part of the custom designed software platform for processing data collected with Dose-3D phantom. Apart from the innovative detection system the software itself may prove to be disruptive in the context of the currently available tools by offering open-source high quality toolkit for wide use in everyday clinical applications.
