To get proficient in diagnostic imaging, radiologists need to practice. However, practicing with a sufficient variety of cases can be challenging due to patient privacy, disease rarity and reduced guidance from medical staff due to COVID-19. [1 – 2] Brain cancer diagnosis, especially tumour grading, is of critical importance since it guides the choice of treatment. [3] Therefore, it is necessary to gain experience with a wide variety of cases representing the true difference in disease presentations as they appear radiologically.
We recently experimented with an approach to generate realistic but artificial radiographic images using Semantic Image Synthesis (SIS) networks. We have shown promising results in 2D lung cancer imaging based on computed tomography, and we are seeking to validate this approach on brain cancer imaging based on magnetic resonance imaging.
The goal of the project is to extend our prior approach to generate brain MRI images with different types of brain cancer using a Semantic Image Synthesis network.
Specifically, this project involves ...
- creating a pre-processing pipeline to get the necessary semantic label information and to pre-process the MRI scans.
- testing multiple, popular semantic image synthesis models and evaluating their performance, i.e. benchmarking.
- improving these models to integrate tumor grading.
The project is suitable for master students from Biomedical Engineering and Electrical Engineering. Contact Elfi Hofmeijer if you are interested in this assignment or would like to have more information (e.i.s.hofmeijer@utwente.nl).
__________________________________________________________________________________
[1] Lotan E, et al. Medical Imaging and Privacy in the Era of Artificial Intelligence: Myth, Fallacy, and the Future. J Am Coll Radiol 2020 17:1159–1162. https://doi.org/10.1016/j.jacr.2020.04.007
[2] Sugi MD, et al. Bridging the gap: interactive, case-based learning in radiology education. Abdom Radiol 2021 46:5503–5508. https://doi.org/10.1007/s00261-021-03147-z
[3] Pereira S, et al. Automatic Brain Tumor Grading from MRI Data Using Convolutional Neural Networks and Quality Assessment. iMIMIC – Workshop on Interpretability of Machine Intelligence in Medical Image Computing 2018.