MSc Thesis: Attention mechanisms for end-to-end therapy response prediction on PDAC CTs
Pancreatic ductal adenocarcinoma (PDAC) is considered the fourteenth most common cancer globally, the seventh on the list of causes of cancer death in 2020, and the fourth in 2021.
To further investigate the causes and therapy for affected patients we aim to extend our previous work. This project is ideal for students who like to work on applied deep learning in the medical field.
In particular in this thesis we aim to evaluate the effect of attention mechanisms to the prediction of therapy response and/or other labels for PDAC patients. Hence, after an in-depth literature research you have the opportunity to implement and evaluate the most promising approaches and compare to previous work. Experience in deep learning and good practices in software engineering are thus a requirement. Note that we do explicitly not require any previous medical knowledge.
If you are interested kindly send us a message including your CV and transcript of records. Please also describe briefly what you are looking for and why you are suited to this thesis. Optionally describe a previous deep learning project you have worked on. We are looking forward to your message.
- Advanced knowledge of machine learning and computer vision. Multiple previous experience on deep learning projects.
- Excellent programming skills in Python and a deep learning framework. Previous work is based on PyTorch.
- Clean coding style and organisational skills.
- Willingness to dive into a new field and ability to research related works.
What we offer:
- Opportunity to introduce your deep learning skills to the benefit of cancer patients.
- Working in a team of experts in image processing, deep learning, biomedical engineering and medicine.
- Freedom to pursue your own project while providing the guidance you need
- GPU Hardware
- Opportunity to participate in lab meetings as well as workspace in TranslaTUM.
 Kaissis, Georgios, et al. “A machine learning algorithm predicts molecular subtypes in pancreatic ductal adenocarcinoma with differential response to gemcitabine-based versus FOLFIRINOX chemotherapy.” PloS one 14.10 (2019): e0218642.  Harder, Felix N., et al. “[18F] FDG PET/MRI enables early chemotherapy response prediction in pancreatic ductal adenocarcinoma.” EJNMMI research 11.1 (2021): 1-11.  Vaswani, Ashish, et al. “Attention is all you need.” Advances in neural information processing systems 30 (2017).  Woo, Sanghyun, et al. “Cbam: Convolutional block attention module.” Proceedings of the European conference on computer vision (ECCV). 2018.