DeeBMED · Deep learning and Bayesian inference for medical imaging
Horizon 2020 — Marie Skłodowska-Curie Actions
- Duration
- 2016-10-01 → 2018-09-30
- EU contribution
- €177,599
- Participants
- 1
- Scheme
- MSCA-IF-EF-ST
Lines connect the coordinator with its partners.
Results in brief
Deep learning and Bayesian inference for medical imaging
In the European Union and worldwide the number of people affected by modern civilization diseases like cancer increases continuously. As a consequence, high rise in treatment related costs is noticed, and a specific work up and care of patients is greatly needed. Unfortunately, the early-stage diagnostics is hampered by significant inter- and intraobserver variability resulting in an adequate treatment delay and the confirmation of the diagnosis. A possible solution is the emerging field of medical imaging. Medical imaging aims at processing and analyzing medical scans to further support physicians in their daily routines. However, there are still many obstacles that limit the wide usage of medical imaging, such as: low number of digitalized cases, large size of images and low availability of annotated images (an image with a description), complicated patters in medical scans. Within the DeeBMED project I proposed to tackle these problems by utilizing a probabilistic framework called Variational Auto-Encoders (VAEs). VAEs allow to model relationships among quantities like an image, x, a disease label, y, and a latent factors, z, using probability distributions and statistical (Bayesian) inference. See the diagrams for details. The DeeBMED project consisted of two main lines of research, namely, the development of the probabilistic framework, and the development of deep learning techniques for medical imaging. Within the first research direction I aimed at exploring possible extensions of the encoder and the prior in order to properly model data representation. The second line of research was focused on adapting deep learning methods to large images like medical scans.
Data: CORDIS, © European Union
Project objective
Diseases characteristic for modern western civilization, such as cancer, diabetes or cardiovascular disorders, lead to millions of deaths per year in the European Union. In order to decrease this enormous quantity, medical imaging should be widely available at early diagnostics and every stage of a therapy. Nowadays, there are various diagnostics techniques including CT, PET, MRI, however, analysis of a medical image is time-consuming and expensive. Development of new effective automatic tool for medical imaging will appear a new strategy in highly specific control of incidences and disease progression. The aim of the DeeBMED project is to develop powerful automatic medical imaging tool that can cope with main problems associated with complex images like medical scans: multimodality of data distribution, large number of dimension and small number of examples, small amount of labeled data, multi-source learning, and robustness to transformations. In this project I will propose a probabilistic framework that combines different deep neural networks (DNN), such as feedforward nets, convolutional nets, Gaussian processes. I will apply DNN to model probabilistic relationships among a medical scan, a disease label, and hidden variables representing latent factors in data. In the case of a small sample size DNN are prone to overfitting. A possible remedy for that is Bayesian learning, however, it is still challenging how to apply it to DNN. In this project I will use various approaches: modelling weights uncertainty, Dropout, Bayesian Distillation. As the result I predict identification of the first highly effective medical imaging analysis tool that in the future will be widely used by radiologists in medical institutes in the whole EU. Novel automation will drastically reduce time and costs of analysis and provide more accessible diagnostics. The project will be carried out at the University of Amsterdam, under the supervision of Prof. Max Welling.
Original text from CORDIS.
Participants
- UNIVERSITEIT VAN AMSTERDAM · AmsterdamCoordinatorNetherlands
Links
Data: CORDIS, © European Union