Department of Biomedical Imaging and Image-guided Therapy
Position: PHD Student
christoph.fuerboeck@meduniwien.ac.at
Keywords
Artificial Intelligence
Research group(s)
- Computational Imaging Research Lab
Head: Georg Langs
Research Area: Research at CIR is roughly grouped around 3 research lines: Machine Learning & Neuroimaging, Computer Aided Diagnosis and Quantification, Computer Vision and Pattern Recognition
Members:
Research interests
Application of machine learning algorithms in medical imaging. The main focus lies on the research of novel deep learning based methods as well as novel applications of these methods - currently concentrating on breast cancer related projects. It is one of my personal goals to enhance the understanding of disease (e.g. breast cancer sub-types) and provide tools to support the decisions of physicians.
Techniques, methods & infrastructure
Deep learning (DL) can be used a variety of applications, including segmentation, classification and regression tasks. In my current research, DL models are used for classification tasks. Medical imaging data of different modalities (Magnet Resonance Imaging, Positron Emission Tomography) is used as the input for the model to predict underlying properties of breast cancer lesions, such as breast cancer sub-types. The identification of sub-types using non-invasive imaging would benefit early assessment. In addition the DL model is used to find visual traits in the form of shared and discriminating phenotypic concepts associated with the sub-types, enabling the linking of image features to underlying molecular characteristics of tumors. The concepts are calculated by clustering intermediate feature representations of the parcellated images and the connection to the sub-types is assessed by testing with concept activation vectors.
Selected publications
- Fürböck, C. et al. (2022) ‘Identifying Phenotypic Concepts Discriminating Molecular Breast Cancer Sub-Types’, Medical Image Computing and Computer Assisted Intervention – MICCAI 2022, pp. 276–286. Available at: http://dx.doi.org/10.1007/978-3-031-16449-1_27.