On August 22, 2022, this year's OCG Promotion Award was presented during the DEXA 2022 conference in Vienna. VRVis researcher Theresa Neubauer received the award for her master thesis "Volumetric Tumor Segmentation on Multimodal Medical Images using Deep Learning", written at VRVis.
Her master thesis, which resulted from a research project of our Biomedical Image Informatics Group, is a great example of best practice in promoting young scientists as well as in interdisciplinary applied research: Under the direction of Katja Bühler and supervised by Maria Wimmer, VRVis was the scientific home of Theresa Neubauer, from where she advanced her research work in close cooperation with her scientific mentors Prof. Thomas Beyer from MedUni Vienna and Eduard Gröller from TU Wien. Theresa Neubauer's research focused on using innovative visual computing and AI techniques to merge multimodal images automatically for more efficient tumor detection to achieve holistic diagnostics.
Soft tissue tumors are tumors of various tissues (musculature, connective tissue, fatty tissue, nerve tissue). An essential part of their diagnosis and the determination of a successful therapy is the correct assessment of the tumors on medical scans. To do this, medical professionals visually analyze the tumor on MR, CT, and PET scans. These images thereby capture the tumor in different anatomical, functional and molecular contexts and thus also provide different information about the tumor, because different aspects and images are relevant for different clinical questions - e.g. biopsy, radiotherapy, or surgical planning. Until now, tumor segmentation has been a manual and time-consuming task requiring a great deal of radiologists' attention. Here, automated tools can be a valuable addition to everyday clinical practice.
However, to date, the potential of multi-modal data for tumor detection has only been exploited by a few established computer-aided image segmentation methods. In order to make a significant contribution here, Theresa Neubauer, an expert in the field of medical informatics, has developed a new method in the course of her diploma thesis "Volumetric Tumor Segmentation on Multi-modal Medical Images using Deep Learning", by means of which multi-modal information from different imaging procedures can be combined for a holistic, optimized diagnostic workflow. This involves fusing magnetic resonance, computed tomography, and positron emission tomography data to obtain a more accurate picture of the tumor. The image segmentation method developed uses machine learning (artificial intelligence) to learn complex image features and relationships between modalities to segment the tumor more efficiently and accurately.
Vienna, August 25, 2022