Login
- Original title:
- Deep learning based analysis of medical imaging data in oncology with a focus on radiation therapy
- Translated title:
- Deep learning an medizinischen Bilddaten in der Onkologie zur Anwendung in der Strahlentherapie
- Author:
- Lang, Daniel Manfred
- Year:
- 2023
- Document type:
- Dissertation
- Faculty/School:
- TUM School of Natural Sciences
- Advisor:
- Wilkens, Jan J. (Prof. Dr.)
- Referee:
- Wilkens, Jan J. (Prof. Dr.); Pfeiffer, Franz (Prof. Dr.)
- Language:
- en
- Subject group:
- PHY Physik
- Keywords:
- Convolutional Neural Networks, Vision Transformer, Self-supervised Learning, Radiation Therapy, Survival Analysis
- Translated keywords:
- Convolutional Neural Networks, Vision Transformer, Selbstlernende Modelle, Strahlentherapie, Ereigniszeitanalyse
- TUM classification:
- MED 540
- Abstract:
- Deep learning algorithms' ability for medical imaging data based prediction of endpoints in radiation therapy has been studied, allowing for non-invasive testing. Networks were customized such that complete tumor volumes could be analyzed. Transfer and self-supervised models have been developed to overcome the need for large datasets. Modern vision transformers have been extended. An architecture able to handle time-to-event data has been established to model temporal endpoints.
- Translated abstract:
- Das Potential von Deep Learning Algorithmen zur Modellierung relevanter Endpunkten der Strahlentherapie wurde untersucht. Dabei wurde die Netzwerkstruktur jeweils so angepasst, dass komplette Tumorvolumina berücksichtigt werden konnte. Transfer und selbstlernende Modelle wurden entwickelt um der Herausforderung kleiner Datensätze zu begegnen. Moderne Vision Transformer wurden erweitert. Ein Intervallzeitmodell wurde erstellt um eine Modellierung temporärer Endpunkte zu ermöglichen.
- WWW:
- https://mediatum.ub.tum.de/?id=1690422
- Date of submission:
- 09.11.2022
- Oral examination:
- 05.07.2023
- File size:
- 11336971 bytes
- Pages:
- 129
- Urn (citeable URL):
- https://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:91-diss-20230705-1690422-1-7
- Last change:
- 04.08.2023
BibTeX