Kalkhof, John ; Gonzalez, Camila ; Mukhopadhyay, Anirban (2022)
Disentanglement Enables Cross-Domain Hippocampus Segmentation.
19th International Symposium on Biomedical Imaging. Kolkata, India (28.-31.03.2022)
doi: 10.1109/ISBI52829.2022.9761560
Konferenzveröffentlichung, Bibliographie
Kurzbeschreibung (Abstract)
Limited amount of labelled training data are a common problem in medical imaging. This makes it difficult to train a well-generalised model and therefore often leads to failure in unknown domains. Hippocampus segmentation from magnetic resonance imaging (MRI) scans is critical for the diagnosis and treatment of neuropsychatric disorders. Domain differences in contrast or shape can significantly affect segmentation. We address this issue by disentangling a T1- weighted MRI image into its content and domain. This separation enables us to perform a domain transfer and thus convert data from new sources into the training domain. This step thus simplifies the segmentation problem, resulting in higher quality segmentations. We achieve the disentanglement with the proposed novel methodology ’Content Domain Disentanglement GAN’, and we propose to retrain the UNet on the transformed outputs to deal with GAN-specific artefacts. With these changes, we are able to improve performance on unseen domains by 6-13% and outperform stateof-the-art domain transfer methods.
| Typ des Eintrags: | Konferenzveröffentlichung |
|---|---|
| Erschienen: | 2022 |
| Autor(en): | Kalkhof, John ; Gonzalez, Camila ; Mukhopadhyay, Anirban |
| Art des Eintrags: | Bibliographie |
| Titel: | Disentanglement Enables Cross-Domain Hippocampus Segmentation |
| Sprache: | Englisch |
| Publikationsjahr: | 26 April 2022 |
| Ort: | Piscataway, NJ |
| Verlag: | IEEE |
| Buchtitel: | IEEE ISBI 2022 Proceedings: 2022 IEEE International Symposium on Biomedical Imaging |
| Veranstaltungstitel: | 19th International Symposium on Biomedical Imaging |
| Veranstaltungsort: | Kolkata, India |
| Veranstaltungsdatum: | 28.-31.03.2022 |
| DOI: | 10.1109/ISBI52829.2022.9761560 |
| Kurzbeschreibung (Abstract): | Limited amount of labelled training data are a common problem in medical imaging. This makes it difficult to train a well-generalised model and therefore often leads to failure in unknown domains. Hippocampus segmentation from magnetic resonance imaging (MRI) scans is critical for the diagnosis and treatment of neuropsychatric disorders. Domain differences in contrast or shape can significantly affect segmentation. We address this issue by disentangling a T1- weighted MRI image into its content and domain. This separation enables us to perform a domain transfer and thus convert data from new sources into the training domain. This step thus simplifies the segmentation problem, resulting in higher quality segmentations. We achieve the disentanglement with the proposed novel methodology ’Content Domain Disentanglement GAN’, and we propose to retrain the UNet on the transformed outputs to deal with GAN-specific artefacts. With these changes, we are able to improve performance on unseen domains by 6-13% and outperform stateof-the-art domain transfer methods. |
| Freie Schlagworte: | feature disentanglement, domain generalisation, distribution shift |
| Fachbereich(e)/-gebiet(e): | 20 Fachbereich Informatik 20 Fachbereich Informatik > Graphisch-Interaktive Systeme |
| Hinterlegungsdatum: | 15 Jun 2023 07:52 |
| Letzte Änderung: | 13 Jul 2023 15:49 |
| PPN: | 509626203 |
| Export: | |
| Suche nach Titel in: | TUfind oder in Google |
 |
Frage zum Eintrag |
Optionen (nur für Redakteure)
 |
Redaktionelle Details anzeigen |
Drucken
Impressum