Пожалуйста, используйте этот идентификатор, чтобы цитировать или ссылаться на этот ресурс: http://elar.urfu.ru/handle/10995/131073
Название: Benefits of Mirror Weight Symmetry for 3D Mesh Segmentation in Biomedical Applications
Авторы: Dordiuk, V.
Dzhigil, M.
Ushenin, K.
Дата публикации: 2023
Издатель: Institute of Electrical and Electronics Engineers Inc.
Библиографическое описание: Dordiuk, V, Dzhigil, M & Ushenin, K 2023, Benefits of Mirror Weight Symmetry for 3D Mesh Segmentation in Biomedical Applications. в 2023 IEEE Ural-Siberian Conference on Computational Technologies in Cognitive Science, Genomics and Biomedicine, CSGB 2023 - Proceedings: book. Institute of Electrical and Electronics Engineers Inc., стр. 100-107, 2023 IEEE Ural-Siberian Conference on Computational Technologies in Cognitive Science, Genomics and Biomedicine (CSGB), 28/09/2023. https://doi.org/10.1109/CSGB60362.2023.10329838
Dordiuk, V., Dzhigil, M., & Ushenin, K. (2023). Benefits of Mirror Weight Symmetry for 3D Mesh Segmentation in Biomedical Applications. в 2023 IEEE Ural-Siberian Conference on Computational Technologies in Cognitive Science, Genomics and Biomedicine, CSGB 2023 - Proceedings: book (стр. 100-107). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CSGB60362.2023.10329838
Аннотация: 3D mesh segmentation is an important task with many biomedical applications. The human body has bilateral symmetry and some variations in organ positions. It allows us to expect a positive effect of rotation and inversion invariant layers in convolutional neural networks that perform biomedical segmentations. In this study, we show the impact of weight symmetry in neural networks that perform 3D mesh segmentation. We analyze the problem of 3D mesh segmentation for pathological vessel structures (aneurysms) and conventional anatomical structures (endocardium and epicardium of ventricles). Local geometrical features are encoded as sampling from the signed distance function, and the neural network performs prediction for each mesh node. We show that weight symmetry gains from 1 to 3% of additional accuracy and allows decreasing the number of trainable parameters up to 8 times without suffering the performance loss if neural networks have at least three convolutional layers. This also works for very small training sets. © 2023 IEEE.
Ключевые слова: 3D MESH SEGMENTATION
BIOMEDICAL SEGMENTATION
HARD CONSTRAINTS
INVERSION INVARIANT
ROTATION INVARIANT
SYMMETRY IN NEURAL NETWORKS
WEIGHT SYMMETRY
CONVOLUTION
CONVOLUTIONAL NEURAL NETWORKS
HEART
MEDICAL APPLICATIONS
MULTILAYER NEURAL NETWORKS
3D MESH SEGMENTATION
3D MESHES
BIOMEDICAL SEGMENTATION
HARD CONSTRAINTS
INVERSION INVARIANT
MESH SEGMENTATION
NEURAL-NETWORKS
ROTATION INVARIANT
SYMMETRY IN NEURAL NETWORK
WEIGHT SYMMETRY
MESH GENERATION
URI: http://elar.urfu.ru/handle/10995/131073
Условия доступа: info:eu-repo/semantics/openAccess
Конференция/семинар: 2023 IEEE Ural-Siberian Conference on Computational Technologies in Cognitive Science, Genomics and Biomedicine, CSGB 2023
Дата конференции/семинара: 28 September 2023 through 29 September 2023
Идентификатор SCOPUS: 85180377048
Идентификатор PURE: 50627272
ISBN: 9798350307979
DOI: 10.1109/CSGB60362.2023.10329838
Сведения о поддержке: Russian Science Foundation, RSF: RSF 22-21-00930
This work has been supported by the grant of the Russian Science Foundation, RSF 22-21-00930. The computations were performed on the Uran supercomputer at the IMM UB RAS.
Карточка проекта РНФ: 22-21-00930
Располагается в коллекциях:Научные публикации ученых УрФУ, проиндексированные в SCOPUS и WoS CC

Файлы этого ресурса:
Файл Описание РазмерФормат 
2-s2.0-85180377048.pdf10,77 MBAdobe PDFПросмотреть/Открыть


Все ресурсы в архиве электронных ресурсов защищены авторским правом, все права сохранены.