TY - GEN
T1 - Diffusion Transformer U-Net for Medical Image Segmentation
AU - Chowdary, G. Jignesh
AU - Yin, Zhaozheng
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.
PY - 2023
Y1 - 2023
N2 - Diffusion model has shown its power on various generation tasks. When applying the diffusion model in medical image segmentation, there are a few roadblocks to remove: the semantic features required for the conditioning of the diffusion process are not well aligned with the noise embedding; and the U-Net backbone employed in these diffusion models is not sensitive to contextual information that is essential during the reverse diffusion process for accurate pixel-level segmentation. To overcome these limitations, we present a cross-attention module to enhance the conditioning from source images, and a transformer based U-Net with multi-sized windows for the extraction of various scales of contextual information. Evaluated on five benchmark datasets with different imaging modalities including Kvasir-Seg, CVC Clinic DB, ISIC 2017, ISIC 2018, and Refuge, our diffusion transformer U-Net achieves great generalization ability and outperforms all the state-of-the-art models on these datasets.
AB - Diffusion model has shown its power on various generation tasks. When applying the diffusion model in medical image segmentation, there are a few roadblocks to remove: the semantic features required for the conditioning of the diffusion process are not well aligned with the noise embedding; and the U-Net backbone employed in these diffusion models is not sensitive to contextual information that is essential during the reverse diffusion process for accurate pixel-level segmentation. To overcome these limitations, we present a cross-attention module to enhance the conditioning from source images, and a transformer based U-Net with multi-sized windows for the extraction of various scales of contextual information. Evaluated on five benchmark datasets with different imaging modalities including Kvasir-Seg, CVC Clinic DB, ISIC 2017, ISIC 2018, and Refuge, our diffusion transformer U-Net achieves great generalization ability and outperforms all the state-of-the-art models on these datasets.
KW - Diffusion model
KW - Medical Image Segmentation
KW - Transformer
KW - U-Net
UR - https://www.scopus.com/pages/publications/85174697835
U2 - 10.1007/978-3-031-43901-8_59
DO - 10.1007/978-3-031-43901-8_59
M3 - Conference contribution
AN - SCOPUS:85174697835
SN - 9783031439001
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 622
EP - 631
BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings
A2 - Greenspan, Hayit
A2 - Greenspan, Hayit
A2 - Madabhushi, Anant
A2 - Mousavi, Parvin
A2 - Salcudean, Septimiu
A2 - Duncan, James
A2 - Syeda-Mahmood, Tanveer
A2 - Taylor, Russell
PB - Springer Science and Business Media Deutschland GmbH
T2 - 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023
Y2 - 8 October 2023 through 12 October 2023
ER -