Multi-Contrast Hippocampal Subfield Segmentation for Ultra-High Field 7T MRI Data using Deep Learning

Daniel Ramsing Lund; Mette Tøttrup Gade; Tina Jensen; Thomas B. Shaw; Maciej Plocharski; Lasse Riis Østergaard; Steffen Bollmann; Markus Barth

Multi-Contrast Hippocampal Subfield Segmentation for Ultra-High Field 7T MRI Data using Deep Learning

Daniel Ramsing Lund, Mette Tøttrup Gade, Tina Jensen, Thomas B. Shaw, Maciej Plocharski, Lasse Riis Østergaard, Steffen Bollmann, Markus Barth

Research output: Contribution to book/anthology/report/conference proceeding › Conference abstract in proceeding › Research › peer-review

Abstract

Ultra-high field 7T MRI and the utilization of multiple MRI contrasts potentially enable a superior hippocampal subfield segmentation. A residual-dense fully convolutional neural network based on U-net, including a dilated-convolutional-block was implemented for hippocampal subfield segmentation. Two data sets were combined for training and mean DSC of 0.7723 was obtained. DSC was higher for larger subfields, which were undersegmented, while smaller subfields were oversegmented. Results were comparable to the atlas-based method ASHS, while providing a substantially faster processing time.

Original language	English
Title of host publication	Proceedings of the 2020 ISMRM & SMRT Virtual Conference & Exhibition, 08-14 August 2020
Publication date	24 Jul 2020
Article number	3521
Publication status	Published - 24 Jul 2020
Event	2020 ISMRM & SMRT Virtual Conference & Exhibition - Duration: 8 Aug 2020 → 14 Aug 2020

Conference

Conference	2020 ISMRM & SMRT Virtual Conference & Exhibition
Period	08/08/2020 → 14/08/2020

AUB Link

Search for the material in Aalborg University Library's search engine

Cite this

@inbook{90bd3879eb7b4e39a59199ac05dcb511,

title = "Multi-Contrast Hippocampal Subfield Segmentation for Ultra-High Field 7T MRI Data using Deep Learning",

abstract = "Ultra-high field 7T MRI and the utilization of multiple MRI contrasts potentially enable a superior hippocampal subfield segmentation. A residual-dense fully convolutional neural network based on U-net, including a dilated-convolutional-block was implemented for hippocampal subfield segmentation. Two data sets were combined for training and mean DSC of 0.7723 was obtained. DSC was higher for larger subfields, which were undersegmented, while smaller subfields were oversegmented. Results were comparable to the atlas-based method ASHS, while providing a substantially faster processing time.",

author = "Lund, {Daniel Ramsing} and Gade, {Mette T{\o}ttrup} and Tina Jensen and Shaw, {Thomas B.} and Maciej Plocharski and {\O}stergaard, {Lasse Riis} and Steffen Bollmann and Markus Barth",

year = "2020",

month = jul,

day = "24",

language = "English",

booktitle = "Proceedings of the 2020 ISMRM & SMRT Virtual Conference & Exhibition, 08-14 August 2020",

note = "2020 ISMRM & SMRT Virtual Conference & Exhibition ; Conference date: 08-08-2020 Through 14-08-2020",

}

Multi-Contrast Hippocampal Subfield Segmentation for Ultra-High Field 7T MRI Data using Deep Learning. / Lund, Daniel Ramsing; Gade, Mette Tøttrup; Jensen, Tina et al.
Proceedings of the 2020 ISMRM & SMRT Virtual Conference & Exhibition, 08-14 August 2020. 2020. 3521.

Research output: Contribution to book/anthology/report/conference proceeding › Conference abstract in proceeding › Research › peer-review

TY - ABST

T1 - Multi-Contrast Hippocampal Subfield Segmentation for Ultra-High Field 7T MRI Data using Deep Learning

AU - Lund, Daniel Ramsing

AU - Gade, Mette Tøttrup

AU - Jensen, Tina

AU - Shaw, Thomas B.

AU - Plocharski, Maciej

AU - Østergaard, Lasse Riis

AU - Bollmann, Steffen

AU - Barth, Markus

PY - 2020/7/24

Y1 - 2020/7/24

N2 - Ultra-high field 7T MRI and the utilization of multiple MRI contrasts potentially enable a superior hippocampal subfield segmentation. A residual-dense fully convolutional neural network based on U-net, including a dilated-convolutional-block was implemented for hippocampal subfield segmentation. Two data sets were combined for training and mean DSC of 0.7723 was obtained. DSC was higher for larger subfields, which were undersegmented, while smaller subfields were oversegmented. Results were comparable to the atlas-based method ASHS, while providing a substantially faster processing time.

AB - Ultra-high field 7T MRI and the utilization of multiple MRI contrasts potentially enable a superior hippocampal subfield segmentation. A residual-dense fully convolutional neural network based on U-net, including a dilated-convolutional-block was implemented for hippocampal subfield segmentation. Two data sets were combined for training and mean DSC of 0.7723 was obtained. DSC was higher for larger subfields, which were undersegmented, while smaller subfields were oversegmented. Results were comparable to the atlas-based method ASHS, while providing a substantially faster processing time.

UR - https://www.ismrm.org/20/program_files/DP09-03.htm

M3 - Conference abstract in proceeding

BT - Proceedings of the 2020 ISMRM & SMRT Virtual Conference & Exhibition, 08-14 August 2020

T2 - 2020 ISMRM & SMRT Virtual Conference & Exhibition

Y2 - 8 August 2020 through 14 August 2020

ER -

Multi-Contrast Hippocampal Subfield Segmentation for Ultra-High Field 7T MRI Data using Deep Learning

Abstract

Conference

AUB Link

Other files and links

Fingerprint

Cite this