Priority attention network with Bayesian learning for fully automatic segmentation of substantia nigra from neuromelanin MRI

Tao Hu; Hayato Itoh; Masahiro Oda; Shinji Saiki; Nobutaka Hattori; Koji Kamagata; Shigeki Aoki; Kensaku Mori

doi:10.1117/12.2655112

3 April 2023 Priority attention network with Bayesian learning for fully automatic segmentation of substantia nigra from neuromelanin MRI

Tao Hu, Hayato Itoh, Masahiro Oda, Shinji Saiki, Nobutaka Hattori, Koji Kamagata, Shigeki Aoki, Kensaku Mori

Author Affiliations +

Proceedings Volume 12464, Medical Imaging 2023: Image Processing; 124643G (2023) https://doi.org/10.1117/12.2655112
Event: SPIE Medical Imaging, 2023, San Diego, California, United States

Conference Poster

Abstract

Neuromelanin magnetic resonance imaging (NM-MRI) has been widely used in the diagnosis of Parkinson’s disease (PD) for its significantly enhanced contrast between the PD-related structure, the substantia nigra (SN) and surrounding tissues. This paper proposes a novel network combining the priority gating attention and Bayesian learning for improving the accuracy of fully automatic SN segmentation from NM-MRI. Different from the conventional gated attention model, the proposed network uses the prior SN probability map for guiding the attention computation. Additionally, to lower the risks of over-fitting and estimate the confidence scores for the segmentation results, Bayesian learning with Monte Carlo dropout is applied in the training and testing phases. The quantitative results showed that the proposed network acquired the averaged Dice score of 79.46% in comparison with the baseline model 77.93%.

Citation Download Citation

Tao Hu, Hayato Itoh, Masahiro Oda, Shinji Saiki, Nobutaka Hattori, Koji Kamagata, Shigeki Aoki, and Kensaku Mori "Priority attention network with Bayesian learning for fully automatic segmentation of substantia nigra from neuromelanin MRI", Proc. SPIE 12464, Medical Imaging 2023: Image Processing, 124643G (3 April 2023); https://doi.org/10.1117/12.2655112

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available