Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Observational Study
. 2025 Sep:88:106993.
doi: 10.1016/j.crad.2025.106993. Epub 2025 Jun 17.

Development and validation of a deep learning-based automatic classification algorithm for the medial temporal lobe atrophy score using a multimodality cascade transformer

S J Lee  1 , D Lee  2 , C H Suh  3 , S Y Jeong  4 , H M Shin  2 , W Jung  2 , J Kim  2 , J-S Lim  5 , H S Kim  6 , S J Kim  6 , J-H Lee  7
Affiliations
Observational Study

Development and validation of a deep learning-based automatic classification algorithm for the medial temporal lobe atrophy score using a multimodality cascade transformer

S J Lee et al. Clin Radiol. 2025 Sep.

Abstract

Aim: The aim of this study was to develop and validate a deep learning-based automatic classification algorithm for the medial temporal lobe atrophy (MTA) score in patients with cognitive impairment.

Materials and methods: This retrospective, observational study included consecutive patients with cognitive impairment from a tertiary hospital between March 2017 and June 2021. We developed a deep learning-based model and a machine learning-based model to automate MTA classification. We reorganised the MTA scores into 3 classes (0/1), (2), and (3/4) then classified the right and left MTA scores separately. The internal testing and external testing datasets were applied and compared to validate the performance of the MTA prediction model.

Results: A total of 1694 patients were evaluated for the training dataset, and 297 patients evaluated for the internal testing dataset. 400 patients were evaluated for the external testing dataset. In the internal testing dataset, the accuracy was 0.82 and 0.87 for the left and right MTA classifications, respectively. In the external testing dataset, the accuracy was 0.82 and 0.85 for the left and right MTA classifications, respectively. When comparing the performance between a deep learning-based model and a machine learning-based model, the results were similar.

Conclusion: The deep learning- and machine learning-based automatic classification algorithms for the MTA score accurately classified the MTA score in patients with cognitive impairment.

PubMed Disclaimer

Conflict of interest statement

Conflict of interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Chong Hyun Suh reports financial support was provided by National Research Foundation of Korea. Woo Hyun Shim reports financial support was provided by Ministry of Science and ICT. Dongsoo Lee reports a relationship with VUNO Inc. that includes employment. Hye Min Shin reports a relationship with VUNO Inc. that includes employment. Wooseok Jung reports a relationship with VUNO Inc. that includes employment. Jinyoung Kim reports a relationship with VUNO Inc. that includes employment. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publication types

LinkOut - more resources