• A Novel Pipeline for Adrenal G...

Treffer: A Novel Pipeline for Adrenal Gland Segmentation: Integration of a Hybrid Post-Processing Technique with Deep Learning.

Title:
A Novel Pipeline for Adrenal Gland Segmentation: Integration of a Hybrid Post-Processing Technique with Deep Learning.
Authors:
Fayemiwo M; School of Computing, Engineering, and Intelligent Systems, Ulster University, Londonderry, Northern Ireland, UK. m.fayemiwo@ulster.ac.uk., Gardiner B; School of Computing, Engineering, and Intelligent Systems, Ulster University, Londonderry, Northern Ireland, UK., Harkin J; School of Computing, Engineering, and Intelligent Systems, Ulster University, Londonderry, Northern Ireland, UK., McDaid L; School of Computing, Engineering, and Intelligent Systems, Ulster University, Londonderry, Northern Ireland, UK., Prakash P; Mike Wiegers Department of Electrical and Computer Engineering, Kansas State University, Manhattan, KS, USA., Dennedy M; School of Medicine, National University of Ireland, Galway, Ireland.
Source:
Journal of imaging informatics in medicine [J Imaging Inform Med] 2025 Dec; Vol. 38 (6), pp. 3698-3710. Date of Electronic Publication: 2025 Mar 04.
Publication Type:
Journal Article
Language:
English
Journal Info:
Publisher: Springer Nature Country of Publication: Switzerland NLM ID: 9918663679206676 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 2948-2933 (Electronic) Linking ISSN: 29482925 NLM ISO Abbreviation: J Imaging Inform Med Subsets: MEDLINE
Imprint Name(s):
Original Publication: [Cham, Switzerland] : Springer Nature, [2024]-
MeSH Terms:
Deep Learning* , Adrenal Glands*/diagnostic imaging , Tomography, X-Ray Computed*/methods , Image Processing, Computer-Assisted*/methods , Radiographic Image Interpretation, Computer-Assisted*/methods, Humans ; Algorithms
References:
Sci Rep. 2020 Mar 19;10(1):5068. (PMID: 32193485)
PeerJ Comput Sci. 2021 Jun 29;7:e607. (PMID: 34307860)
Sci Rep. 2019 Jun 21;9(1):9007. (PMID: 31227772)
Med Image Anal. 2024 Oct;97:103280. (PMID: 39096845)
Comput Methods Programs Biomed. 2021 Oct;210:106375. (PMID: 34500139)
Med Image Anal. 2021 Apr;69:101894. (PMID: 33421919)
Int J Comput Assist Radiol Surg. 2020 Jun;15(6):981-988. (PMID: 32350786)
Sci Rep. 2022 Mar 10;12(1):3948. (PMID: 35273282)
BMC Med Inform Decis Mak. 2020 Jul 9;20(Suppl 3):119. (PMID: 32646419)
Nat Methods. 2021 Feb;18(2):203-211. (PMID: 33288961)
Comput Methods Programs Biomed. 2018 Oct;164:87-100. (PMID: 30195434)
PLoS One. 2014 Nov 17;9(11):e111126. (PMID: 25401954)
Comput Biol Med. 2021 Sep;136:104749. (PMID: 34388467)
Grant Information:
R01 EB028848 United States EB NIBIB NIH HHS; 20/US/3676 Ireland SFI_ Science Foundation Ireland; R01EB028848 Foundation for the National Institutes of Health; MC_PC_20021 United Kingdom MRC_ Medical Research Council; STL/5521/19 HSC Research and Development
Contributed Indexing:
Keywords: Adrenal gland; CT segmentation; Image processing; Test-time augmentation
Entry Date(s):
Date Created: 20250304 Date Completed: 20251213 Latest Revision: 20251215
Update Code:
20251215
PubMed Central ID:
PMC12701139
DOI:
10.1007/s10278-025-01449-y
PMID:
40038136
Database:
MEDLINE

Weitere Informationen

Accurate segmentation of adrenal glands from CT images is essential for enhancing computer-aided diagnosis and surgical planning. However, the small size, irregular shape, and proximity to surrounding tissues make this task highly challenging. This study introduces a novel pipeline that significantly improves the segmentation of left and right adrenal glands by integrating advanced pre-processing techniques and a robust post-processing framework. Utilising a 2D UNet architecture with various backbones (VGG16, ResNet34, InceptionV3), the pipeline leverages test-time augmentation (TTA) and targeted removal of unconnected regions to enhance accuracy and robustness. Our results demonstrate a substantial improvement, with a 38% increase in the Dice similarity coefficient for the left adrenal gland and an 11% increase for the right adrenal gland on the AMOS dataset, achieved by the InceptionV3 model. Additionally, the pipeline significantly reduces false positives, underscoring its potential for clinical applications and its superiority over existing methods. These advancements make our approach a crucial contribution to the field of medical image segmentation.
(© 2025. The Author(s).)

Declarations. Competing Interests: The authors declare no competing interests.