B110-08 An Attention-Enhanced U-Net Neural Network Model for Airway Segmentation in Expiratory Computed Tomography Images

J Park; X Zhang; P K Rajaraman; A P Comellas; E A Hoffman; M Castro; S E Wenzel; N N Jarjour; M L Schiebler; E Israel; B D Levy; J V Fahy; S C Erzurum; K Sumino; C -L Lin

doi:10.1093/ajrccm/aamag162.518

Back

B110-08 An Attention-Enhanced U-Net Neural Network Model for Airway Segmentation in Expiratory Computed Tomography Images

Abstract

Peer reviewed

B110-08 An Attention-Enhanced U-Net Neural Network Model for Airway Segmentation in Expiratory Computed Tomography Images

J Park, X Zhang, P K Rajaraman, A P Comellas, E A Hoffman, M Castro, S E Wenzel, N N Jarjour, M L Schiebler, E Israel, …

American journal of respiratory and critical care medicine, Vol.212(Supplement_1), aamag162518

05/01/2026

DOI: 10.1093/ajrccm/aamag162.518

View Online

Abstract

Rationale This study introduces a new segmentation model for expiratory CT images by integrating a novel attention gate with a U-Net framework to improve airway continuity and completeness. Methods We analyzed data from 120 subjects across four cohorts: asthma, COPD, post-COVID-19, and healthy controls. Each cohort includes 25 subjects with residual volume (RV) scans, which were used for training, validation, and independent testing. We proposed a 3D segmentation framework incorporating an attention gate model, termed Averaged Multi-Gaussian Response (AMGR), integrating with the U-Net architecture, specifically optimized for expiratory airway segmentation. The proposed model was compared against two state-of-the-art methods, which are Fuzzy Attention Neural Network (FANN) and standard U-Net, to evaluate its performance. Quantitative assessments were conducted using Dice, Precision, and Recall metrics. Results The proposed model achieved a Dice score of 0.9629. It outperformed two baseline models: FANN (0.9084) and U-Net (0.9477). In terms of Precision, the proposed model had 0.9621, indicating superior segmentation performance compared with FANN (0.8370) and U-Net (0.9397). While FANN reported the highest Recall (0.9947), the proposed model showed a balanced trade-off between Precision and Recall (0.9647). Conclusion The AMGR approach stabilizes feature fusion by suppressing peripheral noise and enhancing distal airway continuity. The proposed model achieved a high Dice score, while maintaining balanced Precision and Recall for the multi-disease cohort. Overall, the model demonstrated robust noise suppression and improved airway continuity in the RV scans. This abstract is funded by: NIH

Neural networks

Details

Title: Subtitle: B110-08 An Attention-Enhanced U-Net Neural Network Model for Airway Segmentation in Expiratory Computed Tomography Images
Creators: J Park - University of Iowa
X Zhang
P K Rajaraman - University of Iowa
A P Comellas - University of Iowa
E A Hoffman - University of Iowa
M Castro - University of Kansas
S E Wenzel - University of Pittsburgh
N N Jarjour - UW Health University Hospital
M L Schiebler - UW Health University Hospital
E Israel - Brigham and Women's Hospital
B D Levy - Brigham and Women's Hospital
J V Fahy - University of California, San Francisco
S C Erzurum - Cleveland Clinic
K Sumino - Washington University in St. Louis
C -L Lin - University of Iowa
Resource Type: Abstract
Publication Details: American journal of respiratory and critical care medicine, Vol.212(Supplement_1), aamag162518
DOI: 10.1093/ajrccm/aamag162.518
ISSN: 1535-4970
eISSN: 1535-4970
Publisher: Oxford University Press; OXFORD
Grant note: NIH
This abstract is funded by: NIH
Language: English
Date published: 05/01/2026
Academic Unit: Roy J. Carver Department of Biomedical Engineering; Radiology; Pulmonary, Critical Care, and Occupational Medicine; ICTS; IIHR--Hydroscience and Engineering; Mechanical Engineering; Internal Medicine
Record Identifier: 9985164721402771

Metrics

1 Record Views