A comparative study of the state-of-the-art artificial intelligence models for automated plaque measurement in a carotid ultrasound Japanese cohort
Year: 2026
Type: article
Abstract: Background: Automated carotid intima-media thickness (cIMT) measurement from B-mode ultrasound is important for cardiovascular risk stratification. Direct comparisons of deep learning architectures on the same cohort under controlled conditions remain limited. This study aimed to compare the accuracy and computational cost of three deep learning architectures with the commercial AtheroEdge™ 2.0 (AE 2.0) system for carotid ultrasound analysis. Methods: This study evaluated three deep learning architectures: carotid non-attention double-stage convolutional neural network (cNonAtt-dsCNN), carotid attention-based single-stage UNet (cAtt-ssUNet), and carotid attention-based double-stage UNet (cAtt-dsUNet) against the commercial AE 2.0 system on 400 carotid ultrasound images from 200 Japanese patients. All deep learning models were trained using identical preprocessing, augmentation, hyperparameters (Adam optimizer, learning rate =1e-4, batch =32), and 5-fold cross-validation. Evaluation metrics included Dice similarity coefficient (DSC), Jaccard index, lumen-intima error (LIe), media-adventitia error (MAe), intima-media thickness error (IMTe), and computational cost. Results: 8 min) and ~1.7× more graphics processing unit (GPU) memory than cNonAtt-dsCNN (DSC: 81.5%, CC: 0.92, LIe: 0.09 mm, IMTe: 0.124 mm). cAtt-ssUNet was intermediate (DSC: 88.5%, CC: 0.95, LIe: 0.07 mm). AE 2.0 retained the lowest boundary errors (LIe: 0.012 mm, IMTe: 0.011 mm) with 6 ms inference. Conclusions: A clear trade-off between accuracy-efficiency exists across the evaluated methods. cAtt-dsUNet is optimal for high-accuracy research settings with adequate GPU resources, while cNonAtt-dsCNN and AE 2.0 are preferable for speed-critical clinical screening. The single-cohort Japanese dataset limits generalizability; multi-center, multi-ethnic validation is recommended before clinical deployment.
Authors: Nikhil Singh, Pankaj K. Jain, Rubeena Vohra, Arun K. Dubey, Mainak Biswas, Ekta Tiwari, Vinod Anand, Inder M. Singh, Esma R Isenovic, Amer M. Johri, Laura E. Mantella, John R Laird, Narendra N. Khanna, Zoltan Ruzsa, Rajesh Singh, Sanjay Saxena, Mustafa Al Maini, Klaudija Viskovic, Manudeep K Kalra, Andrew Nicolaides, George Kitas, Vijay Viswanathan, Janet N A Ajuluchukwu, Luca Saba, Jasjit S. Suri
Institutions: Bharati Vidyapeeth's College of Engineering, Delhi, Sai University, Indian Institute of Technology BHU, Banaras Hindu University, KIIT University, University of Belgrade, Vinča Institute of Nuclear Sciences, Queen's University, St. Helena Hospital, Adventist Heart Institute, Indraprastha Apollo Hospitals, University of Szeged, Uttaranchal University, International Institute of Information Technology, University Hospital for Infectious Diseases "Dr Fran Mihaljevic", Massachusetts General Hospital, University of Nicosia, University of Manchester, M.V. Hospital for Diabetes and Diabetes Research Centre, University of Cagliari, Idaho State University, Symbiosis International University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7223377927