Attention-Based UNet Deep Learning Model for Plaque Segmentation in Carotid Ultrasound for Stroke Risk
Pankaj K Jain1, Abhishek Dubey1,2, Luca Saba3
1School of Biomedical Engineering, Indian Institute of Technology (BHU), Varanasi 221005, India.
Insights
An AI model using Attention-UNet accurately detects carotid plaques, aiding early stroke and cardiovascular disease (CVD) risk assessment. This deep learning approach improves diagnosis of challenging bright and fuzzy plaque images.
Area of Science:
- Medical Imaging and Artificial Intelligence
- Cardiovascular Disease Diagnostics
Background:
- Stroke and cardiovascular diseases (CVD) are major global health concerns.
- Early detection is crucial to prevent mortality and reduce healthcare costs.
- Current diagnostic methods lack automation and clinical accuracy.
Purpose of the Study:
- To develop an AI-based deep learning model for automated detection and severity prediction of CVD and stroke.
- To identify carotid plaques in internal carotid artery (ICA) and common carotid artery (CCA) images using an attention-channel-based UNet model.
Main Methods:
- Utilized an attention-channel-based UNet deep learning model.
- Trained on multi-ethnic datasets: 970 ICA images (UK), 679 integrated CCA images (Japanese diabetics, Hong Kong post-menopausal women).
- Applied rotation transformation to augment CCA data and K5 cross-validation (80% train, 20% test) for evaluation.
Main Results:
- Attention-UNet demonstrated superior visual plaque segmentation compared to UNet, UNet++, and UNet3P.
- Achieved a correlation coefficient (CC) of 0.96 and an Area Under the Curve (AUC) of 0.97.
- Outperformed benchmark models in CC (0.93-0.92) and AUC (0.964-0.965) values.
Conclusions:
- The Attention-UNet model effectively segments difficult bright and fuzzy carotid plaques.
- This AI approach offers a promising tool for early stroke risk assessment.
- Presents a multi-ethnic, multi-center, and racially unbiased study for stroke risk evaluation.
Abstract:
Stroke and cardiovascular diseases (CVD) significantly affect the world population. The early detection of such events may prevent the burden of death and costly surgery. Conventional methods are neither automated nor clinically accurate. Artificial Intelligence-based methods of automatically detecting and predicting the severity of CVD and stroke in their early stages are of prime importance. This study proposes an attention-channel-based UNet deep learning (DL) model that identifies the carotid plaques in the internal carotid artery (ICA) and common carotid artery (CCA) images. Our experiments consist of 970 ICA images from the UK, 379 CCA images from diabetic Japanese patients, and 300 CCA images from post-menopausal women from Hong Kong. We combined both CCA images to form an integrated database of 679 images. A rotation transformation technique was applied to 679 CCA images, doubling the database for the experiments. The cross-validation K5 (80% training: 20% testing) protocol was applied for accuracy determination. The results of the Attention-UNet model are benchmarked against UNet, UNet++, and UNet3P models. Visual plaque segmentation showed improvement in the Attention-UNet results compared to the other three models. The correlation coefficient (CC) value for Attention-UNet is 0.96, compared to 0.93, 0.96, and 0.92 for UNet, UNet++, and UNet3P models. Similarly, the AUC value for Attention-UNet is 0.97, compared to 0.964, 0.966, and 0.965 for other models. Conclusively, the Attention-UNet model is beneficial in segmenting very bright and fuzzy plaque images that are hard to diagnose using other methods. Further, we present a multi-ethnic, multi-center, racial bias-free study of stroke risk assessment.
