Chamber Attention Network (CAN): Towards interpretable diagnosis of pulmonary artery hypertension using echocardiography

• Published in: Journal of advanced research 2024-09, Vol.63, p.103-115
• Main Authors: Sun, Dezhi, Hu, Yangyi, Li, Yunming, Yu, Xianbiao, Chen, Xi, Shen, Pan, Tang, Xianglin, Wang, Yihao, Lai, Chengcai, Kang, Bo, Bai, Zhijie, Ni, Zhexin, Wang, Ningning, Wang, Rui, Guan, Lina, Zhou, Wei, Gao, Yue
• Format: Article
• Language: English
• Subjects: Adult; Aged; Attention mechanism; Deep learning; Echocardiography; Echocardiography - methods; Female; Humans; Hypertension, Pulmonary - diagnosis; Hypertension, Pulmonary - diagnostic imaging; Male; Mathematics, Engineering, and Computer Science; Middle Aged; Neural Networks, Computer; Pulmonary Arterial Hypertension - diagnostic imaging; Pulmonary artery hypertension
• ISSN: 2090-1232; 2090-1224; 2090-1224
• DOI: 10.1016/j.jare.2023.10.013

[Display omitted] •We introduced a novel chamber attention network (CAN) that utilizes the Grad-CAM technique to assess the relevance of different chambers in identifying PAH.•The attention vector generated by the chamber attention module provides a quantitative measure of the chambers’ importance in PAH diagnosis, which aligns well with clinical knowledge and enhances the model’s interpretability.•Our CAN model achieved outstanding performance on a large training-validation dataset consisting of 13912 individual subjects.•Evaluation on an external test dataset from a different hospital demonstrated the superior generalization ability of our CAN model. Accurate identification of pulmonary arterial hypertension (PAH) in primary care and rural areas can be a challenging task. However, recent advancements in computer vision offer the potential for automated systems to detect PAH from echocardiography. Our aim was to develop a precise and efficient diagnostic model for PAH tailored to the unique requirements of intelligent diagnosis, especially in challenging locales like high-altitude regions. We proposed the Chamber Attention Network (CAN) for PAH identification from echocardiographic images, trained on a dataset comprising 13,912 individual subjects. A convolutional neural network (CNN) for view classification was used to select the clinically relevant apical four chamber (A4C) and parasternal long axis (PLAX) views for PAH diagnosis. To assess the importance of different heart chambers in PAH diagnosis, we developed a novel Chamber Attention Module. The experimental results demonstrated that: 1) The substantial correspondence between our obtained chamber attention vector and clinical expertise suggested that our model was highly interpretable, potentially uncovering diagnostic insights overlooked by the clinical community. 2) The proposed CAN model exhibited superior image-level accuracy and faster convergence on the internal validation dataset compared to the other four models. Furthermore, our CAN model outperformed the others on the external test dataset, with image-level accuracies of 82.53% and 83.32% for A4C and PLAX, respectively. 3) Implementation of the voting strategy notably enhanced the positive predictive value (PPV) and negative predictive value (NPV) of individual-level classification results, enhancing the reliability of our classification outcomes. These findings indicate that CAN is a feasible technique for AI-assisted PAH diagnosis, prov