Dataset related to the article "A token-mixer architecture for CAD-RADS classification of coronary stenosis on multiplanar reconstruction CT images"

This record contains raw data related to the article "A token-mixer architecture for CAD-RADS classification of coronary stenosis on multiplanar reconstruction CT images" A B S T R A C T<br> Background and objective: In patients with suspected Coronary Artery Disease (CAD), the severity of stenosis needs<br> to be assessed for precise clinical management. An automatic deep learning-based algorithm to classify coronary<br> stenosis lesions according to the Coronary Artery Disease Reporting and Data System (CAD-RADS) in multiplanar<br> reconstruction images acquired with Coronary Computed Tomography Angiography (CCTA) is proposed.<br> Methods: In this retrospective study, 288 patients with suspected CAD who underwent CCTA scans were included.<br> To model long-range semantic information, which is needed to identify and classify stenosis with challenging<br> appearance, we adopted a token-mixer architecture (ConvMixer), which can learn structural relationship over<br> the whole coronary artery. ConvMixer consists of a patch embedding layer followed by repeated convolutional<br> blocks to enable the algorithm to learn long-range dependences between pixels. To visually assess ConvMixer<br> performance, Gradient-Weighted Class Activation Mapping (Grad-CAM) analysis was used.<br> Results: Experimental results using 5-fold cross-validation showed that our ConvMixer can classify significant<br> coronary artery stenosis (i.e., stenosis with luminal narrowing ≥50%) with accuracy and sensitivity of 87% and<br> 90%, respectively. For CAD-RADS 0 vs. 1–2 vs. 3–4 vs. 5 classification, ConvMixer achieved accuracy and<br> sensitivity of 72% and 75%, respectively. Additional experiments showed that ConvMixer achieved a better<br> trade-off between performance and complexity compared to pyramid-shaped convolutional neural networks.<br> Conclusions: Our algorithm might provide clinicians