Improving the assessment of vascular complexity in peripheral artery occlusive disease

In this work we propose a method to improve the visual assessment of vascular complexity in cine-angiography images from patients affected by peripheral artery occlusive disease (PAOD); in particular, we aim at evaluating the inter-clinician variability when scoring vascular complexity over "raw" and "processed" cine-angiography videos. The proposed workflow consists of 2 main steps: (i) conversion of the cine-angiographies to single static images with a broader field of view (FOV), and (ii) automatic segmentation of the vascular trees. In order to accomplish steps (i) and (ii) we make use of approaches based on machine-learning and deep-learning, respectively. We assessed the method on 20 cine-angiography acquisitions and asked three experienced interventional cardiologists to visually examine and then score the vascular complexity on cine-angiography (video), static image (grayscale) and segmented image (segmented). The inter-class correlation coefficient (ICC) was computed as inter-observer agreement metric and to account for possible systematic error, that depends on the experience of the raters. Absolute agreement was higher over the segmented image (ICC = 0.956) compared to the video (ICC = 0.761) and the grayscale image (ICC = 0.918). The 95% confidence level was statistically in favour of the segmented image; systematic error among raters was found. Results suggest that extracting the vascular tree from cine-angiography can substantially improve the reliability of visual assessment of vascular complexity in PAOD. Next steps will consist of the identification of a quantitative index as complexity score, in order to both speed up the clinical workflow and reduce the subjective error.