Angiographic dataset for stenosis detection

Published: 11 November 2021| Version 2 | DOI: 10.17632/ydrm75xywg.2
Contributors:
,
,
,
,
,

Description

In this dataset, we present a set of angiographic imaging series of one hundred patients who underwent coronary angiography using Coroscop (Siemens) and Innova (GE Healthcare) image-guided surgery systems at the Research Institute for Complex Problems of Cardiovascular Diseases (Kemerovo, Russia). All patients had angiographically and/or functionally confirmed one-vessel coronary artery disease (≥70% diameter stenosis by quantitative coronary analysis or 50 - 69% with FFR (fractional flow reserve) ≤ 0.80 or stress echocardiography evidence of regional ischemia). For the purpose of our study, significant coronary stenosis was defined according to 2017 US appropriate use criteria for coronary revascularization in patients with stable ischemic heart disease. The study design was approved by the Local Ethics Committee of the Research Institute for Complex Issues of Cardiovascular Diseases (approval letter No. 112 issued on May 11, 2018). All participants provided written informed consent to participate in the study. Coronary angiography was performed by the single operator according to the indications and recommendations stated in the 2018 ESC/EACTS Guidelines on myocardial revascularization. The presence or absence of coronary stenosis was confirmed by the same operator using angiography imaging series according to the 2018 ESC/EACTS Guidelines on myocardial revascularization. Angiographic images of the radiopaque overlaid coronary arteries with stenotic segments were selected and then converted into separate images. An interventional cardiologist rejected non-informative images and selected only those containing contrast passages through a stenotic vessel. A total of 8325 grayscale images (100 patients) of 512×512 to 1000×1000 pixels were included in the dataset. Data were labeled using LabelBox, a free version of SaaS (Software as a Service). We additionally estimated the size of the stenotic region by computing the area of the bounding box. Similar to the Common Objects in Context dataset, we divided objects by their area into three types: small (area < 322), medium (322 ≤ area ≤ 962), and large (area > 962). 2509 small objects (30%), 5704 medium objects (69%), and 113 large objects (1%) were obtained in the input data.

Files

Categories

Cardiovascular Medicine, Object Detection, Machine Learning, Angiography, Cardiovascular Imaging, Stenosis, Deep Learning

License