Encoder-Decoder Architectures for Clinically Relevant Coronary Artery Segmentation: Applications in Stenosis Detection

Coronary X-ray angiography is a crucial clinical procedure for the diagnosis and treatment of coronary artery disease, which accounts for roughly 16% of global deaths every year. However, the images acquired in these procedures have low resolution and poor contrast, making lesion detection and assessment challenging. Accurate coronary artery segmentation not only helps mitigate these problems, but also allows the extraction of relevant anatomical features for further analysis by quantitative methods. Although automated segmentation of coronary arteries has been proposed before, previous approaches have used non-optimal segmentation criteria, leading to less useful results. Most methods either segment only the major vessel, discarding important information from the remaining ones, or segment the whole coronary tree, based mostly on contrast information, producing a noisy output that includes vessels that are not relevant for diagnosis. We adopt a better-suited clinical criterion and segment vessels according to their clinical relevance. Additionally, we simultaneously perform catheter segmentation, which may be useful for diagnosis due to the scale factor provided by the catheter’s known diameter, and is a task that has not yet been performed with good results. To derive the optimal approach, we conducted an extensive comparative study of encoder-decoder architectures trained on a combination of focal loss and a variant of generalized dice loss. Based on the EfficientNet and the UNet++ architectures, we propose a line of efficient and high-performance segmentation models using a new decoder architecture, the EfficientUNet++, whose best-performing version achieves a generalized dice score of 0.9202 ± 0.0356, and artery and catheter class dice scores of 0.8858 ± 0.0461 and 0.7627 ± 0.1812.

Arlindo Oliveira

Arlindo Oliveira was born in Angola and lived in Mozambique, Portugal, Switzerland, California, Massachusetts, and Japan. He obtained his BSc and MSc degrees from Instituto Superior Técnico (IST) and his PhD degree from the University of California at Berkeley. He is a distinguished professor of IST, president of the INESC group, member of the board of Caixa Geral de Depósitos, researcher at INESC-ID, and member of the National Council for Science, Technology and Innovation and of the Advisory Board of the Science and Technology Options Assessment (STOA) Panel of the European Parliament. He authored four books, translated into different languages, and hundreds of scientific and newspaper articles. He has been on the boards of several companies and institutions and is a past president of IST, of the Portuguese Association for Artificial Intelligence, and of INESC-ID. He was the head of the Portuguese node of the European Network for Biological Data (ELIXIR), visiting professor at MIT and at the University of Tokyo, and a researcher at CERN, INESC, Cadence Research Laboratories and Electronics Research Labs of UC Berkeley. He is a member of the Portuguese Academy of Engineering and a senior member of IEEE. He received several prizes and distinctions, including the Technical University of Lisbon / Santander prize for excellence in research, in 2009.IST