Computer Vision News - January 2022
58 Non-Invasive Planning of Coronary Intervention into the coronary arteries while x-ray imaging is acquired. The cardiologist visually assesses whether the coronaries are normal or if a stenosis (narrowing) is present. Upon stenosis diagnosis, the severity is determined, and stent placement is planned. Planning requires accurate measurement of the stenosis length and diameter, taking into account the number of occurrences throughout the coronary branches. Additionally, an adequate stent needs to be selected to prevent misalignment and a need for a repeat procedure. The planning stage may require additional contrast injections to better visualize problem areas, as well as time-consuming overview of the images to verify adequate anatomical understanding. The additional imaging is harmful for the patient as well as the crew in the cath-lab, as X-ray imaging uses ionizing radiation. Also, continuous overview of the images is costly as it adds use-time of the cath-lab. Utilizinganewtechnologyfor3Dreconstruction from 2D x-ray images can provide a solution to the above-described challenge. The input for this is 2 image-sequences of coronary angiography acquired from different angles. Using state-of-the-art deep learning A new Technology for Non-Invasive Planning of Coronary Intervention Heart disease continues being the number one killer in the western world, leading to ~25% of deaths in the USA. Consequently, cardiac procedures, specifically diagnostic catheterization and percutaneous coronary interventions (PCI), amount to$10Bworldwide, making it one of the largest markets in the medical world. Belowwe describe an example of how artificial intelligence (AI) provides nowadays an impressive contribution to improve the treatments provided by modern medicine with exceptional clinical results in an economical manner. When suspicion of cardiac pathology arises, a diagnostic cardiac angiography is performed. During this procedure, contrast dye is injected CORONARY 3D RECONSTRUCTION FROM X-RAY
Made with FlippingBook
RkJQdWJsaXNoZXIy NTc3NzU=