Overcoming Challenging Clinical Scenarios in Non-invasive Coronary Artery Imaging using Next Generation Photon Counting Detector CT and Artificial Intelligence - PROJECT SUMMARY/ABSTRACT Coronary artery disease (CAD) remains the main cause of morbidity and mortality in the United States. CT imaging provides fast non-invasive assessment of CAD with a high sensitivity and negative predictive value when the arterial lumen can be clearly visualized. However, in patients with heavily calcified plaques or coronary stents, the lumen can be obscured, leading to an overestimation of the degree of luminal stenosis. The recent introduction of photon-counting detector CT (PCD-CT) has partially alleviated these concerns because of its excellent spatial resolution (best in-plane spatial resolution of 0.125 mm). However, substantial limitations remain. Excessive image noise occurs in the high-spatial-resolution images, which precludes robust quantitation of percent luminal stenosis and visualization of features associated with high-risk plaques. Further, even with the 66 ms temporal resolution from dual-source technology, motion artifacts remain a major concern, especially in challenging scenarios such as the emergency department, where high heart rates, arrhythmias, and inadequate breath holding are common. In short, methods to control image noise and motion artifacts are critically needed to take full advantage of cardiac PCD-CT’s diagnostic potential in all patients. Additionally, non-ideal properties of photon-counting detectors, particularly charge sharing, limit spectral separation, the accuracy of material decomposition, and the success of plaque characterization. Innovative solutions using coincidence counting technologies have been proposed by our team but need the partnership of a detector/scanner manufacturer to implement. Our objective is to address these limitations in cardiac CT, namely image noise, motion artifact, and limited spectral separation due to charge sharing. Our approach will use the next-generation dual-source PCD-CT system and artificial intelligence (AI) techniques to accurately assess CAD in humans, especially in patients with heavily calcified, stented, or high-risk plaques. Working with our industry partner, Siemens Healthcare, we will develop the next generation of photon counting detectors with coincidence counting. Our proposal is highly significant. Robust, accurate, non-invasive imaging of calcified and stented coronary arteries or high-risk plaques in a single non-invasive exam will greatly reduce the need for invasive diagnostic imaging, reducing the overall time and cost to comprehensively evaluate CAD. To accomplish our objectives will require numerous physics, engineering, and algorithm innovations, including novel coincidence counting and charge-sharing correction, noise reduction, motion correction, and material decomposition algorithms. These advances will culminate in a small clinical study to demonstrate not merely that the images are “better,” as is so often done, but that the next-generation dual-source PCD-CT, in combination with AI techniques, provide clinically significant improvements in the diagnosis and management of challenging to image patients with suspected CAD. Additionally, the technological developments accomplished will benefit all of CT imaging.
