Development of AI assisted intra-cranial and extra-cranial vessel wall imaging and analysis techniques for characterization of high-risk cerebrovascular atherosclerosis

About This Grant

According to a report from the AHA, stroke is a leading cause of long-term disability and the fourth most common cause of mortality. The estimated direct and indirect annual cost of stroke is $68.9 billion, and projected costs in 2050 exceed $2.2 trillion. Cerebrovascular atherosclerotic disease is estimated to be the direct cause of 440,000 ischemic strokes each year . Veterans have a disproportionately high rate of ischemic stroke from cerebrovascular disease reflecting a disproportionate number of cardiovascular risk factors. There is a compelling need to develop efficient but comprehensive carotid and intracranial vessel wall imaging protocols and standardized image analysis techniques to identify and categorize cerebrovascular atherosclerotic plaques that are high risk to cause initial or recurrent ischemic stroke. More advanced tools to identify veterans with unstable atherosclerotic plaques are needed. According to recent surveys of the American Society of Neuroradiology membership, major obstacles to the implementation of extracranial and intracranial vessel wall MRI for the evaluation of atherosclerotic plaque include limited expertise of clinical interpretation (45% and 53.1% respectively) and limited understanding of applications (44% and 50.5% respectively). If these and other obstacles were overcome, 51.1% and 56.4% of neuroradiologists indicated that they would perform the technique to better assess arterial wall pathologies. A standardized scoring system incorporating stratified high-risk plaque features could help overcome these barriers through standardization of interpretation and creating a common lexicon for communication of vulnerable features and their associated stroke risk. While a newly published white paper put forth a carotid PlaqueRADS risk score, there is no similar risk score for intracranial plaques that cause ischemic stroke. Further, the efficacy of PlaqueRADS needs to be clinically evaluated and combined assessment of carotid and intracranial plaques is needed for meaningful stroke risk assessment. To meet these challenges, an efficient combined carotid-brain vessel wall MRI protocol to assess both carotid and intracranial plaque is needed. We have developed specialized carotid MRI coils that can be combined with brain coils and AI-aided plaque analysis tools to facilitate clinical implementation of carotid-brain MRI, and will further develop a clinical Plaque-RADS system to assess risk of carotid and intracranial plaques. Using these advances, we will implement the full suite of technology developed for this study on two major MRI scanner platforms (Siemens and Philips) at two VA medical centers: combined carotid-brain MRI, an AI-aided method that can segment intracranial plaque locations to obtain plaque morphology, an AI- aided carotid plaque segmentation method, and a carotid-intracranial Plaque-Reporting and Data System (Plaque-RADS) score in a cohort of prospectively scanned subjects. The ability to effectively categorize carotid and intracranial plaque components would be of significant benefit, providing reassurance to those veterans with low-risk lesions while supporting intervention for those with unstable plaques. Combined carotid and intracranial MRI should provide a far superior approach for effectively triaging veterans with large vessel disease to appropriate treatment. The development and validation of an intracranial Plaque-RADS score that can be combined with the existing carotid Plaque-RADS score will set the stage for large-scale prospective studies aimed at stratifying stroke risk and reducing the incidence of stroke.