The broader impact/commercial potential of this I-Corps project is the development of a surgical assistive software that will improve overall healthcare provided to patients by ensuring successful endovascular treatments on the first attempt and reducing retreatment procedures in surgical suites. In current clinical practice, neuro-interventionalists cannot guarantee good outcomes, such as the successful healing of intracranial aneurysms immediately following treatment by an endovascular device. In such cases, treated patients have to wait a minimum of 3-6 months before their aneurysm is reassessed on medical imaging and the clinician decides if re-treatment is necessary. During this critical time, patients are still at risk of an aneurysm rupture, which may cause disastrous hemorrhagic strokes. Furthermore, if re-treatment is needed, the additional procedures have higher risk to the patient, and put further financial burden on hospitals and insurance companies. The proposed technology, a digital companion to medical imaging, can predict problem cases intra-procedurally, so that they may be immediately rectified, thus avoiding potential ruptures and reducing rates of re-treatments from 30% to 5%.
This I-Corps project is based on the development of a software tool to analyze intracranial aneurysm (IA) hemodynamics and predict long-term treatment outcomes using quantitative angiographic imaging. This is a completely autonomous method that rapidly performs three tasks for each angiogram. First, artificial intelligence (AI)-based detection and segmentation of the aneurysm dome are performed with a Dice coefficient of 0.84. Next, physics-based quantitative angiographic parameters related to the nature of blood flow are automatically extracted from the aneurysm dome. High correlation is observed with parameters extracted by a human expert (Pearson correlation of 0.87) but the technology is 1,000 times faster. Finally, the parameters are input into an AI prediction algorithm to predict 6-month occlusion probability (successful treatment), which has an area under the receiver operating characteristic curve of 77%. This prototype software is designed to be seamlessly integrated into the imaging suite and provide a prediction in under 1 second. This software is 10–20 times faster than current methods, which involve neuro-interventionalists leaving the operating suite, loading the images on a separate workstation, and then analyzing the device placement with respect to the aneurysm. Moreover, the current time-consuming method does not provide information regarding the future healing of the aneurysm with treatment.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
