C-Arm CBCT with Model-Based Image Reconstruction for Neurosurgical Interventions

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The occurrence of adverse events in neurosurgery and subsequent toll on patients is considerably high, with a reported 16-28% of neurosurgical interventions resulting in adverse events and up to a majority of these events constituting hemorrhagic complications, posing a great risk for devastating neurological outcome. Accurate targeting of neurological soft tissue structures during interventional procedures is a central part of reducing adverse events, but despite the use of preoperative images and navigation systems, changes in intracranial anatomy during surgery continues to contribute to targeting errors. Thus, high-quality intraoperative imaging can play an integral role in guiding neurosurgical interventions - in particular, C-arms are an increasingly prevalent modality that is highly compatible with the surgical setup but are currently limited in soft tissue image quality. Therefore, the proposed research addresses the unmet needs of high-quality imaging sufficient to resolve soft tissue structures at minimum radiation dose, driving advances in three key areas: (a) image quality approaching (and in some contexts even exceeding) diagnostic CT; (b) minimizing dose such that a given imaging task can still be accomplished; and (c) image reconstruction speed compatible with clinical workflow. In particular, the proposed research develops advanced 3D model-based image reconstruction (MBIR) methods for C-arm cone-beam CT (CBCT) specifically for neurosurgical interventions to advance C-arm CBCT image quality beyond limitations of conventional reconstruction to a level suitable to soft tissue visualization (brain tissues, ventricles, and intracranial hemorrhage - ICH) and facilitating reduction in radiation dose. This will be accomplished over three Specific Aims: (1) develop model-based image reconstruction for C-arm CBCT specifically adapted to intraoperative, neurosurgical imaging by implementing a GPU-accelerated reconstruction (1-5 min target reconstruction time) and developing methods for rigorous image quality assessment; (2) improve soft tissue visualization for image-guided neurosurgical interventions by leveraging MBIR to achieve low contrast resolution (5-20 HU target contrast); and (3) reduce radiation dose in C-arm CBCT to target levels of 1-5 mGy. With these capabilities, a series of preclinical experiments will be conducted to evaluate the improvement in soft tissue visualization for tasks such as ICH detection and guidance of interventional devices toward soft tissue targets. Successful completion of the proposed research offers a paradigm shift toward broader utilization of advanced C-arm CBCT imaging methods in neurosurgery, whereby a reduction in complications and need for revision surgery is anticipated by virtue of high-quality 3D intraoperative imaging. The benefits of soft tissue visualization made possible by MBIR, in combination with significantly reduced radiation dose and reconstruction time, will benefit a spectrum of image-guided soft tissue interventions in neurosurgery and beyond, enabling more accurate, more effective, and safer procedures.