Abstract
BACKGROUND
Onco-neurosurgical practice still relies heavily on pre-operatively acquired images to guide intra-operative decision-making for safe tumor removal, a practice with inherent pitfalls such as registration inaccuracy due to brain shift, and lack of real-time (functional) feedback. Exploiting the opportunity for real-time imaging of the exposed brain can improve intra-operative decision-making, neurosurgical safety and patient outcomes. Previously, we described functional Ultrasound (fUS) as a high-resolution, depth-resolved imaging technique able to detect functional regions and vascular morphology during awake resections. Here, we present for the first time fUS as a fully integrated, MRI/CT-registered imaging modality in the OR.
MATERIAL AND METHODS
fUS relies on high-frame-rate (HFR) ultrasound, making the technique sensitive for very small motions caused by vascular dynamics (µDoppler) and allowing measurements of changes in cerebral blood volume (CBV) with micrometer-millisecond precision. This opens up the possibility to 1) detect functional response, as CBV-changes reflect changes in metabolism of activated neurons through neurovascular coupling and 2) visualize in-vivo vascular morphology of tumor and healthy tissue. During a range of anesthetized and awake onco-neurosurgical procedures we acquired images of brain and spinal cord using conventional linear ultrasound probes connected to an experimental acquisition unit. Building on Brainlab’s ‘Cranial Navigation’ and ‘Intra-Operative Ultrasound’ modules, we could co-register our intra-operative Power Doppler Images (PDIs) to patient-registered MRI/CT-data. Using the ‘IGTLink’ research interface, we could access and store real-time tracking data for informed volume reconstructions in post-processing.
RESULTS
Intra-operative fUS could be registered to MRI/CT-images in real-time, showing overlays of PDIs at imaging depths of >5 centimeters. During meningioma resections, these co-registered PDIs revealed fUS’ ability to visualize the tumor’s feeding vessels and surrounding healthy vasculature prior to durotomy, with a level of detail unprecedented by conventional MRI-sequences. Comparing post-operatively reconstructed 3D-vascular maps of pre- and post-durotomy acquisitions, further confirmed the dural dependency of the vascular network feeding the tumor. During awake resections, fUS revealed distinct functional areas as activated during motor and language tasks.
CONCLUSION
fUS is a new real-time, high-resolution and depth-resolved imaging technique, combining characteristics uniquely beneficial for a potential image-guided resection tool. The successful integration of fUS in the onco-neurosurgical OR demonstrated by our team, is an essential step towards clinical integration of fUS, as well as the technique’s validation against modalities such as MRI and CT.
A Hardware Accelerator for Real Time Sliding Window Based Pedestrian Detection on High Resolution Images
