A novel, inexpensive method of image coregistration for applications in image-guided surgery using augmented reality

Augmented reality (AR) is a technique in which an overlay of a virtual image to a live picture is performed to create a new image in which both original images coexist as a single image. This results in the visualization of internal structures through overlying tissues. The objective was to describe an easy, inexpensive, and successful method to coregister with AR in an image-guided surgery setting using the resources at hand.

Cortical information was obtained with a volumetric acquisition of 200 0.8-mm thick, cerebral magnetic resonance imaging scans in an axial T1-weighted sequence. For the venous anatomy, a contrast phase at 7 mm/s velocity was used. This data was reconstructed in a three-dimensional fashion using MRIcro software (v. 1.37, freeware, courtesy of Chris Rorden) and was overlaid to a digital image of the cerebral cortex either pre- or intraoperatively.

Eight patients were studied. There was an adequate coregistration in seven of the patients as confirmed by intraoperative ultrasound, frame-based stereotaxy, or obvious anatomic homology between the three-dimensional magnetic resonance imaging scan virtual reconstruction and the live image obtained during surgery. AR was not possible in one case of a cerebellar lesion.

AR coregistration capabilities are adequate when revised by other intraoperative guidance devices. When performed with "freeware" software and conventional digital cameras, it is relatively inexpensive, which makes it a potential tool for surgical planning and noncontinuous intraoperative guidance in neurosurgery. Its largest drawbacks are the inability to function in deep-seated lesions and its lack of tracking devices, which gives it a noncontinuous coregistration nature.