Real time estimation of brain tissue dislocation and deformation : a phantom study

The main source of error during targeting of abnormalities in the brain during surgery is the phenomenon of brain shift. Brain shift makes the pre-operative treatment plan, which is developed based on the pre-operative images, outdated during the course of the surgery. Hence, there is a need to accurately estimate the brain tissue shift and to update the treatment plan in real time. In this paper, a method is proposed to track the dislocations and deformations of the brain tissue in real time using various image processing techniques with the objective of updating on-line planning for non-invasive procedures such as tissue ablation using High Intensity Focused Ultrasound (HIFU). The energy delivery is planned through a precise craniotomy by coupling HIFU source(s) directly on to dura-mater without opening it (thereby, non-invasive to brain tissue). Coherent Point Drift method is used for the non-rigid registration of ultrasound images in a suitably designed skull and brain phantoms. The method gave a maximum rms error of less than 4.582 mm and the computation took approximately 72.7 seconds, which is sufficient in the context of HIFU based ablation of point-by-point lesioning in the targeted regions. For the estimation of the dislocation of a targeted feature in the phantom, a correlation method, for ultrasound reference image to subsequent on-line images is proposed. Various laboratory trials showed a maximum error of 0.4 mm and an average computation time of under 0.5 seconds