Compromised Accuracy of Stereotactic Target Delineation Associated with Computed Tomography-Based Frame Registration: A Comparative Analysis of Magnetic Resonance Imaging-Computed Tomography Fusion

Abstract

<b><i>Introduction:</i></b> Recent advancements in stereotactic neurosurgical techniques have become increasingly reliant on image-based target planning. We devised a case-phantom comparative analysis to evaluate the target registration errors arising during the magnetic resonance imaging (MRI)-computed tomography (CT) image fusion process. <b><i>Methods:</i></b> For subjects whose preoperative MRI and CT images both contained fiducial frame localizers, we investigated discrepancies in target coordinates derived from frame registration based on either MRI or CT. We generated a phantom target through an image fusion process, merging the framed CT images with their corresponding reference MRIs after masking their fiducial indicators. This phantom target was then compared with the original during each instance of target planning. <b><i>Results:</i></b> In our investigative study with 26 frame registrations, a systematic error in the <i>y</i>-axis was observed as −0.89 ± 0.42 mm across cases using either conventional CT and/or cone-beam CT (O-arm). For the <i>z</i>-axis, errors varied on a case-by-case basis, recording at +0.64 ± 1.09 mm with a predominant occurrence in those merged with cone-beam CT. Collectively, these errors resulted in an average Euclidean error of 1.33 ± 0.93 mm. <b><i>Conclusion:</i></b> Our findings suggest that the accuracy of frame-based stereotactic planning is potentially compromised during MRI-CT fusion process. Practitioners should recognize this issue, underscoring a pressing need for strategies and advancements to optimize the process.