gms | German Medical Science

Objective: Spatial image distortion in magnetic resonance imaging is a source of inaccuracy in stereotactic procedures like biopsies or deep brain stimulation. Among other factors, B0-field strength plays an important role in this phenomenon. Currently, 3-Tesla scanners are continuously replacing the former generation of 1.5-Tesla scanners in the clinical field. In order to obtain the advantages of 3T-MRI without losing accuracy, a suitable distortion correction is becoming more important. We evaluated the quality of the built-in distortion correction in a 1.5T and a 3T-scanner against the gold standard of CT-imaging.

Method: A cylindrical phantom with seven preset points was mounted on a stereotactic frame with a combined MRI/CT localizer. We then performed a CT- and two MRI-scans (Siemens Somatom Espree (1.5T) and Siemens Somatom Verio (3T)) to acquire an isotropic (1 x 1 x 1 mm) dataset. Two commonly used 3D-sequences (VIBE, MPRAGE) were acquired and automatic distortion correction was performed by the scanner hardware. Using stereotactic planning software (Praezis Plus, Precisis AG, Heidelberg), global stereotactic transformation of all nine datasets was performed. Finally, coordinates of all seven preset points were independently identified and recorded by two investigators and compared against the gold standard (CT).

Results: The smallest deviation from the gold standard (mean ± SD) was achieved by 1.5-tesla distortion corrected VIBE (1.02 ± 0.35 mm) whereas the corresponding 3-Tesla sequence (VIBE corrected) reached a deviation of 1.78 ± 0.89 mm. Interestingly, the most major deviation was found in the uncorrected 1.5-tesla MPRAGE sequence with 2.94 ± 1.91 mm (corrected 1.5-tesla MPRAGE 2.05 ± 0.86mm). Overall and regardless of field strength, spatial accuracy was higher in VIBE than in MPRAGE sequences. As expected, targets near the isocenter reached a higher accuracy (1.39 ± 0.51mm) than targets located in the periphery of the phantom (2.77 ± 1.31 mm).

Conclusions: We recommend the mandatory use of spatial distortion correction for stereotactic procedures even when using 1.5-Tesla scanners. By using distortion corrected VIBE sequences, spatial deviations can be reduced down to reasonable tolerances on 3-Tesla hardware.