Spatial distortion in MRI with application to stereotactic neurosurgery

The aim of this work was to implement a thorough method for quantifying the errors introduced to frame-based neurosurgical stereotactic procedures by the use of MRI. Chang & Fitzpatrick's reversed gradient distortion correction method was used, in combination with a phantom, to measure these errors. Spatial distortion in MR images of between 1 mm and 2 mm was measured. Further analysis showed that this typically introduced an additional error in the coordinate of the actual treatment point of 0.7 mm. The implications of this are discussed. The main source of distortion in the MR images used for stereotaxis was found to be the head ring. A comparison between imaging sequences and MR scanners revealed that the spatial distortion depends mainly on the bandwidth per pixel of the sequence rather than other differences in the imaging sequences. By comparison with a phase map distortion correction technique, the imaging parameters required to allow successful distortion correction with the reversed gradient method were identified. The most important was the use of full Fourier spin echo acquisitions. The reversed gradient correction method was applied to two contemporary EPI techniques. Considerable improvement was seen in the production of ADC maps after the images had been corrected for distortion. The method also was shown to be valid in application to BOLD fMRI data