Accuracy of stereotactic localisation with magnetic resonance compared to CT scan: experimental findings

Stereotactic localisation of proper targets can be obtained with Magnetic Resonance imaging (MRI), pending correction of the well-known inaccuracy of MRI in reproducing exact geometrical dimension. The aim of the experimental work was to determine the real extent of the MRI distortion and to correct it.

PVC phantoms and fixed human brains were used as target simulators: stereotactic targeting was obtained with CT and MRI: the results were compared and, finally, the real stereotactic values were measured on a commercial stereotactic device. To optimise MRI targeting it was necessary to abate the "chemical-shift", to modify the head-coil receiver (tuning the "off-resonance" frequencies) and to correct the targeting directly on the images (especially for the Y axis values); careful quality control and environmental tests were also necessary to maintain good performances along the time.

172 measurements were carried out using both CT, plain X-ray and MRI on PVC phantom. The values obtained from CT were considered as reference. Significant geometrical accuracy was found with CT targeting, except for the Z co-ordinate, due to the slice thickness. Mean value differences between the targets on MRI and their real geometric position are about 1 mm on X and Z axises and about 2 mm on Y axis. Similar results were obtained on fixed brains, where absolute values of X and Y co-ordinates of the simulators were measured using a millimetre grid placed over the brain slice.

Experimental findings concerning stereotactic targeting with MRI suggest accuracy adequate for clinical practice, even when extreme geometrical precision is required, such as in radiosurgery or functional stereotaxy. Careful quality control and intensive experimental setting of the MRI device are mandatory to obtain satisfactory results.