A phantom study of the geometric accuracy of computed tomographic and magnetic resonance imaging stereotactic localization with the Leksell stereotactic system
C. Yu et al., A phantom study of the geometric accuracy of computed tomographic and magnetic resonance imaging stereotactic localization with the Leksell stereotactic system, NEUROSURGER, 48(5), 2001, pp. 1092-1098
OBJECTIVE: To assess the spatial accuracy of magnetic resonance imaging (MR
I) and computed tomographic stereotactic localization with the Leksell ster
eotactic system.
METHODS: The phantom was constructed in the shape of a box, 164 mm in each
dimension, with three perpendicular arrays of solid acrylic rod, 5 mm in di
ameter and spaced 30 mm apart within the phantom. In this study, images fro
m two different MRI scanners and a computed tomographic scanner were obtain
ed using the same Leksell (Elekta Instruments, Stockholm, Sweden) head fram
e placement. The coordinates of the rod images in the three principal plane
s were measured by using a tool provided with Leksell GammaPlan software (E
lekta Instruments, Norcross, CA) and were compared with the physical phanto
m measurements.
RESULTS: The greatest distortion was found around the periphery, and the le
ast distortion (<1.5 mm) was present in the middle and most other areas of
the phantom. In the phantom study using computed tomography, the mean value
s of the maximum errors for the x, y, and z axes were 1.0 mm (range, 0.2-1.
3 mm), 0.4 mm (range, 0.1-0.8 mm), and 3.8 mm (range, 1.9-5.1 mm), respecti
vely. The mean values of the maximum errors when using the Philips MRI scan
ner (Philips Medical Systems, Shelton, CT) were 0.9 mm (range, 0.4-1.7 mm),
0.2 mm (range, 0.0-0.7 mm), and 1.9 mm (range, 1.3-2.3 mm), respectively.
Using the Siemens MRI scanner (Siemens Medical Systems, New York, NY), thes
e values were 0.4 mm (range, 0.0-0.7 mm), 0.6 mm (range, 0.0-1.0 mm), and 1
.6 mm (range, 0.8-2.0 mm), respectively. The geometric accuracy of the MRI
scans when using the Siemens scanner was greatly improved after the impleme
ntation of a new software patch provided by the manufacturer. The accuracy
also varied with the direction of phase encoding.
CONCLUSION: The accuracy of target localization for most intracranial lesio
ns during stereotactic radiosurgery can be achieved within the size of a vo
xel, especially by using the Siemens MRI scanner at current specifications
and with a new software patch. However, caution is warranted when imaging p
eripheral lesions, where the distortion is greatest.