Improvement of the total clinical accuracy of gamma knife radiosurgery and assessment of the clinical impact of MR spatial distortions





Keywords: image guidance, magnetic resonance imaging, stereotactic assisted, gamma knife, technique

Interactive Manuscript

Ask Questions of this Manuscript:

   



Abstract

     The total clinical accuracy of GK treatment delivery may be impaired due to the spatial distortion inherent in the MR images used for target volume definition in 3D space. T
     his work presents a time-efficient method to improve the total clinical accuracy of GK radiosurgery by correcting the MR images for background field distortions which, as we have published previously, are a major contributor to accuracy degradation (~1mm).
     
     The proposed method is based on the acquisition of an extra MR sequence that adds only minutes to the total patient scan-time and an appropriate software tool written in Matlab to process the “corrected” MR series. The method was benchmarked at 1.5 T using phantoms and then used to assess the effect of the sequence–dependent spatial distortions in 12 patient treatments with a total number of 80 relatively small metastases (less than 2cm in diameter, with 88 % of them less than 1cm and 55% less than 0.5cm). Using the extra MR sequence acquired for all patients, a “corrected” MR series was produced in one-to-one correspondence for every “original” Gd-enhanced T1W series routinely used for target definition. Then the targets corresponding to the different metastases were contoured on both the “original” and “corrected” series. Patient plans based on the “original” MR series were finally produced.
     The prescription dose was 25 Gy and the target coverage more than 99% for all targets contoured in the “original” MR images. For the same plans, the 25Gy target coverage was found to be less than 93% for 41% of the targets contoured in the “corrected” MR images (54% of the targets with diameter <0.5cm, 25% with diameter 0.5cm-1cm and 20% with diameter >1cm). For these targets, a dose of less than 23Gy (18–23Gy) instead of the prescribed 25Gy covered the 99% of the target volume (dose difference >8%).
     This evaluation did not compare MR images to CT images.
     There is a fundamental inaccuracy induced by background field spatial distortions on MRI. Small targets especially may be incorrectly covered by routine planning methods. 
     The proposed methodology was found to minimize the distortions and thus to improve significantly (in sub-voxel dimensions) the total clinical accuracy.


Acknowledgements

Project Roles: