MRI-Based Treatment Planning with Pseudo CT Generated Through Atlas Registration
J Uh*, T Merchant, C Hua, St. Jude Childrens Research Hospital, Memphis, TNTH-C-WAB-11 Thursday 10:30AM - 12:30PM Room: Wabash Ballroom
Purpose: To develop a magnetic resonance imaging (MRI)-based treatment planning method without the need of acquiring patient-specific computed tomography (CT) dataset.
Methods: A pseudo CT, providing electron density information for dose calculation, was generated by deforming atlas CT images of other patients. Two approaches, arithmetic averaging and Gaussian process regression (Hofmann 2008), were used to combine the deformed images into a single dataset. The required deformation was derived from registering corresponding atlas MR images to those of the subject of interest. We tested this method on 7 patients aged 8-12 years with common pediatric brain tumors. We used a leave-one-out procedure, i.e., the 6 pairs of MR/CT images other than those of each patient of interest served as atlas images. The original patient-specific CT was replaced with the pseudo CT and the dose distribution was recalculated to quantify the difference. We used Chi-evaluation (Bakai 2003) for dose distribution comparison. Doses to the planning target volume (PTV) and organs at risk (OARs) were compared.
Results: The pseudo CT generated by arithmetic averaging showed a high similarity to the patient-specific CT (correlation coefficient, 0.75-0.85). The calculated dose distribution was in close agreement with original dose. Nearly the entire patient volume (98.4 - 99.8%) satisfied the criteria of Chi-evaluation (<3mm and 3% maximum dose). Dose indices of V95 and D95 in PTV differed by up to 2.2%. The average differences in maximum and mean doses to OARs ranged from 0.3% to 1.0%. The pseudo CT by Gaussian process showed similar performances for the images included in this study (P=0.27, Chi-evaluation) despite longer computation time.
Conclusion: MR-based treatment planning with pseudo CT generated through atlas registration is feasible for pediatric brain tumor patients. The doses calculated from pseudo CT agreed well with those from patient-specific CT.