A comparative study to evaluate the efficacy of on board imaging with cone beam CT using target registration in patients with lung tumors undergoing stereotactic body radiation therapy and comparison with ExacTrac using skeletal registration on Novalis Tx

Background: Stereotactic body radiation therapy is an advanced technique, which delivers ablative doses to lung lesions. Target verification is done either by orthogonal x-rays or cone beam CT. This study was undertaken to compare these two verification methods. Aim: To evaluate the efficacy of ExacTrac and Cone Beam Computed Tomography (CBCT) for target repositioning while delivering Stereotactic Body Radiation Therapy (SBRT) for lung lesions and derive the population-based margin. Materials and Methods: All patients who had undergone SBRT for lung lesions from February to September 2009 were involved. Patients were immobilized using the BodyFix double vacuum immobilization system, indexed to the computed tomography (CT) simulator and treatment machine. Four-dimensional (3-D) scan was done to generate internal target volume (ITV) and a free breathing CT scan for planning was done on the BrainLab iPlan 4.1 software. During treatment, patient′s position was verified using ExacTrac and CBCT. The resulting vertical, lateral, and longitudinal shifts were noted. The random and systematic error were calculated and the margin recipe derived using the Van Herk formula. Results: Sixteen patients had undergone SBRT for lung tumors from February to September 2009. Data from eight patients who had undergone 34 sessions of SBRT was analyzed. The systematic error for lateral, longitudinal, and vertical shifts for ExacTrac and CBCT were 3.68, 4.27, 3.5 mm and 0.53, 0.38, 0.70 mm, respectively. The random error were 1.10, 1.51, 1.96 mm and 0.32, 0.81, 0.59 mm. The lateral, longitudinal and vertical Van Herk margin recipe for ExacTrac were 9.98, 11.72, 10.18 mm, respectively, and for CBCT was 2.17, 1.53,1.55 mm. Conclusions: The systematic and random errors for CBCT were significantly lesser as compared to the errors with Exactrac