Isocenter optimal matching shift algorithm to verify the dose distribution in intensity-modulated radiation therapy through the stochastic property

Dose distribution verification using a film for assuring the quality of intensity-modulated radiation therapy for patients with head and neck cancer using an m3 (Brain Lab.) did not complement the planned dose distribution at points having large gradients in dose variations, although the isocenter position had few discrepancies. An effective way to solve this problem would be precisely investigate the reason for this unexpected shift and then to recover it to the correct position. Using Mathlab 7.1, we have developed a new software program called "isocenter optimal matching shift algorithm" (ISOMSA). To find the optimal ISOMSA for finding isocenter shifts, we compared two methods, one based on a statistical method of minimizing the variance (SMMV) and the other based on a statistical method of minimizing the mean (SMMM) of the setup error. To evaluate the two methods, we compared verification results for seven patients at the National Cancer Center by using both methods. The results showed that the number of points revealing more than a 9 % difference in SMMV was 2.86 %, which was less than the percentage, 8.0 %, from SMMM. While the average of the maximum percentage difference between the plan and the film in SMMV was 14.06 %, the percentage was 15.23 % in SMMM. These experimental results suggest that SMMV works better than the SMMM in finding setup errors. Clinically, this result is very important because the dose difference between the plan and the film may be analyzed more precisely by using the proposed method to reduce systematic errors.