Effects of treatment setup variation on beam's eye view dosimetry for radiation therapy using the multileaf...

The purpose of this study is to quantify and compare retrospectively the effects of treatment setup variation on beam's eye view (BEV) dosimetry for radiation therapy using a multileaf collimator (MLC) vs. cerrobend block.

A study was performed on a group of 18 patients with cancer of the head and neck, lung, and pelvis who were treated with irregularly shaped fields. The BEV dosimetry of the fields shaped with cerrobend blocks and the MLC was measured with films at the depth of dose prescription in a solid water phantom. A "one-half-leaf" insertion convention was used to shape the MLC. In addition, an average of 15 sequential daily port films was taken per patient during the course of radiotherapy. The port films were aligned with the prescription film for each patient. Systematic error and random error of treatment setup for each patient were calculated. The effects of setup variation were incorporated by convolving the patient portal imaging data with the corresponding BEV film dosimetry. Two parameters were used to quantify the BEV dosimetry. First, the field penumbra width was calculated, which represented the average of the normal separations between 20 and 80% isodose lines along the prescription outline. Second, the ratio of areas covered by the 90 and 20% isodose lines, A90/20, was determined. The BEV dosimetry was then characterized with and without the effects of treatment setup variation. In addition, the difference in BEV dosimetry between the cerrobend block and the MLC was used to estimate the corresponding changes in tumor control probability (TCP). These changes were also compared to the changes in TCP for the treatment with or without the effects of random setup variation.

With or without daily setup variation, the use of cerrobend block was more favorable than the MLC in terms of the field penumbra width and A90/20 for all treatment sites. In the absence of daily variation, the MLC field penumbra width was on average 1.3 mm larger than that of the cerrobend block, and 0.9 mm larger in the presence of daily setup variation. Similarly, the ratio A90/20 of the cerrobend block was on average 0.03 larger than that of MLC without daily setup variation, and 0.02 with daily setup variation. The difference in field penumbra width and A90/20 between the MLC and the cerrobend block was slightly reduced due to the effects of daily setup variation. For both the cerrobend block and the MLC, daily setup variation produced a significant increase in the field penumbra width, 2.3 mm for the cerrobend block and 1.9 mm for the MLC, and a decrease in the A90/20, 0.06 for the former and 0.05 for the latter. The change due to the daily setup variation was about a factor of 2 larger than the changes due to replacing the cerrobend block with the MLC. Using the TCP model, the change in TCP due to the daily setup variation was more than a factor of 3 larger than the change in TCP due to replacing the cerrobend block with the MLC. It was noted that the average changes in the penumbra, the A90/20 and the TCP calculated for the patient population did not adequately describe the changes for the individual patient.

Our results do not show significant dosimetric differences between the MLC and the cerrobend block in conventional radiation treatment, whether or not daily setup variation was taken into consideration. The effects of daily setup variation alone produced a larger dosimetric change. The same results were obtained when the data were applied to calculate changes in TCP. For optimal radiation therapy, efforts should be concentrated on reducing daily setup variation. Our results also demonstrate the importance of frequent evaluation of MLC treatment using electronic portal imaging devices.