INTRAFRACTIONAL AND INTERFRACTIONAL REPRODUCIBILITY OF TANGENTIAL BREAST FIELDS - A PROSPECTIVE ONLINE PORTAL IMAGING STUDY

Purpose: A perception exists that weekly verification films accurately reflect the setup of the tangential breast portals. This prospective study was undertaken to assess patient movement during treatment and s etup reproducibility of tangential breast fields using electronic on-l ine portal imaging. Methods and Materials: Thirteen patients with carc inoma of the breast were treated on a linear accelerator equipped with an on-line portal imaging system. Patients were immobilized daily wit h an alpha cradle. The medial and lateral tangential fields were image d and 139 fractions, 225 portal fields, and 4450 images were obtained. Images were then analyzed off line and 22,250 measurements were made from these images. Anatomical features recorded include the lung area (LA), central lung distance (CLD), central breast distance (CBD), cent ral flash distance (CFD), and inferior central margin (ICM). Intrafrac tional variations were calculated for every portal held and fraction f or each patient. Interfractional variations were determined by finding the variance of intrafractional means for each patient. A population standard deviation for each of the five parameters for intra- and inte rfractional variations were determined. The simulation to treatment se tup errors were calculated for all five variables. Results: Lung area variation was 1.50 and 4.19 cm(2) [1 standard deviation (SD)] for intr a- and interfractional movement. Intrafractional variation for the oth er four variables ranged from 0.85 mm for ICM to 2.1 mm (1 SD) for CBD , while interfractional variations ranged from 3.2 to 6.25 mm for CBD and ICM, respectively. The simulation-to-treatment setup variation was greater than the interfractional variation for three of the five vari ables and was similar for the other two. Conclusions: On-line verifica tion of intrafractional variation shows a moderate deviation from the treatment setup position for all five parameters studied, while interf ractional variation showed even greater deviations for these five para meters. To cover the breast target in 95% of cases, margins of 7.70, 7 .70, and 10.30 mm corresponding to the CLD, CFD, and ICM distances, re spectively, are required.