Relations of image quality in on-line portal images and individual patient parameters for pelvic field radiotherapy

We have previously demonstrated that on-line portal imaging (OPI) can detect and correct significant errors in field set-up. Such errors occurred very frequently when irradiating the pelvic region and were typically detected after 10% of the field dose was delivered. The image quality on pelvic fields was, however, disappointing. The aims of the present study involving 566 pelvic fields on 13 patients were: 1. To study the machine- and patient-related factors influencing image quality. 2. To study the factors related to machine, patient and patient set-up, influencing the errors of field set-up. 3. To develop a method for predicting the camera settings. The OPI device consisted of a fluorescent screen scanned by a video camera. An image quality score on a scale 0-5 was given for 546/566 fields. In a univariate analysis, open field subtraction adversely affected the score (P < 0.001). The image score of anterior fields was significantly better than that of posterior fields (P < 0.001). Multivariate stepwise logistic regression showed that, in addition to anterior or posterior field (P < 0.001) and subtraction (P = 0.003), gender (P = 0.02) was also a significant predictor of image score. Errors requiring field adjustments were detected on 289/530 (54.5%) evaluable fields or 229/278 (82.4%) evaluable patient set-ups. Multivariate logistic regression showed that the probability of performing an adjustment was significantly related to gender (P < 0.001), image quality (P = 0.001) and AP-PA diameter (P = 0.04). The magnitude of adjustments made in the lateral direction correlated significantly (P < 0.0001) with patient bulk. The camera kV level with gain held constant showed an exponential dependency on dose rate at the image detector plate and can thus be predicted by treatment planning. © 1992 Springer-Verlag.