Pii: S0167-8140(99)00171-1

Purpose: The goals of this study were to evaluate the use of electronic portal imaging device (EPID) paper images as off-line veri®cation tools and to assess the feasibility of replacing portal ®lms by EPID printed images. Materials and methods: Electronic portal images were acquired using a video-based imaging system. After contrast enhancement, these images were printed and compared to portal ®lms when prescribed, and judged about their usefulness for off-line veri®cation. A total of 2025 images were acquired from 322 ®elds on 137 patients. The images were shown to eight radiation oncologists and two senior residents in radiation oncology, each one of them judging ®elds relevant to his (her) daily practice. The questions asked were related to the choice of important anatomical structures and the visibility of such structures, the usefulness of the printed images, the comparison with portal ®lms and the possible replacement of such ®lms by paper images. Results: Answers to the different questions were treated as quantitative scores. For the visibility question, means and standard deviations were calculated for each individual structure, then a global score was obtained for a given treatment site. Means and standard deviations were also computed for the comparison question. Proportions and con®dence intervals were used for the other questions. The results show that EPID paper images are useful for some treatment sites such as breast, thorax, prostate, abdomen, pelvis (other than rectum) and axilla. The image quality remains insuf®cient for some other sites such as head and neck and spine. Conclusion: Although global anatomical landmarks scores are good, the usefulness score is not always as high because some essential anatomical structures scores must be taken into account. There is also a strong habit factor related to acceptance of EPID printed images as veri®cation tools. As long as they see more and more images, radiation oncologists can more easily visualize anatomical structures and are less stringent when evaluating the ef®ciency of EPID paper images as off-line veri®cation tools. q 2000 Elsevier Science Ireland Ltd. All rights reserved.