[Evolution of the use of the portal imaging device: prospective study over three years]

To describe the evolution of the use of the electronic portal imaging device (EPID) over three periods.

From 1990, as part of the quality assurance research programs, the radiotherapy department of the G.-F. Leclerc Centre of Dijon used EPID systems in a prospective fashion. During the first of the three periods (PER 1:1990-1993), the study consisted of analysis criteria determination, software efficiency improvement and a selection of patients who could benefit from the method. Eight hundred and forty-five images of 40 patients were analysed qualitatively and quantitatively. Two verifications per week were planned, and the action level for correction was 10 mm. Head and neck images were also displayed in 'cinema' presentation for internal movements analysis. From 1994 to 1995 (PER 2), off-line procedure (OLP) based upon early correction of the systematic error and the rules calculated from our previous experience were tested for checking the brain, head and neck (LOC 1: 396 images) and many of the pelvic irradiations (LOC 2: 260 images). A double-exposure procedure and/or movie loop presentation was reserved for other patients. During the last period (PER 3: 1996-1997), the OLP procedure was routinely performed in 54 patients (images: 321 LOC 1, 680 LOC 2).

LOC 1: deviations of < 3 mm increased from 75.5% during PER 1 to 81% during PER 2 to 83% during PER3. Conversely, deviations of 3-5 mm dropped from 19.5 to 13%, while deviations of more than 5 mm remained stable, around 5%. The actual standard error of the mean deviation observed was 2 mm. LOC 2: deviations of < 5 mm were observed in 81% of the cases during PER 1 and in 91% during PER 3 (89.5% in PER 2). These good results led to a decrease in deviation of 5 to 7 mm (11 to 6%) and also to a significant drop in deviations of more than 7 mm, 8 to 3% respectively. The actual precision obtained was 2.5 mm +/- 3 mm SD.

The OLP based upon the early correction of the systematic error led to a significant increase of setup accuracy of patients irradiated for the brain, head and neck, and especially for pelvic lesions.