David F. Lewis 1, Andre Micke 1, Xiang Yu 1 and Maria F. Chan 2 1. Advanced Materials Group, Ashland Inc., Wayne, NJ; 2. Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, Basking Ridge, NJ 1.0 Purpose: This work explores post-exposure changes in GafChromic EBT3 film with the objective of eliminating the usual long post- exposure wait between exposure and scanning. 2.0 Materials and Methods: Film strips 3.81 x 20.3 cm 2 were obtained from a production lot of EBT3. The initial work was done by exposing films to five doses between 30 and 480 cGy using 160 KV x-rays filtered through 1 mm of Cu. Before exposure a 3.81 x 5 cm 2 area was marked on each sample to facilitate alignment with a 5 x 5 cm 2 target centered on a uniform ( ±1%) radiation field. The exposures were completed in <10 min. Scanning was done at 72 dpi in 48-bit transmission mode on an Epson 10000XL RGB scanner. Color corrections were disabled. At various times-after-exposure, the films, plus an unexposed film from the same lot, were scanned with their 3.81 cm side parallel to the scan direction. Time-after-exposure is the elapsed time between exposure of the last film and scanning. Other images were collected by digitizing the films on different 10000XL and Epson V700 scanners using the orientation described above. Irradiation of calibration films at the center of a 10x10cm 2 field was also done using the 6 MV photon beam of a Varian Trilogy Linac. Films were irradiated at depth of 5 cm and 100 cm SAD in a polystyrene phantom. Irradiations of an IMRT treatment plan were done at 6MV on the Trilogy Linac using the same phantom and a 20.3 x 25.4 cm 2 sheet of film. Immediately after the IMRT plan exposure a reference filmstrip from the same production lot was placed in the phantom at the center of a 10x10 cm 2 field and exposed to a known dose. The reference and IMRT films were irradiated within a 5-minute time window. 3. Measurement: Measurements and analyses of images were done with FilmQAPro 3.0 software (Ashland Inc., Wayne, NJ) using the triple- channel dosimetry method 1. For calibration films, the RGB measurements were taken in 3.5 x 4.5 cm 2 areas at the centers of the exposures. Dose-response data sets were fitted to the rational function X(D) = A + B/(C-D) where X(D) is the response at dose D and A, B and C are constants. Dose maps were obtained by inverting the function and converting images to dose space. Measured dose maps were compared to the treatment plans and to one another using gamma analysis with test criteria of 2% at 2mm. 4. Results: 4.1 Response Equivalence and Post-exposure Changes: Figure 1 shows dose-response data for the red color channel measured from images at times-after-exposure between 65 min. and 80 hr. The inset shows the measured response of the unexposed film varied from time-to-time. From extensive experience we have found the inter-scan variability of the 10000XL scanner is about ±0.075%. Absent a pattern to the responses of our unexposed film we attribute the observed differences to inter-scan variability. To factor out the variability we normalized the responses of each film in an image to the response of the unexposed film in that same image. Figure 2 shows the results. The shapes of the curves look similar and are related by a scaling factor. Consider the ratio of the net responses NR T1/ NR T2 for the highest dose (480 cGy) at any two times, T 1 and T 2, after exposure where net response is the difference in films response at 480 cGy and 0 cGy. When the scaling factor NR T1/ NR T2 was applied to the net responses of all other doses at time-after-exposure T 2, the responses at the two times-after- exposure were equivalent, i.e. calibration curves at any two times-after-exposure are related by a scaling factor dependent on two measured responses. Responses in the green and blue channels behaved similarly. 4.2 Response Equivalence - Scanner model: When images of the calibration films were acquired in a variety of ways the measurements revealed similar equivalences to those previously described. Thus, the net, normalized film responses on one scanner were found to be related to the net, normalized responses on a second scanner and the data sets could be made to correspond by linear scaling using the ratio of the net response values at 480 cGy on the two scanners. We found this true for all color channels. The example in Figure 3 shows the equivalence of the normalized and scaled red-channel response data from films scanned on four 10000XL scanners. We extended the measurements to different scanner models and found the same pattern of equivalence between data acquired on Epson 700 scanners and 10000XL scanners. 4.3 Response Equivalence - Photon energy, temperature and orientation: We found the same pattern of equivalence in three other cases: between response data acquired from films exposed at different photon energies (6 MV and 160kVp); between response data acquired from films scanned at 10°C and 21°C; and between response data acquired from films scanned in different orientations i.e cm side of film parallel to thescan direction (landscape orientation) and perpendicular to the scan direction (portrait orientation). In this last case the correspondence of the data for the two orientations after scaling is illustrated in Figure 4. The scaled response values in the two orientations are equivalent to ±0.3% in all color channels. 5. New Protocol an Original Calibration Curve: The results indicated that dose-response data for a film production lot could be fit to a set of related rational functions leading to the description of a generic calibration curve. A simplified protocol was established where dose-response data for a specific scanner, scanning conditions (time-after exposure, temperature, orientation) and exposure source could be derived from a generic calibration curve using one film exposed to a known dose and an unexposed film to adapt the generic curve to the specific case. The normalized response X of the system with respect to dose can be correlated using rational functions of the form, X(D) = A + B/(C-D) or X(D) = A + BD/(D+C) where A, B, C are parameters that can be fitted to calibration data using least square approach. For measured data (n i, D i ) with =1(1)I,n normalized system response and D dose, the equation ∑ i (N(D i ) - n i ) 2 → min A,B,C is minimized to determine the calibration parameters A, B, C. A specific calibration can be derived from the normalized system response N using the rescaling relation X(D)= α + β N(D) where X is the response in one of the color channels R, G or B. The two parameters α and β can be calculated as α = (N 1 X 2 – N 1 X 1 )/(N 1 -N 2 ) and β = (X 1 – X 2 )/(N 1 -N 2 ) if two data points (X i, D i ), i=1,2 are available using N i = N(D i ). This new protocol promises to ease post-exposure waiting restrictions to just a few minutes before scanning. It requires the application film to be scanned with two reference films from the same production lot, one reference film unexposed and the other exposed to a dose similar to the highest dose on the application film. To minimize the post-exposure wait the films should be exposed within a narrow time window (Section 7). If the time window is t, the minimum time between exposure and scanning should be 4t to keep dose error <0.5%. Testing the Protocol-1: The protocol was tested using the same calibration films described in Section 4. Measurements two hours after exposure were fit to the rational function X(D) = a + b/(c-D). For each image at different times-after-exposure measurements of the unexposed film and the film exposed to the highest dose were used to apply two-point re-scaling to adapt the original function for the specific scan. The adapted functions were inverted for to calculate dose maps. The data in Table 1 shows the values for each of the films are substantially independent of time-after-exposure providing a convincing demonstration that post-exposure changes have been compensated by using the reference films to re-scale the calibration function. Without re-scaling calculated doses increase with increasing time-after-exposure reflecting the well known post exposure changes of radiochromic film. Testing the Protocol-2: The protocol was also evaluated by comparing dose maps calculated from EBT3 films exposed with an IMRT treatment field and scanned at various times from 30 min. to 115 hrs after exposure. A set of calibration films was made by exposing 10 x 10 cm 2 fields with a 6MV beam within a 10 minute time window. Together with an unexposed film from the same lot they were scanned 2 hrs later The dose-response data was fitted as before to provide a generic calibration function. Using a 6MV beam, a film was exposed to an IMRT field and within 5 minutes a reference film was exposed to 100 cGy. At various intervals the IMRT film and exposed and unexposed reference films were scanned together. Using measurements from the reference films in each image, the generic calibration function was re-scaled and adapted for the specific scan. The IMRT images were converted to dose maps and compared to the treatment plan using gamma evaluation with test criteria of 2% at 2mm. In all cases, from 30 min. to 115 hr after exposure, the values had 97.2±1% agreement with plan (Table 2). In contrast, dose maps calculated without re-scaling had poorer agreement with plan and sometimes much poorer. For instance the map 30 min. after exposure had doses about 1-2% below plan and 93% gamma agreement, while the map 115 hr after exposure had doses 5% above plan and a gamma agreement <60%. 7. Post-exposure change: Measurements show the post-exposure changes are linear with log(time-after-exposure). This becomes a source of error for films scanned at different times-after-exposure. The new protocol requires scanning application and reference films together. But they can’t be exposed simultaneously so the times between their exposure and their measurement will be different. If films are exposed in a narrow time window, error from exposure timing diminishes rapidly as the time between exposure and scanning increases. From the data in Figure 5, we find that a scan delay of 30 minutes with a 5-minute exposure time difference leads to dose error of ~ 0.3%. Changing the scan delay by a factor of two doubles or halves the error. As a rule of thumb, if the post-exposure delay before scanning is four times longer than the exposure time window, dose errors will be <0.5%, e.g. for a 5-minute time window scanning could be done with 20 minutes delay, or any time later. 8. Conclusion : We have used a simplified protocol to measure doses on EBT3 film by scanning an application film, an exposed reference film and an unexposed film together to provide measurements to adapt a generic calibration curve for a specific scan. The simplification and timesaving provide a practical solution for using radiochromic film for fast therapy plan QA without sacrificing spatial resolution for convenience. By combining measurement and calibration in one scan the protocol promises to eliminate the effects of inter-scan variability. The consistency of results after re-scaling the generic calibration function to compensate for post-exposure change shows the value of the new protocol. The work also raises the possibility of publishing a generic calibration curve for every production lot of EBT3. Figure 1Figure 2 Figure 3 Figure 4 Figure 5 Table 1Table 2 A new protocol for fast radiochromic film dosimetry