Measurements of beam performance during IMRT

(Abstract for the DGMP-Meeting in Passau, Germany, Sept. 15 - 18, 1999
The DGMP is the German Association for Medical Physics)

H. Krauss, S. Mitterer, A.U. Schratter-Sehn
Institute of Radiooncology, Kaiser-Franz-Josef Spital (KFJ), Kundratstraße 3, A-1100 Vienna



Before Intensity Modulated Radiotherapy (IMRT) can be introduced into clinical routine, the physicist has to verify, among many other things, the beam performance during such dynamic treatment techniques. During treatments of intensity modulated beams with Dynamic Multileaf Collimators (DMLC, sliding window technique), the servo control of the Linac may, under certain circumstances, lead to sharp variations of dose rate during the whole treatment. Therefore, beam characteristics (mainly beam symmetry) have to be verified not only during the startup phase of the Linac, but for the entire treatment.

Material and Methods

With our two linear accelerators (Varian Clinac 2300C/D) it is possible to deliver DMLC treatments in Clinical mode, with full verification by the VARiS verification system. After treatment planning with the Cadplan extension Helios ("inverse planning") all treatment parameters including the DMLC file are exported to the VARiS system and can be treated on both machines.

A static MLC file is a text file that contains a single shape, whereas the dynamic MLC file consists of a sequence of up to 320 shapes, each with an associated dose fraction (values between 0.000 and 1.000). During execution, the control system enforces a linear path of all leaves within a segment, with different leaves generally travelling at different speeds. Figure 1 shows snapshots of four out of 200 segments of a prostate field, planned with Cadplan/Helios, together with the resulting fluency distribution. The four segments have the corresponding dose fractions 0.2513, 0.4523, 0.5578 and 0.7437, respectively.

As long as the required MLC motions do not result in any of the leaves exceeding their maximum speed, the operator-selected dose rate is used (in the range of 100 – 600 MU/min). Otherwise, the controller drops beam pulses, which leads to a sharp fluctuation of dose rate. Figure 2 shows the central axis dose rate for a DMLC test treatment, where leaves move faster and faster until, after 1.2 seconds of beam, the maximum leaf speed is reached. The array was positioned inplane at SSD = 70 cm and 10 cm depth in water. Figure 3 shows the relative dose rate distribution in 3D, normalized to 100% at dose rate maximum.

The sharp fluctuations pose no problem a priori, as long as the beam characteristics are acceptable under these conditions. The influence on total treatment time is moderate: the measured treatment time of the field shown in Fig. 1 was 15 seconds (84MU @ 600 MU/min). The time to position the patient on the couch would still be considerable longer than the pure beam-on time. Therefore a complete IMRT/DMLC-treatment can be delivered in about the same time as a "conventional" conformal treatment with the same number of fields.

With our fast measuring system (PTW Linear Array LA48, fast measurement option) it is possible to acquire full beam profiles within 2 milliseconds and a repetition time of 10 ms [1]. With this system, the time dependence of dose rate and symmetry during DMLC treatments were studied.

Results and Discussion

The measuring system has a high time resolution. Within 2.5 s, 250 profiles can be measured and stored. Analysis of symmetry was done with a commercial spread sheet program, using the IEC 976 definition of symmetry. But one has to be careful with the analysis, because on Linacs with a short beam rise time, an effect of the measuring system can be observed that slightly distorts the results. The curve shown in Fig. 2 and 3 has a maximum change of dose rate ("dose acceleration") of 4 Gy/s2. Since the 47 ionisation chambers of the LA48 are read out sequentially within 2 ms, this leads to a delay of 43 microseconds per channel and can lead to a maximum error of the dose rate value of 5% for channel 47, if the full field is used. By rotating the array by 180°, it can be shown that the effect originates from the measuring system, not from the Linac. To find the "true" symmetry, the raw data has to be corrected numerically. Corrections of first order were made and showed that the symmetry is in the range of the "static" values (see poster for details).


[1] St. Rassow, J. Moog, M. Buchgeister, E. Schüle, F. Nüsslin, Profilmessungen in 10 ms-Intervallen mit einem Linear-Array, in: Medizinische Physik 1998, Hrsg. L. Voigtmann und P. Geyer, S. 99f