Enhanced Accuracy of IMRT Photon Fluence Profile Surveillance:Iterative Resolution Correction

Enhanced accuracy of IMRT photon fluence profile surveillance:iterative resolution correction of the DAVID chamber

H.K.Looe1, 2, D. Harder3, A. Rühmann1, 2, K. C. Willborn2 and B. Poppe1, 2

1 Working Group Medical Radiation Physics, Carl von Ossietzky University Oldenburg, Germany

2 Pius-Hospital, Clinic for Radiotherapy and Oncology, Oldenburg, Germany

3 Prof. em., Medical Physics and Biophysics, Georg August University, Göttingen, Germany

Introduction

The development of advanced treatment delivery techniques such asIMRT and rotational therapy has called for comprehensive qualityassurance procedures in the clinical routine. Inaddition to the pretreatment plan dosimetric verification, the permanent compliance of thetreatment delivery with the planned one has to be guaranteed for thewhole course of the therapy which may be extended over several weeks.The DAVID system (PTW-Freiburg, Germany) is an in-vivo dosimetrythroughout all fractions [1]. The system consists of a translucent multiwireionization chamber, permanently placed in the accessory holder of thetreatment head, directly downstream of the MLC and the reticle. Each wireis exactly adjusted along the midline of its associated leaf pair, therebygenerating a signal proportional to the leaf pair aperture. This system hasbeen successfully implemented in the clinical routine to monitor all IMRTtreatment deliveries in our clinic [2].

The reading of each single channel of the DAVID chamber is proportionalto the integral of the ionisation density along the detection wire in question.The signal recorded by each wire is mainly owed to the secondaryelectrons originating from the front and back plates irradiated by photonspassing the leaf pair opening associated with this wire. However,secondary electrons with an origin in other parts of the front and back platewithin the irradiated field, scattered within the plates and moving laterally inthe air gap, also contribute to the signal of this wire. Thus, the lateraltransport of secondary electrons influences the lateral resolution functionof the DAVID chamber. This effect can be characterized by measuring theresponses of all detection wires when only the leaf pair associated with asingle wire is opened. Once the lateral response function is known, it ispossible to calculate the true photon fluence profile from the measuredsignal profile by a deconvolution algorithm.

Material and methods

Two versions of DAVID chambers DAVID58 and DAVID160, designed eitherfor the Siemens MLC which has 29 leaf pairs (MLC58, Siemens, Concord,CA, USA) or for that with 80 leaf pairs (MLC160, Siemens, Concord, CA,USA) respectively, have been investigated. Figure 1 shows the lateralresponse functions fξ(x) measured with DAVID58 and DAVID160, both at twophoton energies. Here ξ is the running number of the wire whoseassociated leaf pair is opened, and x is the running number of all signalwires whose signals are recorded. The measured signal profile S(x) of theDAVID system in the x-direction is regarded as the result of theconvolution of the true photon fluence profile P(x) with the lateral responsefunction fξ(x). For each given detection wire, this function was measuredwhile only the MLC leaf pair directly above this wire was opened, and wasstored as a look-up table labeled by number ξ. The measured lateralresponse functions fξ(x) were then used in an iteration algorithm tocalculate the true photon fluence profiles P(x) from the measured signalprofiles S(x). The iteration process consists in a sequence ofapproximations Pn(x) which quickly converges towards the desired trueP(x). Each approximation Pn(x) is numerically convolved with fξ(x) and fromthe comparison of the result with S(x), the next approximation Pn+1(x) isderived.

Results

Figure 2 illustrates the reconstruction of the true fluence profiles P(x) fromthe blurred signal profiles S(x) of an IMRT segments with the DAVID160chamber. After the deconvolution, the fine structure of the fluence profilesis restored. The changes in the measured signal profiles and in thedeconvolved fluence profiles due to artifially introduced leaflet errors areindicated by the arrows. The magnitudes of the measured signaldeviations at the detection wires directly associated with the leaflet errors.

and at the next adjacent wires are at the limit of visibility, whereas after thedeconvolution the associated detection wires clearly show the leaflet errors,and the adjacent wires exhibit no sig

Figure 1: Lateral response functions, at two photon energies, of the DAVID58 chamber (upper
panel) and the DAVID160 chamber (lower panel) for three selected leaf pairs at different
positions (upper panel: ξ= 9, 19 and 29; lower panel: ξ= 16, 40 and 65)

F i g u r e 2: Reconstruction of the true photon fluence profiles and enhancement of the error detection efficiency of the DAVID160 chamber by iterative deconvolution. Artificial 2 mm leaflet errors (e) were introduced into channels 36, 41 and 45 (see arrows). Upper panel: Measured signal profiles, S(x) and Se(x). Lower panel: Deconvolved photon fluence profiles, P(x) and Pe(x)

Literature

[1] Poppe B, Thieke C, Beyer D, Kollhoff R, Djouguela A, Rühmann A, Willborn KC and Harder D. DAVID-a translucent multi-wire transmission ionisation chamber for in vivo verification of IMRT and conformal irradiation techniques. Phys Med Biol 2006; 51:1237-48.

[2] Poppe B, Looe H K, Chofor N, Rühmann A, Harder D and Willborn K. Clinical Performance of a Transmission Detector Array for the Permanent Supervision of IMRT Deliveries. Radiother Oncol 2010; 95:158-65