In patient dose reconstruction using a cine acquisition for dynamic arc radiation therapy

Abstract

An amorphous silicon (a-Si) electronic portal imaging device (EPID) was implemented to perform transit in vivo dosimetry for dynamic conformal arc therapy (DCAT). A set of images was acquired for each arc irradiation using the EPID cine acquisition mode, that supplies a frame acquisition rate of one image every 1.66 s, with a monitor unit rate equal to 100 UM/min. In these conditions good signal stability, ±1% (2SD) evaluated during 3 months, signal reproducibility within ±0.8% (2SD) and linearity with dose and dose rate within ±1% (2SD) were obtained. The transit signal, S _t, due to the transmitted radiotherapy beam below a solid phantom, measured by the EPID cine acquisition mode was used to determine, (1) a set of correlation functions, F(w, L), defined as the ratio between S _t and the dose at half thickness, D _m, measured in solid water phantoms of different thicknesses, w and with square fields of side L, (2) a set of factors, f(d, L), that take into account the different x-ray scatter contribution from the phantom to the S _t signal as a function of the variation, d, of the air gap between the phantom and the EPID. The reconstruction of the isocenter dose, D _iso, for DCAT was obtained convolving the transit signal values, obtained at different gantry angles, with the respective reconstruction factors determined by a house-made software. The method was applied to a first patient and the results show that the reconstructed D _iso values can be obtained with an accuracy within ±5%. In conclusion, it was assessed that an a-Si EPID with the cine acquisition mode is suitable to perform transit in vivo dosimetry for the DCAT therapy.

Appendix: Determination of C α

The CT scans of the patients, generally 3 mm thick, have been uploaded in the MATrix LABoratory (MATLAB version 7.1, The MathWorks Inc., Natick, MA, USA) and processed by a home-made software CIRIO. In particular, the CT scan, containing the isocenter, is initially contoured.

The input data of the CIRIO software for every arc are

• the gantry initial angle α _i and the final angle α _f of the arc θ;

• the isocenter position;

• the mean equivalent square field of the arc;

• the couch angle β as respect to the gun-target direction;

• the x-ray beam quality.

If the couch angle β = 0, only the CT slice that contains the isocenter is processed by CIRIO software. When the couch angle is not equal to 0 (for non-coplanar arcs) a number, n, of CT slices have to be processed

$$ n = \frac{{{\text{FOV}} \cdot {\text{sen}}\beta }}{\tau } $$

(1A)

where the FOV is the field of view diameter of the CT scanner (equal to 48.8 cm) and τ is the thickness of the patient’s CT slice. Using the n CT slices a non-coplanar image along the β direction is obtained interpolating its values between the two nearest CT slices that contain each given section of the transversal plane along the β direction, then the non-coplanar image is processed by the CIRIO software.

For each of the n gantry angles, α, the program supplies

• the geometric patient’s thickness, z, the distance, d, between z/2 and the isocenter and the depth of the isocenter point z _iso.

• the mean relative physical densities along the depth z, z/2, z _iso. These data are determined using the stechiometric calibration of the Hounsfield numbers. Indeed the pixel values, along a rectangular strip coincident with the beam central axis (5 mm wide), were converted firstly in mean electronic density and successively in mean physical density.

• the water-EPLs w, w/2 w _iso determined multiplying the geometrical thicknesses z, z/2 and z _iso, by their mean relative physical densities.

By these data, the function F(w,L), the factors f(d,L) and the TMR can be selected to obtain the dose reconstruction factors C _α and the D _iso values. The execution of this phase takes no more than 5 min.