Improvement of dose calculation in radiation therapy due to metal artifact correction using the augmented likelihood image reconstruction

Abstract Background Metal artifacts caused by high‐density implants lead to incorrectly reconstructed Hounsfield units in computed tomography images. This can result in a loss of accuracy in dose calculation in radiation therapy. This study investigates the potential of the metal artifact reduction algorithms, Augmented Likelihood Image Reconstruction and linear interpolation, in improving dose calculation in the presence of metal artifacts. Materials and Methods In order to simulate a pelvis with a double‐sided total endoprosthesis, a polymethylmethacrylate phantom was equipped with two steel bars. Artifacts were reduced by applying the Augmented Likelihood Image Reconstruction, a linear interpolation, and a manual correction approach. Using the treatment planning system Eclipse™, identical planning target volumes for an idealized prostate as well as structures for bladder and rectum were defined in corrected and noncorrected images. Volumetric modulated arc therapy plans have been created with double arc rotations with and without avoidance sectors that mask out the prosthesis. The irradiation plans were analyzed for variations in the dose distribution and their homogeneity. Dosimetric measurements were performed using isocentric positioned ionization chambers. Results Irradiation plans based on images containing artifacts lead to a dose error in the isocenter of up to 8.4%. Corrections with the Augmented Likelihood Image Reconstruction reduce this dose error to 2.7%, corrections with linear interpolation to 3.2%, and manual artifact correction to 4.1%. When applying artifact correction, the dose homogeneity was slightly improved for all investigated methods. Furthermore, the calculated mean doses are higher for rectum and bladder if avoidance sectors are applied. Conclusion Streaking artifacts cause an imprecise dose calculation within irradiation plans. Using a metal artifact correction algorithm, the planning accuracy can be significantly improved. Best results were accomplished using the Augmented Likelihood Image Reconstruction algorithm.

image data strongly depends on the mass number of the material used in the implant. In this way, they showed that the error in dose calculation was up to 23.56% for the planning target volume (PTV) region in the case of cerrobend. However, the error could be reduced to 0.11% by a MAR algorithm. In conclusion, it is stated that the dose calculation is more accurate when the reconstructed density information is less distorted. 7 In this context, the ability to retrieve correct HU values should be the main evaluation criterion for a MAR algorithm. [6][7][8][9][10][11] In a previous study, we investigated the MAR algorithm Augmented Likelihood Image Reconstruction (ALIR) regarding its ability to retrieve attenuation coefficients in the presence of streaking artifacts. 12 It was shown that ALIR is able to correct distorted HU values to a high accuracy. Furthermore, structures that were not perceptible due to streaking artifacts were reconstructed accurately using ALIR.
In the current study, we investigate the impact of corrected HU values on the dose calculation after applying different MAR algorithms. A PMMA phantom that imitates a patient with double-sided total endoprosthesis (TEP) was created. Two different case scenarios were investigated. In the first scenario, we calculate volumetric modulated arc therapy (VMAT) plans with double arc rotation, without taking into account possible influence of the steel inserts on the dose distribution during the planning process. In the second scenario, dose effects near the implant and the shadowing characteristic of the steel inserts are minimized using avoidance sectors, which avoid direct irradiation through the implants. 13 In order to investigate the dosimetric effects, plans based on metal artifact corrected and original reconstructed images are compared. Furthermore, a measurement of the applied dosage within the isocenter of the phantom based on ionization chambers is used as a reference.

2.A | Phantom
In order to simulate a pelvis with a double-sided endoprosthesis for the hip, a PMMA phantom (

2.B | CT imaging
For the acquisition of images, a 40-slice CT scanner type Biograph mCT (Siemens AG, Erlangen, Germany) was used. The scans were acquired sequentially with 120 kVp, a field of view of 500 mm, and a slice thickness of 4 mm. The images were reconstructed using the filtered backprojection with a ramp filter (FBP) and the iterative algorithm ALIR (see next paragraph).
In addition to a dataset with the two metal rods, an image dataset without steel rods was acquired in order to have an artifact-free image set available. This was used for the contouring of the ionization chamber.
The Hounsfield scale in clinical use is usually limited to a maximal value of 3071 HU. This is sufficient to represent the organs and bones according to their specific densities. However, in order to cover the steel inserts, the used Hounsfield scale was extended to 13,500 HU. This corresponds to a material density of q (Steel) = 7.9 g/cm³.

2.C | Artifact correction
The reduction in metal artifacts is performed utilizing three different approaches. The linear interpolation approach (LI) represents a simple and easy applicable reference method for the reduction in artifacts. Here, for every angle, projections that pass through the metal object are replaced by a linear interpolation between uncorrupted projection values. 14 Based on the resulting raw data, the reconstruction of the image is performed using the FBP.
As a second approach, the recently proposed ALIR algorithm is used. 15 The method is based on an iterative scheme and integrates two different ideas in order to reduce streaking artifacts. In a first step, the algorithm formulates the reconstruction of an image as an optimization problem based on the negative log-likelihood function for transmission CT. 16,17 The optimization process is complemented by constraints that force the reconstruction to assign certain attenuation values in the region of the metal implant. In the present case,

| 229
The main dose planning objective is to achieve a PTV dose coverage according to the ICRU Report 50 with 107%/95% of the prescribed dose within the PTV. 1 The VMAT is chosen, as it is the routine method for this entity. In comparison to intensity-modulated radiotherapy, its ability in sparing organs at risk is more efficient, the account of monitor units is lower, and the target dose distribution is more homogeneous. [18][19][20] VMAT plans were created in such a way that the 95% isodose covers the PTV. The absorbing effect of the steel rods is taken into account by two different approaches. In the first approach, the issue with the dense material of the rods was left entirely to the optimization tool of the planning software. Here, the linear accelerator rotates 360°with an activated beam around the phantom. In the second approach, avoidance sectors were selected for the rotational angles at which the PTV was covered by the absorbing steel rods [250°to 290°and 70°to 110°, see Figs. 3(a) and 3(b)]. Within these sectors, the object was not irradiated (Figs. 4 a & 4b).
The dose optimization was performed with the VMAT optimization module of Eclipse TM , whereby the optimization process was performed in two steps. First, the intermediate dose calculation was performed using the pencil beam algorithm (PBC 10028) 21 and second, the final dose calculation was performed using the anisotropic analytical algorithm (AAA 10.0.28). 22 The used dose optimization objectives are listed in Table 1. The upper and lower objectives describe the percentage, which should not be exceeded (upper objective) or not fall below (lower objective). A preset priority value weights the importance of the volume in question during the optimization process.

3.A | Dose homogeneity index HI
For determination of the homogeneity index, we used the following where the parameter D x% represents the absorbed dose received by x% of the PTV. An HI of close to zero indicates that the absorbed dose distribution is almost homogenous. 23,24  bladder, the best reduction in mean dose is achieved with LI (À2.7%) and ALIR (À2.3%). Manual correction leads to a reduction of À1.7%.

3.B | Doses in bladder and rectum
For the rectum, the most reduction in the mean dose is achieved by ALIR (À3.1%) and LI (À2.7%). Manual correction leads to a reduction of À1.8%. (Table 2a & b) For the planning scenario of two rotations with avoidance sectors, we observe a dose shift after applying artifact correction as well. However, the manifestation is not as distinctive as in the scenario without avoidance sectors. The bladder mean dose shifts for ALIR by À0.4%, for LI by À0.5%, and for manual by À1.5% after correction. The mean doses at the rectum are reduced for ALIR by À1.9%, for LI by À0.8%, and for manual by À1.3%.

CONF LICTS OF INTEREST
All authors declare that there are no conflicts of interest.