Home About us Editorial board Ahead of print Current issue Search Archives Submit article Instructions Subscribe Contacts Login 

 Table of Contents  
ORIGINAL ARTICLE
Year : 2020  |  Volume : 16  |  Issue : 4  |  Page : 878-883

Improvement of metallic artifacts in computed tomography in the absence of artifact reduction algorithms for spinal treatment planning applications


1 Department of Medical Physics, Guilan University of Medical Sciences, Rasht, Iran
2 Department of Radiation Oncology, St. Jude Children's Research Hospital Memphis, TN, USA
3 Department of Oncology, Milad Hospital, Isfahan, Iran
4 Department of Medical Physics, Isfahan University of Medical Sciences, Isfahan, Iran

Date of Submission28-Dec-2016
Date of Acceptance25-Feb-2018
Date of Web Publication24-Oct-2018

Correspondence Address:
Parvaneh Shokrani
Department of Medical Physics and Engineering, School of Medicine, Isfahan University of Medical Sciences, Isfahan
Iran
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/jcrt.JCRT_1446_16

Rights and Permissions
 > Abstract 


Aim of Study: The goal of this research was to investigate if application of optimized imaging parameters, recommended in literature, would be effective in producing the image quality required for treatment planning of spinal radiation fields with metallic implants.
Materials and Methods: CT images from an anthropomorphic torso phantom with and without spinal implants were acquired using different imaging protocols: raising kVp and mAs, reducing the pitch and applying an extended CT scale (ECTS) technique. Profiles of CT number (CT#) were produced using DICOM data of each image. The effect of artifact on dose calculation accuracy was investigated using the image data in the absence of implant as a reference and the recommended electron density tolerance levels (Δρe).
Results: Raising the kVp was the only method that produced improvement to some degree in CT# in artifact regions. Application of ECTS improved CT# values only for metal.
Conclusions: Although raising the kVp was effective in reducing metallic artifact, the significance of this effect on Δρe values in corrected images depends on the required tolerance for treatment planning dose calculation accuracy. ECTS method was only successful in correcting the CT number range in the metal. Although, application of ECTS method did not have any effect on artifact regions, its use is necessary in order to improve delineation of metal and accuracy of attenuation calculations in metal, provided that the treatment planning system can use an extended CT# calibration curve. Also, for Monte Carlo calculations using patient's images, ECTS-post-processed-CT images improve dose calculation accuracy for impure metals.

Keywords: Computed tomography image artifact, extended computed tomography scale, spinal metallic implant, treatment planning


How to cite this article:
Alinejad M, Pourmoghaddas A, Amouheidari A, Shokrani P. Improvement of metallic artifacts in computed tomography in the absence of artifact reduction algorithms for spinal treatment planning applications. J Can Res Ther 2020;16:878-83

How to cite this URL:
Alinejad M, Pourmoghaddas A, Amouheidari A, Shokrani P. Improvement of metallic artifacts in computed tomography in the absence of artifact reduction algorithms for spinal treatment planning applications. J Can Res Ther [serial online] 2020 [cited 2020 Sep 30];16:878-83. Available from: http://www.cancerjournal.net/text.asp?2020/16/4/878/243485




 > Introduction Top


Accurate dose delivery in three-dimensional (3D) conformal radiotherapy is a complex process that begins with creating 3D digital data sets of patient anatomy used in treatment planning. Hounsfield unit (HU) values for different materials are calculated from computed tomography (CT) images and are used for organ delineation and dose calculations. Image artifacts created by the presence of metallic implants can cause changes in the HU values. These artifacts may prevent the accurate evaluation of position, size, and shape of regions of interest near the implants and jeopardize the accuracy of dose calculation in artifact areas.[1],[2],[3],[4],[5] Therefore, it is necessary to minimize and/or correct CT images for metal artifacts. From diagnostic point of view, the followings are recommended to minimize metal artifact: using metals with lower attenuation,[6] application of optimal imaging parameters, i.e., high kVp and mAs settings, narrow collimation, thin section imaging protocols[1],[7],[8] and postprocessing techniques such as extended CT scale (ECTS).[9] In the ECTS technique, the maximum HU (4096) is expanded to 40,960 to include values for metallic implants (8000–20,000 HU).[10]

Artifact reduction algorithms have also been used to improve both the diagnostic and dose calculation accuracy. It has been shown that application of artifact correction algorithms resulted in higher level of accuracy in dose calculated by Monte Carlo simulations using the EGSnrc/DOSXYZnrc CTCREATE code[11] with extended material calibration in patients with hip prostheses.[2] However, none of the correction artifact algorithms are clinically available for everyday use. Furthermore, to the best of our knowledge, the effect of metallic spinal artifacts for treatment planning purposes has not been evaluated. Compared to hip prostheses, spinal implants are smaller and are in proximity of each other, the planning target volume and the spinal cord. Furthermore, lower energies are used for treatment of lesions in and around the vertebral column. It has been demonstrated that the tolerance for electron density quality assurance to maintain a certain level of dose accuracy (2%) is a function of tissue type, thickness, and photon beam energy. For lower energies, a relative electron density tolerance of 0.1 results in a planning dose error of 2%, for typical tissue thicknesses in and around the vertebral column.[12]

The aim was to investigate if application of optimized imaging parameters by itself is effective in producing the image quality required for treatment planning of spinal radiation fields with metallic implants.


 > Materials and Methods Top


CT images from an anthropomorphic torso phantom, with and without implants, were acquired using different imaging protocols and an ECTS to the reconstructed images. The CT numbers in the absence of the metallic implants were used as a reference to study the change in CT number in each pixel of image with metallic artifact and corrected images.

Phantom

The anthropomorphic phantom is made of the following materials: Plexiglas as soft tissue (ρ = 1.18 g/cm3), Teflon as bone (ρ = 2.2 g/cm3), and Polyethylene as spinal cord (ρ = 0.9 g/cm3). An insert was added to the phantom to represent the spinal column containing the implants. Two pure titanium (Ti) rods (Synthes company, Switzerland, ρ = 4.5 g/cm3, Ø = 6 mm diameter) were used as implants. Three holes were machined in the Teflon cylinder (Ø = 30 mm) where the Polyethylene rod (Ø =10 mm) and Ti rods (Ø = 6 mm) were inserted [Figure 1]a.
Figure 1: The anthropomorphic torso phantom used in this study, (a) a photograph of phantom containing spinal implants, (b) phantom computed tomography image without metallic implant (titanium rods) shows no streaking artifact, (c) phantom implanted with metallic rods leads to streaking artifacts. Arrows in b and c highlight the axes used to generate computed tomography number profiles in Figures 3-6

Click here to view


Imaging protocols

A clinical CT scanner (Siemens, Somatom Sensation 40/64) was used to obtain images of the phantom. Phantom images were acquired using the following 8 scan protocols: Protocol A: 120 kVp, automatic dose control option (tube current modulation [TCM]), 200 mAs, pitch = 1.4, rotation time = 0.5s, slice thickness = 3 mm, and B30 medium smooth filtration (this is the protocol routinely used in our department for scans of spine of radiotherapy patients), Protocol B: 140 kVp, other parameters similar to Protocol A, Protocol C: TCM turned off, 250 mAs, other parameters similar to Protocol A, Protocol D: pitch = 0.45, other parameters similar to protocol A, Protocol E: ECTS option applied, other parameters similar to Proto col A, Protocol F: 140 kVp, ECTS, other parameters similar to Protocol A, Protocol G: TCM turned off, 250 mAs, ECTS option, other parameters similar to Protocol A, and Protocol H: pitch = 0.45, ECTS, other parameters similar to Protocol A. For each scan protocol, the reconstructed DICOM data were exported offline for image analysis.

Computed tomography number calibration

[Figure 2] shows the calibration curve for CT number versus electron density relative to water scanned using different imaging protocols and normal and extended CT number ranges. For values higher than 3000 HU, CT calibration curves with standard number and ECTS ranges have different gradients. In this study, for different phantom materials used, the range of CT numbers was determined to be −1024–3071 and −10,240–30,710 using protocols A, B, C, D and E, F, G, H, respectively.
Figure 2: Calibration curve for computed tomography number versus electron density relative to water for normal range using Protocols A (120 kVp, 200 mAs, 0.6 mm collimation, 3 mm reconstructed section thickness, kernel value [B30f]), B (140 kVp, other parameters similar to Protocol A), C (tube current modulation turned off, 250 mAs, other parameters similar to Protocol A) and D (pitch = 0.45, other parameters similar to Protocol A) and extended computed tomography ranges scanned using Protocol E to H, i.e., Protocols A to D and extended computed tomography scale postprocessing technique

Click here to view


Computed tomography number profiles for different scanning techniques.

Profiles of CT number were used to demonstrate the effect of imaging parameters on the metallic artifact. For each protocol, DICOM data with voxel size of (x, y, z) 0.087 cm × 0.087 cm × 0.3 cm of one image of a fixed position (one slice) in phantom was used. For each image, CT number profiles along the perpendicular and horizontal axes shown in [Figure 1]b and c were generated. These profiles were used to study the change in CT numbers (ΔHU) relative to image without implant, in each tissue region. For each image, ΔHU at each point was then converted to Δρe(the change in relative electron density in that point compared to the image without artifact) using CT number calibration curve in [Figure 2]. To study the effect of Δρe on accuracy of dose calculations in treatment planning, the Δρe tolerance levels related to a 2% dose accuracy level presented by Kilby et al.[12] were used. From their study, we used a tolerance range of 0.18–0.09 for Δρe at 6MV for tissue depths ranging from 3 to 6 cm in a typical posterior spinal radiation field, as our benchmark. The effect of different imaging protocols on dose calculation accuracy was investigated, in each tissue region, by comparing the relative number of pixels (as a percentage of total pixels in each region) that have Δρe values above the tolerance level in that region.


 > Results Top


Application of different imaging techniques

Results of image analysis using reconstructed DICOM data are presented as line profiles of CT number in the two directions as shown in [Figure 1]b and [Figure 1]c, hereafter, called horizontal (H) and perpendicular (P) profiles. The H profile, going from left to right of the image, intercepts the following regions: soft tissue, bone, titanium, bone, titanium, bone, and soft tissue. The P profile, going from top to bottom, intercepts soft tissue, bone, spine, bone, and soft tissue regions. In [Figure 3], [Figure 4], [Figure 5], [Figure 6], CT number profiles for the image with artifact (Protocol A with Ti) and profiles for corrected images (imaging Protocols B to D and F with Ti) are compared to profile for the reference image (Protocol A without Ti). For each image, change in CT number at each point in the phantom (along the profile) is calculated by taking the difference between the value of profile at that point for that image and the reference image in HU. In [Figure 3], maximum change in CT number (ΔHU) due to metallic artifacts in the soft tissue, spinal cord, and bone regions was −540, −151, and −836, respectively. The highest change in CT number is seen in regions with severe artifact, i.e., soft-tissue region in H profile and bone regions in-between Ti rods in both H and P profiles. Using the higher kVp (Protocol B), a notable improvement in ΔHU values was seen: −398, −90, and −683 for soft tissue, spine, and bone regions, respectively. However, application of neither of the higher mAs (Protocol C) nor the lower pitch (Protocol D) showed improvement in metallic artifact CT number values, compared to the reference image [Figure 4] and [Figure 5].
Figure 3: Effect of increase in kVp on metallic artifact shown by profiles of computed tomography numbers in different regions of phantom using reconstructed DICOM data of phantom images with and without Ti rods, scanned using Protocol A (120 kVp, 200 mAs, 0.6 mm collimation, 3 mm reconstructed section thickness, kernel value [B30f]) and Protocol B (140 kVp, other parameters similar to Protocol A), in the two directions shown in the insert: (a) the perpendicular profile, (b) the horizontal profile

Click here to view
Figure 4: Effect of increase in mA on metallic artifact shown by profiles of computed tomography numbers in different regions of phantom using reconstructed DICOM data of phantom images with and without Ti rods, scanned using Protocol A (120 kVp, 200 mAs, 0.6 mm collimation, 3 mm reconstructed section thickness, kernel value [B30f]) and Protocol C (tube current modulation turned off, 250 mAs, other parameters similar to Protocol A), in the two directions shown in the insert: (a) the perpendicular profile, (b) the horizontal profile

Click here to view
Figure 5: Effect of decrease in pitch on metallic artifact shown by profiles of computed tomography numbers in different regions of phantom using reconstructed DICOM data of phantom images with and without Ti rods, scanned using Protocol A (120 kVp, 200 mAs, 0.6 mm collimation, 3 mm reconstructed section thickness, kernel value [B30f]) and Protocol D (pitch = 0.45, other parameters similar to Protocol A), in the two directions shown in the insert: (a) the perpendicular profile, (b) the horizontal profile

Click here to view
Figure 6: Effect of extended computed tomography scale postprocessing technique on metallic artifact shown by profiles of computed tomography numbers in different regions of phantom using reconstructed DICOM data of phantom images with and without Ti rods, scanned using Protocols A (120 kVp, 200 mAs, 0.6 mm collimation, 3 mm reconstructed section thickness, kernel value [B30f]), B (kVp = 140, other parameters similar to Protocol A), E (Protocol A and extended computed tomography scale) and F (Protocol B and extended computed tomography scale); in the two directions shown in the insert: (a) the perpendicular profile, (b) the horizontal profile

Click here to view


Application of extended computed tomography scale

[Figure 6] presents the results of applying ECTS postprocessing method for correction of metallic artifact. Application of ECTS improved CT number range in the metallic regions from 3071 HU (in Protocols A to D with Ti) to 8140–8620 for Protocols E, G, and H and 7110–7500 for Protocol F.

Effect of different imaging protocols on accuracy of dose

[Figure 7] shows a comparison of the number of pixels that contain CT number values with error levels above the tolerance necessary to keep calculated dose error to <2%[12] due to image metallic artifact for Protocols A to D. The number of pixels is reported as a percentage of total number of pixels in spine and bone regions.
Figure 7: Comparison of number of pixels that contain computed tomography number values with error levels above the tolerance necessary to keep calculated dose error to <2% (12) due to image metallic artifact in computed tomography images scanned with Protocols A, B, C, and D: Protocol A (120 kVp, 200 mAs, 0.6 mm collimation, 3 mm reconstructed section thickness, kernel value [B30f]), Protocol B (140 kVp, other parameters similar to Protocol A), Protocol C (tube current modulation turned off, 250 mAs, other parameters similar to Protocol A), and Protocol D (pitch = 0.45, other parameters similar to Protocol A)

Click here to view



 > Discussion Top


When planning a radiation treatment using a treatment planning system (TPS), CT images are used to contour the target and the critical organs and to perform dose calculations. Accuracy of dose calculations is affected by error in CT numbers. The goal of this research was to investigate if in the absence of artifact reduction algorithms, clinically available metallic artifact reduction methods are effective in producing the image quality required for treatment planning of spinal radiation fields? We studied the effect of raising kVp and mAs, reducing pitch and application of extended CT number scale (ECTS) by comparing related images.

The maximum change in CT number (ΔHU) due to metallic artifacts was −540, −151, and −836 HU in the soft tissue, spine, and bone regions, respectively. From [Figure 1], these values correspond to error in relative electron density (Δρe) of 0.6, 0.15, and 0.85 for soft tissue, spine, and bone regions, respectively. According to Kilby et al.,[12] such levels of uncertainty in ρ will correspond to uncertainty in calculated dose of higher than 2%, when images containing artifacts are used for treatment planning calculations. In their study, they showed that tolerance range for Δρe was 0.18–0.09 for 6MV in 3–6 cm depths. It should be mentioned that this tolerance has been reported for a certain treatment planning situation, including field specifications and TPS. Therefore, the significance of effect of improvement in metallic artifact on Δρe values in corrected images depends on the specific tolerance requirement for treatment planning dose calculation accuracy.

As was expected, raising kVp improved image artifact: maximum ΔHU (Δρe) was −398 (0.43), −90 (0.09), and −683 (0.7) for soft tissue, spine, bone regions, respectively. Although it is recommended to raise mAs to reduce metallic artifact,[1],[7] in our study, such effect was not seen. The reason is that in our department, for patients with metal implants, TCM option is used. Therefore, further increase in mA did not notably affect image quality. Using a lower pitch did not improve ΔHU either, as recommended.[8] The reason is that pitch reduction with fixed mA, kV, and slice thickness will slow down the table which will lead to increase in imaging time and therefore in mAs. Effect of raising mAs on artifact in systems with TCM was discussed above.

ECTS option was used to postprocess original metal artifact images, higher kVp, mA images, and finally images with lower pitch. Results showed improvement in CT number for metal regions for all protocols. Compared to other protocols, combination of higher kVp and ECTS (Protocol F) resulted in lower CT number values for metal regions, due to higher transmission in metal. The ECTS option did not correct CT numbers in metallic artifact in tissues around the metal regions. In their Monte Carlo study, Bazalova et al.[2] used artifact-corrected images and showed that application of extended CT number calibration instead of the standard DOSXYZnrc/CTCREATE calibration[11] reduced dose calculation errors. However, in their study, the real CT number for metal was not used, i.e., the maximum extent of CT number for metal was limited to the highest CT number defined by the scanner in routine imaging, i.e., 3071 instead of 8000–20,000, the common range for metals.[10] In DOSXYZnrc/CTCREATE, a linear relationship between CT number and density in each tissue is assumed, from a minimum value to a maximum value. The minimum CT number/density in each region is equal to the maximum value of the same parameters in the preceding tissue region. In this study, it was shown that for values higher than 3000 HU, CT calibration curves with standard and extended CT number ranges have different gradients [Figure 2]. Different gradients may assign different densities in one pixel, for metallic alloys or metals containing impurities. In an ECTS postprocessing method, real CT number values are assigned to each pixel. Therefore, this study demonstrates the need for postprocessing CT images with the ECTS method, when using DOSXYZnrc/CTCREATE to calculate dose distributions in the presence of metal implants.

Therefore, our results do not agree with studies that report significant reduction of metal artifacts using ECTS.[5] Assigning correct CT numbers to metal in treatment planning calculations can result in precise delineation of metal and accurate calculation of attenuation in metal, provided that the TPS can use an extended CT number calibration curve. For Monte Carlo calculations, ECTS postprocessed CT images will provide more accurate results, especially if metal has impurities or is an alloy.


 > Conclusions Top


The goal of this research was to investigate if clinically available metallic artifact reduction methods are effective in producing the image quality required for treatment planning of spinal radiation fields? We studied the effect of raising kVp and mAs, reducing pitch and application of ECTS by comparing original and artifact-corrected CT images of a thoracic phantom containing titanium spinal implants. The results showed that raising kVp was the only effective method in reducing metallic artifact in regions surrounding the implants. However, electron density values derived from CT numbers in corrected image did not meet the required tolerance for treatment planning dose calculation accuracy.

ECTS postprocessing method did not have any effect on artifact reduction in regions surrounding the implants. This method was only successful in correcting the CT number range in the metal regions. Assigning correct CT numbers to metal in treatment planning calculations can result in precise delineation of metal and accurate calculation of attenuation in metal, provided that the TPS can use an extended CT number calibration curve. For Monte Carlo calculations, ECTS postprocessed CT images will provide more accurate results, especially if metal has impurities or is an alloy.

Financial support and sponsorship

This research was supported by Isfahan University of Medical Sciences in Isfahan, Iran.

Conflicts of interest

There are no conflicts of interest.



 
 > References Top

1.
Barrett JF, Keat N. Artifacts in CT: Recognition and avoidance 1. Radiographics 2004;24:1679-91.  Back to cited text no. 1
    
2.
Bazalova M, Beaulieu L, Palefsky S, Verhaegena F. Correction of CT artifacts and its influence on monte carlo dose calculations. Med Phys 2007;34:2119-32.  Back to cited text no. 2
    
3.
Keall PJ, Chock LB, Jeraj R, Siebers JV, Mohan R. Image reconstruction and the effect on dose calculation for hip prostheses. Med Dosim 2003;28:113-7.  Back to cited text no. 3
    
4.
Pawlicki T, Ma C. Effect of CT Streaking Artifacts in Monte Carlo dose Distributions for Head and Neck Cancer. In: Proceedings of the 13th International Conference on the Use of Computers in Radiotherapy, Heidelberg; 2000. p. 414-6.  Back to cited text no. 4
    
5.
Yazdi M, Gingras L, Beaulieu L. An adaptive approach to metal artifact reduction in helical computed tomography for radiation therapy treatment planning: Experimental and clinical studies. Int J Radiat Oncol Biol Phys 2005;62:1224-31.  Back to cited text no. 5
    
6.
Hunter TB, Yoshino MT, Dzioba RB, Light RA, Berger WG. Medical devices of the head, neck, and spine. Radiographics 2004;24:257-85.  Back to cited text no. 6
    
7.
Lee MJ, Kim S, Lee SA, Song HT, Huh YM, Kim DH, et al. Overcoming artifacts from metallic orthopedic implants at high-field-strength MR imaging and multi-detector CT. Radiographics 2007;27:791-803.   Back to cited text no. 7
    
8.
Yazdi M, Beaulieu L. Artifacts in spiral X-ray CT scanners: Problems and solutions. Int J Biol Med Sci 2009;4:135-9.  Back to cited text no. 8
    
9.
Klotz E, Kalender WA, Sokiransky R, Felsenberg D. Algorithms for the Reduction of CT Artifacts Caused by Metallic Implants. In: Medical Imaging'90, Newport Beach, 4-9 Feb 90: International Society for Optics and Photonics; 1990. p. 642-50.  Back to cited text no. 9
    
10.
Link TM, Berning W, Scherf S, Joosten U, Joist A, Engelke K, et al. CT of metal implants: Reduction of artifacts using an extended CT scale technique. J Comput Assist Tomogr 2000;24:165-72.  Back to cited text no. 10
    
11.
Walters B, Kawrakow I, Rogers D. DOSXYZnrc Users Manual. NRC Report PIRS; 2005. p. 794.  Back to cited text no. 11
    
12.
Kilby W, Sage J, Rabett V. Tolerance levels for quality assurance of electron density values generated from CT in radiotherapy treatment planning. Phys Med Biol 2002;47:1485-92.  Back to cited text no. 12
    


    Figures

  [Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

  >Abstract>IntroductionMaterials and Me...>Results>Discussion>Conclusions>Article Figures
  In this article
>References

 Article Access Statistics
    Viewed829    
    Printed257    
    Emailed0    
    PDF Downloaded253    
    Comments [Add]    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]