Comparison and evaluation of different deep learning models of synthetic CT generation from CBCT for nasopharynx cancer adaptive proton therapy

BACKGROUND

Cone-beam computed tomography (CBCT) scanning is used for patient setup in image-guided radiotherapy. However, its inaccurate CT numbers limit its applicability in dose calculation and treatment planning.

PURPOSE

This study compares four deep learning methods for generating synthetic CT (sCT) to determine which method is more appropriate and offers potential for further clinical exploration in adaptive proton therapy for nasopharynx cancer.

METHODS

CBCTs and deformed planning CT (dCT) from 75 patients (60/5/10 for training, validation and testing) were used to compare cycle-consistent Generative Adversarial Network (cycleGAN), Unet, Unet+cycleGAN and conditionalGenerative Adversarial Network (cGAN) for sCT generation. The sCT images generated by each method were evaluated against dCT images using mean absolute error (MAE), structural similarity (SSIM), peak signal-to-noise ratio (PSNR), spatial non-uniformity (SNU) and radial averaging in the frequency domain. In addition, dosimetric accuracy was assessed through gamma analysis, differences in water equivalent thickness (WET), and dose-volume histogram metrics.

RESULTS

The cGAN model has demonstrated optimal performance in the four models across various indicators. In terms of image quality under global condition, the average MAE has been reduced to 16.39HU, SSIM has increased to 95.24%, and PSNR has increased to 28.98. Regarding dosimetric accuracy, the gamma passing rate (2%/2 mm) has reached 99.02%, and the WET difference is only 1.28 mm. The D95 value of CTVs coverage and Dmax value of spinal cord, brainstem show no significant differences between dCT and sCT generated by cGAN model.

CONCLUSIONS

The cGAN model has been shown to be a more suitable approach for generating sCT using CBCT, considering its characteristics and concepts. The resulting sCT has the potential for application in adaptive proton therapy.

Does the shape of the field-of-view influence the magnitude of artefacts from high-density materials in cone-beam computed tomography?

OBJECTIVE

To compare cylindrical and convex triangular field-of-views (FOVs) concerning the magnitude of artefacts from high-density materials in cone-beam CT (CBCT).

METHODS AND MATERIALS

Cylinders of amalgam, chromium-cobalt, titanium, and zirconia were individually placed in the anterior and posterior regions of a polymethylmethacrylate phantom and scanned using cylindrical and convex triangular FOVs of the Veraview X800 CBCT device. Using the Image J software, 15 square regions of interest (ROIs) were placed in the axial reconstruction around the middle level of the cylinder and at distances of 0.5, 1.0, and 1.5 cm from the centre of the cylinder. Mean grey value and standard deviation of each ROI were averaged for each distance and subtracted from the values of a control ROI to calculate the magnitude of the artefacts by the grey value mean difference (GVMD) and grey value standard deviation (GVSD). Multiway analysis of variance with Tukey post-hoc test with a significance level of 5% evaluated the effect of the shape of the FOV, position inside the FOV, high-density material, and the distance of the artefact from the material.

RESULTS

The convex triangular FOV increased the GVSD for all materials in the anterior and posterior regions at 0.5 cm compared to the cylindrical FOV (p < 0.0001). The convex triangular FOV showed greater GVMD for chromium-cobalt and zirconium in the anterior region and all materials in the posterior region at all distances (p < 0.0001).

CONCLUSION

The FOV shape influences the magnitude of artefacts from high-density materials. The convex triangular FOV showed greater artefact magnitude with variability among the high-density materials, region in the FOV, and distance from the material.

Comparison of artifact values of prosthodontic materials with 2 different cone beam computed tomography devices

STATEMENT OF PROBLEM

Prosthodontic materials may cause unexpected artifacts in cone beam computed tomography (CBCT) images, but studies quantifying these artifacts are sparse.

PURPOSE

The purpose of this in vitro study was to compare the artifact expression of fixed prosthodontic materials with different CBCT devices.

MATERIAL AND METHODS

Ten prosthodontic materials (Co-Cr-Mo alloy, interim acrylic resin, polyetheretherketone, feldspathic ceramic, lithium disilicate glass-ceramic, zirconia-reinforced lithium silicate ceramic, zircon core, and 3 monolithic zirconias) were scanned with 2 CBCT devices. The materials were placed in polymethyl methacrylate resin to simulate clinical conditions. To assess the impact of the devices on artifacts, the gray values of 8 areas in each material image were analyzed. The data were analyzed with the Kruskal-Wallis and Wilcoxon Signed-Rank tests (α=.05).

RESULTS

Statistically significant differences were found in the artifact expression of the materials (P<.001) and between CBCT devices (P<.001).

CONCLUSIONS

The artifact expression of polymeric and ceramic materials in CBCT images was less than that of other materials. The milliampere-second (mAs) value of CBCT devices had a significant impact on the artifact level.

Optimization of CBCT data with image processing methods and production with fused deposition modeling 3D printing

The present study has investigated the effect of the removal of artifacts in cone beam computed tomography (CBCT) images with image processing techniques to dental implant planning. The aim of this study has been to benefit from the novel image processing techniques and additive manufacturing technologies in order to change the existing approach in the usage of the 3D model in the orthogonal surgery, traumatic cases, and tumor operations and to solve the restrictions in surgical operations. In the study, firstly, 3 × 3, 5 × 5, and 7 × 7 kernel values were determined on the CBCT image data of the patient. The determined kernel values were applied on CBCT images by choosing median, median-mean-Gaussian (MMG), and bilateral filters, which are quite successful in removing noise in medical images. A thresholding process to separate teeth and bones from soft tissue regions on CBCT images, histogram normalization for a balanced color distribution, morphology operations to reduce noise areas, and tooth and bone boundaries were determined as closely as possible to patient anatomy. The original image and the images obtained from image enhancement techniques were compared. Results showed that the 3 × 3 median filtering method from three different kernel values out of three different image processing methods used in the study greatly improved the artifacts. It has also been shown that the availability of image processing and additive manufacturing methods on CBCT images has been shown to be a highly important factor before dental surgery planning.

Comparative Evaluation of Artifacts Originated by Four Different Post Materials Using Different CBCT Settings

The aim of this study was to evaluate whether cone beam computed tomography (CBCT) images in the presence of four different post materials, obtained from different kVps with varying resolutions and varying metal artifact reduction (MAR) algorithms, differed in artifact estimation, and to compare tooth regions in terms of artifact value.

MATERIALS AND METHODS

Forty premolar teeth were used in this study. Root canals were treated, and teeth were randomly distributed into four subgroups (n = 10) for the preparation of post materials: titanium, gold (Nordin), quartz fiber (Bisco DT Light), and glass fiber (Rely X). The CBCT images were taken with two different kVps, three different metal artifact reduction (MAR) algorithm options, and two different resolutions. For each protocol, the effective dose was calculated according to the dose area production (DAP) value. The standard analysis of variance technique and the Tukey multiple comparison adjustment method were used to assess interactions among material types, kVp, MAR, and voxel settings.

RESULTS

More artifacts were found in the middle third than in the cervical third (p < 0.05). The mean value of artifacts was highest for gold (Nordin), 90 kVp, no MAR, and 100 voxel size. Glass or quartz fiber posts at low resolution, with high MAR and 96 kVp, originated fewer artifacts. Moreover, the use of 90 and 96 kVp with 200 voxel size and high MAR provided the least amount of radiation.

CONCLUSION

The best setting for radiographic follow-up of post materials on the Planmeca ProMax is 96 kVp with low resolution and high MAR; this setting produced one of the lowest effective doses.

CLINICAL SIGNIFICANCE

This study estimated the best scanning protocol by lowering the effective dose to a minimum level according to the "as low as reasonably achievable" principle, as well as assessing the tooth region and the post material generating the fewest artifacts, in order to prevent image interpretation challenges such as false-positive and false-negative results stemming from the deterioration of the visibility of the root canal due to perforation, fractures, and voids in the root canal region.

Comparison of the expression of the volumetric alteration artifact in cylindrical and triangular fields of view in two cone-beam computed tomography devices

OBJECTIVE

To compare the expression of the volumetric alteration (VA) artifact between cylindrical and convex triangular fields of view (FOVs) using high-density materials in different positions in two cone-beam computed tomography (CBCT) devices.

MATERIALS AND METHODS

Cylinders of five high-density materials (amalgam, chromium-cobalt, gutta-percha, titanium, and zirconium) with known physical volume were individually submitted to CBCT acquisition in four positions inside a polymethylmethacrylate phantom using two different FOV shapes (convex triangular and cylindrical) on the Veraviewepocs® R100 (R100) and Veraview® X800 (X800) devices. Two oral radiologists obtained the tomographic volumes by segmenting each cylinder. The difference between the tomographic and physical volumes corresponded to the VA. These values were analyzed by intraclass correlation coefficient and analysis of variance for repeated measures with Tukey post hoc test (α = 5%).

RESULTS

The FOV influenced the VA only in the X800 device (p = 0.014): the VA in the triangular FOV was greater than in the cylindrical FOV. The VA in the triangular FOV of the X800 device was greater than the R100 device (p < 0.0001). The material influenced the expression of the VA only in the R100 device (p < 0.0001); gutta-percha presented the highest VA, being underestimated, and differing from the other materials (p < 0.01).

CONCLUSION

The triangular FOV increased the VA of high-density materials in the X800 device.

CLINICAL RELEVANCE

It is important to know if there is an influence on the volumetric alteration artifact of dental materials due to the different image formation geometry in the convex triangular FOV.

Synthetic CT generation from CBCT images via deep learning

PURPOSE

Cone-beam computed tomography (CBCT) scanning is used daily or weekly (i.e., on-treatment CBCT) for accurate patient setup in image-guided radiotherapy. However, inaccuracy of CT numbers prevents CBCT from performing advanced tasks such as dose calculation and treatment planning. Motivated by the promising performance of deep learning in medical imaging, we propose a deep U-net-based approach that synthesizes CT-like images with accurate numbers from planning CT, while keeping the same anatomical structure as on-treatment CBCT.

METHODS

We formulated the CT synthesis problem under a deep learning framework, where a deep U-net architecture was used to take advantage of the anatomical structure of on-treatment CBCT and image intensity information of planning CT. U-net was chosen because it exploits both global and local features in the image spatial domain, matching our task to suppress global scattering artifacts and local artifacts such as noise in CBCT. To train the synthetic CT generation U-net (sCTU-net), we include on-treatment CBCT and initial planning CT of 37 patients (30 for training, seven for validation) as the input. Additional replanning CT images acquired on the same day as CBCT after deformable registration are utilized as the corresponding reference. To demonstrate the effectiveness of the proposed sCTU-net, we use another seven independent patient cases (560 slices) for testing.

RESULTS

We quantitatively compared the resulting synthetic CT (sCT) with the original CBCT image using deformed same-day pCT images as reference. The averaged accuracy measured by mean absolute error (MAE) between sCT and reference CT (rCT) on testing data is 18.98 HU, while MAE between CBCT and rCT is 44.38 HU.

CONCLUSIONS

The proposed sCTU-net can synthesize CT-quality images with accurate CT numbers from on-treatment CBCT and planning CT. This potentially enables advanced CBCT applications for adaptive treatment planning.

Application of radiopaque micro-particle fillers for 3-D imaging of periodontal pocket analogues using cone beam CT

BACKGROUND

Periodontitis is an infectious/inflammatory disease most often diagnosed by deepening of the gingival sulcus, which leads to periodontal pockets (PPs) conventional manual periodontal probing does not provide detailed information on the three-dimensional (3-D) nature of PPs.

OBJECTIVES

To determine whether accurate 3-D analyses of the depths and volumes of calibrated PP analogues (PPAs) can be obtained by conventional cone beam computed tomography (CBCT) coupled with novel radiopaque micro-particle fillers (described in the companion paper) injected into the PPAs.

METHODS

Two PPA models were employed: (1) a human skull model with artificial gingiva applied to teeth with alveolar bone loss and calibrated PPAs, and (2) a pig jaw model with alveolar bone loss and surgically-induced PPAs The PPAs were filled with controlled amounts of radiopaque micro-particle filler using volumetric pipetting Inter-method and intra-method agreement tests were then used to compare the PPA depths and volumes obtained from CBCT images with values obtained by masked examiners using calibrated manual methods.

RESULTS

Significant inter-method agreement (0.938-0.991) and intra-method agreement (0.94-0.99) were obtained when comparing analog manual data to digital CBCT measurements enabled by the radiopaque filler.

SIGNIFICANCE

CBCT imaging with radiopaque micro-particle fillers is a plausible means of visualizing and digitally assessing the depths, volumes, and 3-D shapes of PPs This approach could transform the diagnosis and treatment planning of periodontal disease, with particular initial utility in complex cases Efforts to confirm the clinical practicality of these fillers are currently in progress.

Effect of anatomical region on the formation of metal artefacts produced by dental implants in cone beam computed tomographic images

OBJECTIVES

To quantitatively compare metal artefacts produced by implants in different maxillomandibular regions on cone beam CT (CBCT) images.

METHODS

A total of 200 implants selected from CBCT examinations were divided into four groups: Group 1 (n = 50)-implants located in the anterior maxilla; Group 2 (n = 50)-implants located in the posterior maxilla; Group 3 (n = 50)-implants located in the anterior mandible and Group 4 (n = 50)-implants located in the posterior mandible. The implants were further classified as isolated or adjacent to other implants. Three axial reconstructions were selected for each sampled implant (apical, middle and cervical). On each slice, the artefacts produced by the implants were counted. The Mann-Whitney test was used to compare the variables between groups. The Kruskal-Wallis and Student-Newman-Keuls tests were used to compare the axial reconstructions.

RESULTS

The mandible showed a greater number of artefacts than the maxilla (apical image: p = 0.0024; middle image: p < 0.0001). The anterior region produced more artefacts than the posterior region (apical image: p = 0.0105; middle image: p < 0.0316). There was no significant difference in the number of artefacts between isolated and adjacent implants, and the cervical image was most affected by artefacts.

CONCLUSIONS

Dental implants always produce metal artefacts in CBCT images, and these artefacts are affected by the anatomical location in the dental arch.

Quantitative assessment of image artifacts from root filling materials on CBCT scans made using several exposure parameters

Purpose

To quantify artifacts from different root filling materials in cone-beam computed tomography (CBCT) images acquired using different exposure parameters.

Materials and Methods

Fifteen single-rooted teeth were scanned using 8 different exposure protocols with 3 different filling materials and once without filling material as a control group. Artifact quantification was performed by a trained observer who made measurements in the central axial slice of all acquired images in a fixed region of interest using ImageJ. Hyperdense artifacts, hypodense artifacts, and the remaining tooth area were identified, and the percentages of hyperdense and hypodense artifacts, remaining tooth area, and tooth area affected by the artifacts were calculated. Artifacts were analyzed qualitatively by 2 observers using the following scores: absence (0), moderate presence (1), and high presence (2) for hypodense halos, hypodense lines, and hyperdense lines. Two-way ANOVA and the post-hoc Tukey test were used for quantitative and qualitative artifact analysis. The Dunnet test was also used for qualitative analysis. The significance level was set at P<.05.

Results

There were no significant interactions among the exposure parameters in the quantitative or qualitative analysis. Significant differences were observed among the studied filling materials in all quantitative analyses. In the qualitative analyses, all materials differed from the control group in terms of hypodense and hyperdense lines (P<.05). Fiberglass posts did not differ statistically from the control group in terms of hypodense halos (P>.05).

Conclusion

Different exposure parameters did not affect the objective or subjective observations of artifacts in CBCT images; however, the filling materials used in endodontic restorations did affect both types of assessments.

Three-dimensional virtual simulation of alar width changes following bimaxillary osteotomies

The aim of this study was to evaluate the accuracy of three-dimensional (3D) soft tissue simulation of nose width changes following bimaxillary osteotomies and to identify patient- and surgery-related factors that may affect the accuracy of simulation. Sixty patients (mean age 26 years) who underwent bimaxillary osteotomies participated in this study. Cone beam computed tomography scans were acquired preoperatively and at 1-year postoperative. The 3D hard and soft tissue rendered preoperative and postoperative virtual head models were superimposed, after which the maxilla and mandible were segmented and aligned to the postoperative position. The postoperative changes in alar width were simulated using a mass tensor model (MTM)-based algorithm and compared with the postoperative outcome. 3D cephalometric analyses were used to quantify the simulation error. The postoperative alar width was increased by 1.6±1.1mm and the mean error between the 3D simulation and the actual postoperative alar width was 1.0±0.9mm. The predictability was not correlated to factors such as age, sex, alar cinch suture, VY closure, maxillary advancement, or a history of surgically assisted rapid maxillary expansion. The MTM-based simulation model of postoperative alar width change was found to be reasonably accurate, although there is room for further improvement.

A retrospective, semi-quantitative image quality analysis of cone beam computed tomography (CBCT) and MSCT in the diagnosis of distal radius fractures

OBJECTIVE

To compare image quality and diagnostic validity of CBCT and MSCT for distal radius fractures.

METHODS

35 CBCT and 33 MSCT scans were retrospectively reviewed with a visual grading scale regarding the depiction of cortical bone, trabecular bone, articular surfaces, and soft tissue. The extent and type of artefacts was analyzed. Agreement on AO classification and measurement of cortical disruption and length of the fracture gap was determined. Fracture reduction was evaluated in post-treatment x-rays. Statistical analysis was performed with visual grading characteristics (VGC), chi square tests, and Kendall's coefficient of concordance.

RESULTS

CBCT performed significantly worse for cortical bone, articular surfaces, and especially soft tissue. Trabecular bone showed no significant difference. Significantly more CBCT images showed artefacts. Physics-based artefacts were the most common. CBCT scans also showed motion artefacts. There was no significant difference in agreement on AO classification. The agreement on measurements was substantial for both modalities. Slightly more fractures that had undergone MSCT imaging showed adequate reduction.

CONCLUSION

This initial study of an orthopaedic extremity CBCT scanner showed that the image quality of a CBCT scanner remains inferior for most structures at standard settings. Diagnostic validity of both modalities for distal radius fractures seems similar.

KEY POINTS

• Subjectively, CBCT remains inferior to MSCT in depicting most structures. • Similar diagnostic validity for CBCT and MSCT imaging of distal radius fractures. • CBCT is a possible alternative to MSCT in musculoskeletal imaging. • Visual grading characteristics (VGC) analysis proves useful in analyzing visual grading scales.

Assessment of metal artefact reduction around dental titanium implants in cone beam CT

OBJECTIVES

The aim of this study was to investigate if the metal artefact reduction (MAR) tool used in the software of the ORTHOPANTOMOGRAPH(®) OP300 (Instrumentarium Dental, Tuusula, Finland) can improve the gray value levels in post-operative implant scans.

METHODS

20 potential implant sites were selected from 5 edentulous human dry mandibles. Each mandible was scanned by a CBCT scanner, and images were produced under three different conditions: implant sites drilled but no implants inserted, implants inserted without application of MAR and implants inserted with application of MAR. Using Geomagic(®) Studio 2012 (Geomagic, Morrisville, NC) and 3Diagnosys(®) v. 5.3.1 (3Diemme(®) SRL, Cantù, Italy) software, three scans of each mandible were superimposed. The mean gray value of identical regions of bone around the implants was derived for each condition. The differences between gray value measurements at implant sites derived from different conditions were assessed.

RESULTS

A significant difference was found between mean gray values from the scans with no implants inserted and with implants inserted (with and without MAR) (p = 0.012). No significant difference was revealed for gray values measured from scans with and without MAR (p = 0.975).

CONCLUSIONS

The MAR tool in the software of the ORTHOPANTOMOGRAPH OP300 CBCT scanner does not significantly correct the voxel gray values affected by the metal artefact in the vicinity of an implant in human dry mandibles.