Comparison of different deep learning architectures for synthetic CT generation from MR images

PurposeAmong the different available methods for synthetic CT generation from MR images for the task of MR-guided radiation planning, the deep learning algorithms have and do outperform their conventional counterparts. In this study, we investigated the performance of some most popular deep learning architectures including eCNN, U-Net, GAN, V-Net, and Res-Net for the task of sCT generation. As a baseline, an atlas-based method is implemented to which the results of the deep learning-based model are compared.MethodsA dataset consisting of 20 co-registered MR-CT pairs of the male pelvis is applied to assess the different sCT production methods' performance. The mean error (ME), mean absolute error (MAE), Pearson correlation coefficient (PCC), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR) metrics were computed between the estimated sCT and the ground truth (reference) CT images.ResultsThe visual inspection revealed that the sCTs produced by eCNN, V-Net, and ResNet, unlike the other methods, were less noisy and greatly resemble the ground truth CT image. In the whole pelvis region, the eCNN yielded the lowest MAE (26.03 ± 8.85 HU) and ME (0.82 ± 7.06 HU), and the highest PCC metrics were yielded by the eCNN (0.93 ± 0.05) and ResNet (0.91 ± 0.02) methods. The ResNet model had the highest PSNR of 29.38 ± 1.75 among all models. In terms of the Dice similarity coefficient, the eCNN method revealed superior performance in major tissue identification (air, bone, and soft tissue).ConclusionsAll in all, the eCNN and ResNet deep learning methods revealed acceptable performance with clinically tolerable quantification errors.     

T1-weighted and T2-weighted MRI image synthesis with convolutional generative adversarial networks

BackgroundThe objective of this study was to propose an optimal input image quality for a conditional generative adversarial network (GAN) in T1-weighted and T2-weighted magnetic resonance imaging (MRI) images.Materials and methodsA total of 2,024 images scanned from 2017 to 2018 in 104 patients were used. The prediction framework of T1-weighted to T2-weighted MRI images and T2-weighted to T1-weighted MRI images were created with GAN. Two image sizes (512 × 512 and 256 × 256) and two grayscale level conversion method (simple and adaptive) were used for the input images. The images were converted from 16-bit to 8-bit by dividing with 256 levels in a simple conversion method. For the adaptive conversion method, the unused levels were eliminated in 16-bit images, which were converted to 8-bit images by dividing with the value obtained after dividing the maximum pixel value with 256.ResultsThe relative mean absolute error (rMAE ) was 0.15 for T1-weighted to T2-weighted MRI images and 0.17 for T2-weighted to T1-weighted MRI images with an adaptive conversion method, which was the smallest. Moreover, the adaptive conversion method has a smallest mean square error (rMSE) and root mean square error (rRMSE), and the largest peak signal-to-noise ratio (PSNR) and mutual information (MI). The computation time depended on the image size.ConclusionsInput resolution and image size affect the accuracy of prediction. The proposed model and approach of prediction framework can help improve the versatility and quality of multi-contrast MRI tests without the need for prolonged examinations.     

Conditional generative adversarial networks to generate pseudo low monoenergetic CT image from a single-tube voltage CT scanner

PurposeTo generate pseudo low monoenergetic CT images of the abdomen from 120-kVp CT images with cGAN.Materials and MethodsWe retrospectively included 48 patients who underwent contrast-enhanced abdominal CT using dual-energy CT. We reconstructed paired data sets of 120 kVp CT images and virtual low monoenergetic (55-keV) CT images. cGAN was prepared to generate pseudo 55-keV CT images from 120-kVp CT images. The pseudo 55 keV CT images in epoch 10, 50, 100, and 500 were compared to the 55 keV images generated using peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM).ResultsThe PSNRs were 28.0, 28.5, 28.6, and 28.8 at epochs 10, 50, 100, and 500, respectively. The SSIM was approximately constant from epochs 50 to 500.ConclusionPseudo low monoenergetic abdominal CT images were generated from 120-kVp CT images using cGAN, and the images had good quality similar to that of monochromatic images obtained with DECT software.     

Elimination of biological and physical artifacts in abdomen and brain computed tomography procedures using filtering techniques

IntroductionMedical images are usually affected by biological and physical artifacts or noise, which reduces image quality and hence poses difficulties in visual analysis, interpretation and thus requires higher doses and increased radiographs repetition rate.ObjectivesThis study aims at assessing image quality during CT abdomen and brain examinations using filtering techniques as well as estimating the radiogenic risk associated with CT abdomen and brain examinations.Materials and MethodsThe data were collected from the Radiology Department at Royal Care International (RCI) Hospital, Khartoum, Sudan. The study included 100 abdominal CT images and 100 brain CT images selected from adult patients. Filters applied are namely: Mean filter, Gaussian filter, Median filter and Minimum filter. In this study, image quality after denoising is measured based on the Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), and the Structural Similarity Index Metric (SSIM).ResultsThe results show that the images quality parameters become higher after applications of filters. Median filter showed improved image quality as interpreted by the measured parameters: PSNR and SSIM, and it is thus considered as a better filter for removing the noise from all other applied filters.DiscussionThe noise removed by the different filters applied to the CT images resulted in enhancing high quality images thereby effectively revealing the important details of the images without increasing the patients’ risks from higher doses.ConclusionsFiltering and image reconstruction techniques not only reduce the dose and thus the radiation risks, but also enhances high quality imaging which allows better diagnosis.     

Quantifying near metal visibility using dual energy computed tomography and iterative metal artifact reduction in a fracture phantom

PurposeTo quantitatively assess CT image quality and fracture visibility using virtual monochromatic imaging and iterative metal artifact reduction (iMAR) in a femoral bone fracture phantom with different fixation implants.MethodsA custom made phantom was scanned at 120-kVp and 140-kVp single-energy and 100/150-kVp dual-energy. Three stainless steel and two titanium implants with different thicknesses were placed on the phantom containing simulated one and two mm fractures. Single-energy CT images were reconstructed with and without iMAR, while DECT images were reconstructed at monochromatic energies between 70 and 190 keV. Non-metal scans were used as a reference. A Fourier power spectrum method and fracture model were used to analyze several anatomical areas.ResultsCT-value deviations of titanium implants were much lower compared to stainless steel implants. These deviations decreased for both DECT and iMAR. Fracture visibility, measured with the fracture model, improved the most when DECT was used while artifact reduction benefitted more from iMAR. The optimal monochromatic energy for metal artifact reduction, based on CT-value deviation, varied for each metal between 130 and 150 keV. The fracture model provided a signal-to-noise ratio for the near metal fracture visibility, providing the optimal keV.ConclusioniMAR and high keV monochromatic images extracted from DECT both reduce metal artifacts caused by different metal fixation implants. Quantitative femoral phantom results show that DECT is superior to iMAR regarding fracture visualization adjacent to metal fixation implants. The introduction of new artifacts when using iMAR impedes its value in near metal fixation implant imaging.     

Estimation and evaluation of pseudo-CT images using linear regression models and texture feature extraction from MRI images in the brain region to design external radiotherapy planning

AimThe aim of this study is to construct and evaluate Pseudo-CT images (P-CTs) for electron density calculation to facilitate external radiotherapy treatment planning.BackgroundDespite numerous benefits, computed tomography (CT) scan does not provide accurate information on soft tissue contrast, which often makes it difficult to precisely differentiate target tissues from the organs at risk and determine the tumor volume. Therefore, MRI imaging can reduce the variability of results when registering with a CT scan.Materials and methodsIn this research, a fuzzy clustering algorithm was used to segment images into different tissues, also linear regression methods were used to design the regression model based on the feature extraction method and the brightness intensity values. The results of the proposed algorithm for dose-volume histogram (DVH), Isodose curves, and gamma analysis were investigated using the RayPlan treatment planning system, and VeriSoft software. Furthermore, various statistical indices such as Mean Absolute Error (MAE), Mean Error (ME), and Structural Similarity Index (SSIM) were calculated.ResultsThe MAE of a range of 45–55 was found from the proposed methods. The relative difference error between the PTV region of the CT and the Pseudo-CT was 0.5, and the best gamma rate was 95.4% based on the polar coordinate feature and proposed polynomial regression model.ConclusionThe proposed method could support the generation of P-CT data for different parts of the brain region from a collection of MRI series with an acceptable average error rate by different evaluation criteria.     

Utility of the Iodine Overlay Technique and Virtual Nonenhanced Images for the Preoperative T Staging of Colorectal Cancer by Dual-Energy CT with Tin Filter Technology

Objectives

To evaluate the diagnostic accuracy and the potential radiation dose reduction of dual-energy CT (DECT) for tumor (T) staging of colorectal cancer (CRC) using iodine overlay (IO) and virtual nonenhanced (VNE) images.

Materials and Methods

This retrospective study included 103 consecutive patients who underwent nonenhanced CT and enhanced DECT for preoperative CRC staging. Enhanced weighted-average (WA), IO and VNE images were reconstructed from enhanced 80 kVp and Sn140 kVp scans. Two radiologists assessed image qualities of the true nonenhanced (TNE) and VNE images. For T-staging, another two radiologists independently interpreted all scans in two separate reading sessions: in the first session, only images derived from the single phase DECT acquisition (IO and VNE images) were read. In the second reading session after 30 to 50 (average:42) days, the same assessment was again performed with the TNE and enhanced WA images thereby simulating conventional dual-phase single-energy CT. The tumor node metastasis (TNM) system was used for staging with histopathologic reports as gold standard. Analysis of variance was used for statistical analysis.

Results

The signal-to-noise ratios (SNRs) of the tumors and normal reference tissues showed significant correlation between the TNE and VNE images (P<0.01). The mean iodine overlay value (48.4 HU±12.2) and enhancement (49.4 HU±11.8) value of CRCs had no significant difference (P = 0.52).The mean image noise on TNE (5.0±1.1) and VNE (5.3±1.1) images were similar (P = 0.07). The quantitative qualities of the VNE images were mildly inferior to the TNE images. Overall accuracy of T-stage CRC when using single-phase acquisition was slightly better than the dual-phase acquisition (90.3% vs 87.4%) (P = 0.51). The mean dose of the single-phase DECT acquisition was 6.2mSv comparing with 14.3mSv of dual-phase.

Conclusion

Single-phase DECT using IO and VNE images yields a high accuracy in T-staging of CRCs. Thereby, the radiation exposure of the patients can be reduced.     

A comparison of denoising methods for X-ray fluoroscopic images

Fluoroscopic images exhibit severe signal-dependent quantum noise, due to the reduced X-ray dose involved in image formation, that is generally modelled as Poisson-distributed. However, image gray-level transformations, commonly applied by fluoroscopic device to enhance contrast, modify the noise statistics and the relationship between image noise variance and expected pixel intensity. Image denoising is essential to improve quality of fluoroscopic images and their clinical information content. Simple average filters are commonly employed in real-time processing, but they tend to blur edges and details. An extensive comparison of advanced denoising algorithms specifically designed for both signal-dependent noise (AAS, BM3Dc, HHM, TLS) and independent additive noise (AV, BM3D, K-SVD) was presented. Simulated test images degraded by various levels of Poisson quantum noise and real clinical fluoroscopic images were considered. Typical gray-level transformations (e.g. white compression) were also applied in order to evaluate their effect on the denoising algorithms. Performances of the algorithms were evaluated in terms of peak-signal-to-noise ratio (PSNR), signal-to-noise ratio (SNR), mean square error (MSE), structural similarity index (SSIM) and computational time. On average, the filters designed for signal-dependent noise provided better image restorations than those assuming additive white Gaussian noise (AWGN). Collaborative denoising strategy was found to be the most effective in denoising of both simulated and real data, also in the presence of image gray-level transformations. White compression, by inherently reducing the greater noise variance of brighter pixels, appeared to support denoising algorithms in performing more effectively.     

Non-invasive characterization of coronary artery atherosclerotic plaque using dual energy CT: Explanation in ex-vivo samples

PurposeIn this study non-calcified plaque composition is evaluated by Dual Energy CT (DECT). Energy Dispersive X-ray Spectroscopy (EDS) has been used to study the Plaque composition. An attempt has been made to explain the DECT results with EDS analysis.MethodsThirty-two ex-vivo human cadaver coronary artery samples were scanned by DECT and data was evaluated to calculate their effective atomic number and electron density (Z_eff & ρ_e) by inversion method. Result of DECT was compared with pathology to assess their differentiating capability. The EDS study was used to explain DECT outcome.ResultsDECT study was able to differentiate vulnerable plaque from stable with 87% accuracy (area under the curve (AUC):0.85 [95% confidence interval {CI}:0.73–0.98}] and Kappa Coefficient (KC):0.75 with respect to pathology. EDS revealed significant compositional difference in vulnerable and stable plaque at p < .05. The weight percentage of higher atomic number elements like F, Na, Mg, S, Si, P, Cl, K and Ca was found to be slightly more in vulnerable plaques as compared to a stable plaque. EDS also revealed a significantly increased weight percentage of nitrogen in stable plaques.ConclusionsThe EDS results were able to explain the outcomes of DECT study. This study conclusively explains the physics of DECT as a tool to assess the nature of non-calcified plaques as vulnerable and stable. The method proposed in this study allows for differentiation between vulnerable and stable plaque using DECT.     

Artificial neural networking to estimate the leaf area for invasive plant Wedelia trilobata

Leaf area are very important parameter for the understanding of growth and physiological responses of invasive plant species under different environmental factors. This study was conducted to build non-destructive leaf area model of Wedelia trilobata that were grown in greenhouse. Regression analysis and artificial neural network (ANN) approaches were used for the development of leaf area model with the help of leaf length and width of 262 plants samples. In selection of best method under both techniques, the lower value of mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE) and higher value of R² were considered. According to the results it was found that ANN have higher value of (R² = 0.96) and lower value of error (MAE = 0.023, RMSE = 0.379, MAPE = 0.001) than regression analysis (R² = 0.94, MAE = 0.111, RMSE = 1.798, MAPE = 0.0005). It was concluded that error between predicted and actual value was less under ANN. Therefore, ANN model approach can be used as an alternating method for the estimation of leaf area. Through estimation of leaf area, invasive plant growth can predict under different environment conditions.     

Geometric inaccuracy and co-registration errors for CT,DynaCT and MRI images used in robotic stereotactic radiosurgery treatment planning

PurposeTo measure the combined errors due to geometric inaccuracy and image co-registration on secondary images (dynamic CT angiography (dCTA), 3D DynaCT angiography (DynaCTA), and magnetic resonance images (MRI)) that are routinely used to aid in target delineation and planning for stereotactic radiosurgery (SRS).MethodsThree phantoms (one commercial and two in-house built) and two different analysis approaches (commercial and MATLAB based) were used to quantify the magnitude of geometric image distortion and co-registration errors for different imaging modalities within CyberKnife’s MultiPlan treatment planning software. For each phantom, the combined errors were reported as a mean target registration error (TRE). The mean TRE’s for different intramodality imaging parameters (e.g., mAs, kVp, and phantom set-ups) and for dCTA, DynaCTA, and MRI systems were measured.ResultsOnly X-ray based imaging can be performed with the commercial phantom, and the mean TRE ± standard deviation values were large compared to the in-house analysis using MATLAB. With the 3D printed phantom, even drastic changes in treatment planning CT imaging protocols did not greatly influence the mean TRE (<0.5 mm for a 1 mm slice thickness CT). For all imaging modalities, the largest mean TRE was found on DynaCT, followed by T2-weighted MR images (albeit all <1 mm).ConclusionsThe user may overestimate the mean TRE if the commercial phantom and MultiPlan were used solely. The 3D printed phantom design is a sensitive and suitable quality assurance tool for measuring 3D geometric inaccuracy and co-registration errors across all imaging modalities.     

Stomach Virtual Non-Enhanced CT with Second-Generation,Dual-Energy CT: A Preliminary Study

Objectives

To compare the true non-enhanced (TNE) and virtual non-enhanced (VNE) data sets in patients who underwent gastric preoperative dual-energy CT (DECT) and to evaluate potential radiation dose reduction by omitting a TNE scan.

Methods

A total of 74 patients underwent gastric DECT. The mean CT values, length, image quality and effective radiation doses for VNE and TNE images were compared.

Results

There was no statistical difference in maximal thickness of gastric tumors and maximal diameter of enlarged lymph nodes among the TNE and VNE images (P>0.05). The mean CT value differences between TNE and VNE were statistically significant for all tissue types, except for aorta attenuation measurements (P<0.05), but the absolute differences were under 10 HU. Lower noise was found for VNE images than TNE images (P<0.01). Image quality of VNE was diagnostic but lower than that of TNE (P<0.01). The dose reduction achieved by omitting the TNE acquisition was 21.40±4.44%.

Conclusion

VNE scan may potentially replace TNE as part of a multi-phase gastric preoperative staging imaging protocol with consequent saving in radiation dose.     

Comparison of Two Hybrid Models for Forecasting the Incidence of Hemorrhagic Fever with Renal Syndrome in Jiangsu Province,China

Background

Cases of hemorrhagic fever with renal syndrome (HFRS) are widely distributed in eastern Asia, especially in China, Russia, and Korea. It is proved to be a difficult task to eliminate HFRS completely because of the diverse animal reservoirs and effects of global warming. Reliable forecasting is useful for the prevention and control of HFRS.

Methods

Two hybrid models, one composed of nonlinear autoregressive neural network (NARNN) and autoregressive integrated moving average (ARIMA) the other composed of generalized regression neural network (GRNN) and ARIMA were constructed to predict the incidence of HFRS in the future one year. Performances of the two hybrid models were compared with ARIMA model.

Results

The ARIMA, ARIMA-NARNN ARIMA-GRNN model fitted and predicted the seasonal fluctuation well. Among the three models, the mean square error (MSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of ARIMA-NARNN hybrid model was the lowest both in modeling stage and forecasting stage. As for the ARIMA-GRNN hybrid model, the MSE, MAE and MAPE of modeling performance and the MSE and MAE of forecasting performance were less than the ARIMA model, but the MAPE of forecasting performance did not improve.

Conclusion

Developing and applying the ARIMA-NARNN hybrid model is an effective method to make us better understand the epidemic characteristics of HFRS and could be helpful to the prevention and control of HFRS.     

Dual-energy CT imaging of orbits during episcleral brachytherapy with Ru-106 plaques: A phantom study on its potential for plaque position verification

PurposeTo investigate the potential of dual energy CT (DECT) to suppress metal artifacts and accurately depict episcleral brachytherapy Ru-106 plaques after surgical placement.MethodsAn anthropomorphic phantom simulating the adult head after surgical placement of a Ru-106 plaque was employed. Nine DECT acquisition protocols for orbital imaging were applied. Monochromatic 140 keV images were generated using iterative reconstruction and an available metal artifact reduction algorithm. Generated image datasets were graded by four observers regarding the ability to accurate demarcate the Ru-106 plaque. Objective image quality and visual grading analysis (VGA) was performed to compare different acquisition protocols. The DECT imaging protocol which allowed accurate plaque demarcation at minimum exposure was identified. The eye-lens dose from orbital DECT, with and without the use of radioprotective bismuth eye-shields, was determined using Monte Carlo methods.ResultsAll DECT acquisition protocols were judged to allow clear demarcation of the plaque borders despite some moderate streaking/shading artifacts. The differences between mean observers’ VGA scores for the 9 DECT imaging protocols were not statistically significant (p > 0.05). The eye-lens dose from the proposed low-exposure DECT protocol was found to be 20.1 and 22.8 mGy for the treated and the healthy eye, respectively. Bismuth shielding was found to accomplish >40% reduction in eye-lens dose without inducing shielding-related artifacts that obscure plaque delineation.ConclusionsDECT imaging of orbits after Ru-106 plaque positioning for ocular brachytherapy was found to allow artifact-free delineation of plaque margins at relatively low patient exposure, providing the potential for post-surgery plaque position verification.     

Fast Second Degree Total Variation Method for Image Compressive Sensing

This paper presents a computationally efficient algorithm for image compressive sensing reconstruction using a second degree total variation (HDTV2) regularization. Firstly, a preferably equivalent formulation of the HDTV2 functional is derived, which can be formulated as a weighted L ₁-L ₂ mixed norm of second degree image derivatives under the spectral decomposition framework. Secondly, using the equivalent formulation of HDTV2, we introduce an efficient forward-backward splitting (FBS) scheme to solve the HDTV2-based image reconstruction model. Furthermore, from the averaged non-expansive operator point of view, we make a detailed analysis on the convergence of the proposed FBS algorithm. Experiments on medical images demonstrate that the proposed method outperforms several fast algorithms of the TV and HDTV2 reconstruction models in terms of peak signal to noise ratio (PSNR), structural similarity index (SSIM) and convergence speed.     

A Novel Weighted Total Difference Based Image Reconstruction Algorithm for Few-View Computed Tomography

In practical applications of computed tomography (CT) imaging, due to the risk of high radiation dose imposed on the patients, it is desired that high quality CT images can be accurately reconstructed from limited projection data. While with limited projections, the images reconstructed often suffer severe artifacts and the edges of the objects are blurred. In recent years, the compressed sensing based reconstruction algorithm has attracted major attention for CT reconstruction from a limited number of projections. In this paper, to eliminate the streak artifacts and preserve the edge structure information of the object, we present a novel iterative reconstruction algorithm based on weighted total difference (WTD) minimization, and demonstrate the superior performance of this algorithm. The WTD measure enforces both the sparsity and the directional continuity in the gradient domain, while the conventional total difference (TD) measure simply enforces the gradient sparsity horizontally and vertically. To solve our WTD-based few-view CT reconstruction model, we use the soft-threshold filtering approach. Numerical experiments are performed to validate the efficiency and the feasibility of our algorithm. For a typical slice of FORBILD head phantom, using 40 projections in the experiments, our algorithm outperforms the TD-based algorithm with more than 60% gains in terms of the root-mean-square error (RMSE), normalized root mean square distance (NRMSD) and normalized mean absolute distance (NMAD) measures and with more than 10% gains in terms of the peak signal-to-noise ratio (PSNR) measure. While for the experiments of noisy projections, our algorithm outperforms the TD-based algorithm with more than 15% gains in terms of the RMSE, NRMSD and NMAD measures and with more than 4% gains in terms of the PSNR measure. The experimental results indicate that our algorithm achieves better performance in terms of suppressing streak artifacts and preserving the edge structure information of the object.     

Evaluation of a cycle-generative adversarial network-based cone-beam CT to synthetic CT conversion algorithm for adaptive radiation therapy

PurposeImage-guided radiation therapy could benefit from implementing adaptive radiation therapy (ART) techniques. A cycle-generative adversarial network (cycle-GAN)-based cone-beam computed tomography (CBCT)-to-synthetic CT (sCT) conversion algorithm was evaluated regarding image quality, image segmentation and dosimetric accuracy for head and neck (H&N), thoracic and pelvic body regions.MethodsUsing a cycle-GAN, three body site-specific models were priorly trained with independent paired CT and CBCT datasets of a kV imaging system (XVI, Elekta). sCT were generated based on first-fraction CBCT for 15 patients of each body region. Mean errors (ME) and mean absolute errors (MAE) were analyzed for the sCT. On the sCT, manually delineated structures were compared to deformed structures from the planning CT (pCT) and evaluated with standard segmentation metrics. Treatment plans were recalculated on sCT. A comparison of clinically relevant dose-volume parameters (D₉₈, D₅₀ and D₂ of the target volume) and 3D-gamma (3%/3mm) analysis were performed.ResultsThe mean ME and MAE were 1.4, 29.6, 5.4 Hounsfield units (HU) and 77.2, 94.2, 41.8 HU for H&N, thoracic and pelvic region, respectively. Dice similarity coefficients varied between 66.7 ± 8.3% (seminal vesicles) and 94.9 ± 2.0% (lungs). Maximum mean surface distances were 6.3 mm (heart), followed by 3.5 mm (brainstem). The mean dosimetric differences of the target volumes did not exceed 1.7%. Mean 3D gamma pass rates greater than 97.8% were achieved in all cases.ConclusionsThe presented method generates sCT images with a quality close to pCT and yielded clinically acceptable dosimetric deviations. Thus, an important prerequisite towards clinical implementation of CBCT-based ART is fulfilled.     

A study to determine the impact of IMPT optimization techniques on prostate synthetic CT image sets dose comparison against CT image sets

BackgroundThe objective of this study is to determine the impact of intensity modulated proton therapty (IMPT) optimization techniques on the proton dose comparison of commercially available magnetic resonance for calculating attenuation (MRCA T) images, a synthetic computed tomography CT (sCT) based on magnetic resonance imaging (MRI) scan against the CT images and find out the optimization technique which creates plans with the least dose differences against the regular CT image sets.Material and methodsRegular CT data sets and sCT image sets were obtained for 10 prostate patients for the study. Six plans were created using six distinct IMPT optimization techniques including multi-field optimization (MFO), single field uniform dose (SFUD) optimization, and robust optimization (RO) in CT image sets. These plans were copied to MRCA T, sCT datasets and doses were computed. Doses from CT and MRCA T data sets were compared for each patient using 2D dose distribution display, dose volume histograms (DVH), homogeneity index (HI), conformation number (CN) and 3D gamma analysis. A two tailed t-test was conducted on HI and CN with 5% significance level with a null hypothesis for CT and sCT image sets.ResultsAnalysis of ten CT and sCT image sets with different IMPT optimization techniques shows that a few of the techniques show significant differences between plans for a few evaluation parameters. Isodose lines, DVH, HI, CN and t-test analysis shows that robust optimizations with 2% range error incorporated results in plans, when re-computed in sCT image sets results in the least dose differences against CT plans compared to other optimization techniques. The second best optimization technique with the least dose differences was robust optimization with 5% range error.ConclusionThis study affirmatively demonstrates the impact of IMPT optimization techniques on synthetic CT image sets dose comparison against CT images and determines the robust optimization with 2% range error as the optimization technique which gives the least dose difference when compared to CT plans.     

Phase correlation applied to the 3D registration of CT and CBCT image volumes

PurposeIn this study, a 3D phase correlation algorithm was investigated to test feasibility for use in determining the anatomical changes that occur throughout a patient's radiotherapy treatment. The algorithm determines the transformations between two image volumes through analysis in the Fourier domain and has not previously been used in radiotherapy for 3D registration of CT and CBCT volumes.MethodsVarious known transformations were applied to a patient's prostate CT image volume to create 12 different test cases. The mean absolute error and standard deviation were determined by evaluating the difference between the known contours and those calculated from the registration process on a point-by-point basis. Similar evaluations were performed on images with increasing levels of noise added. The improvement in structure overlap offered by the algorithm in registering clinical CBCT to CT images was evaluated using the Dice Similarity Coefficient (DSC).ResultsA mean error of 2.35 (σ = 1.54) mm was calculated for the 12 deformations applied. When increasing levels of noise were introduced to the images, the mean errors were observed to rise up to a maximum increase of 1.77 mm. For CBCT to CT registration, maximum improvements in the DSC of 0.09 and 0.46 were observed for the bladder and rectum, respectively.ConclusionsThe Fourier-based 3D phase correlation registration algorithm investigated displayed promising results in CT to CT and CT to CBCT registration, offers potential in terms of efficiency and robustness to noise, and is suitable for use in radiotherapy for monitoring patient anatomy throughout treatment.     

Deep learning methods to generate synthetic CT from MRI in radiotherapy: A literature review

PurposeIn radiotherapy, MRI is used for target volume and organs-at-risk delineation for its superior soft-tissue contrast as compared to CT imaging. However, MRI does not provide the electron density of tissue necessary for dose calculation. Several methods of synthetic-CT (sCT) generation from MRI data have been developed for radiotherapy dose calculation. This work reviewed deep learning (DL) sCT generation methods and their associated image and dose evaluation, in the context of MRI-based dose calculation.MethodsWe searched the PubMed and ScienceDirect electronic databases from January 2010 to March 2021. For each paper, several items were screened and compiled in figures and tables.ResultsThis review included 57 studies. The DL methods were either generator-only based (45% of the reviewed studies), or generative adversarial network (GAN) architecture and its variants (55% of the reviewed studies). The brain and pelvis were the most commonly investigated anatomical localizations (39% and 28% of the reviewed studies, respectively), and more rarely, the head-and-neck (H&N) (15%), abdomen (10%), liver (5%) or breast (3%). All the studies performed an image evaluation of sCTs with a diversity of metrics, with only 36 studies performing dosimetric evaluations of sCT.ConclusionsThe median mean absolute errors were around 76 HU for the brain and H&N sCTs and 40 HU for the pelvis sCTs. For the brain, the mean dose difference between the sCT and the reference CT was <2%. For the H&N and pelvis, the mean dose difference was below 1% in most of the studies. Recent GAN architectures have advantages compared to generator-only, but no superiority was found in term of image or dose sCT uncertainties. Key challenges of DL-based sCT generation methods from MRI in radiotherapy is the management of movement for abdominal and thoracic localizations, the standardization of sCT evaluation, and the investigation of multicenter impacts.