Acceptable Image Quality Research Articles

Low-Dose Cone-Beam Computed Tomography in Swedish Pediatric Patients With Alveolar Clefts Following Alveolar Bone Grafting-A Clinical Study.

The aim of this study was to investigate whether a low-dose cone-beam computed tomography (CBCT) protocol provides diagnostically acceptable image quality for assessing bone healing after alveolar bone grafting. The study cohort comprised 11 patients (aged 7-14 years) with orofacial clefts who had undergone alveolar bone grafting at Skåne University Hospital in Malmö, Sweden. During the postsurgical follow-up at 6 months, each patient was assessed twice: once with a standard-dose CBCT protocol and once with a low-dose CBCT protocol, which in total corresponds to one CBCT examination made with the exposure settings recommended by the manufacturer. Among others, the assessed parameters included subjective image quality, as well as bone graft height, thickness, and integration. No significant differences were found between the standard- and low-dose protocols for most parameters (p > 0.05). Exceptions included subjective image quality (one observer, p = 0.05) and confidence levels during the assessment (three observers, p = 0.01, 0.01, 0.02). The low-dose protocol yielded adequate image quality for postoperative CBCT healing assessment in patients who have undergone alveolar bone grafting. However, the confidence level of observers during the assessment with the low-dose protocol was reduced. This study is registered on ClinicalTrials.gov (NCT06395077). This study is registered on ClinicalTrials.gov (NCT06395077).

Deep learning reconstruction of diffusion-weighted brain MRI for evaluation of patients with acute neurologic symptoms

Purpose: We aimed to evaluate whether the deep-learning (DL) accelerated diffusion weighted image (DWI) is clinically feasible for evaluating patients with acute neurologic symptoms, regarding its shorter study time and acceptable image quality. Materials and methods: In this retrospective study, brain images obtained at DWI with a b-value of 0 s/mm2 and DWI with a b-value of 1000 s/mm2 (DWI 1000) from 321 consecutive patients with acute stroke-like symptom were reconstructed with and without DL algorithm. We compare the diagnostic performance between DL-DWI and conventional DWI for detecting brain lesions, including acute infarction. We assessed the diagnostic accuracy of conventional DWI and DL-DWI and compared the results. Qualitative analysis based on image quality was assessed and compared using a five-point visual scoring system. Apparent diffusion coefficients (ADCs) from DWI with and without DL were also compared. Results: The mean acquisition time for the DL-DWI (49 s) was significantly shorter (P < 0.001) than conventional DWI (165 s). Both DWI with and without DL showed similar performance in diagnosing brain lesions especially sensitivity (98.8% in both DWI and DL-DWI) and specificity (99.5% in both DWI and DL-DWI). Overall image quality, gray-white matter and deep gray matter differentiation of two sequences were similar. DL DWI showed more artifacts than DWI. Lesion conspicuity, especially smaller than 5 mm, was better with DL DWI than conventional DWI (p = 0.03). ADC values of white matter, deep gray matter, and pons with DL were lower than conventional DWI. Conclusions: Compared to conventional DWI, DL-DWI achieved comparable image quality and brain lesion visualization for acute neurological symptoms, with a significantly shorter scan time.

Image Quality and Lesion Detectability of Low-Concentration Iodine Contrast and Low Radiation Hepatic Multiphase CT Using a Deep-Learning-Based Contrast-Boosting Model in Chronic Liver Disease Patients.

This study investigated the image quality and detectability of double low-dose hepatic multiphase CT (DLDCT, which targeted about 30% reductions of both the radiation and iodine concentration) using a vendor-agnostic deep-learning-based contrast-boosting model (DL-CB) compared to those of standard-dose CT (SDCT) using hybrid iterative reconstruction. The CT images of 73 patients with chronic liver disease who underwent DLDCT between June 2023 and October 2023 and had SDCT were analyzed. Qualitative analysis of the overall image quality, artificial sensation, and liver contour sharpness on the arterial and portal phase, along with the hepatic artery clarity was conducted by two radiologists using a 5-point scale. For quantitative analysis, the image noise, signal-to-noise ratio, and contrast-to-noise ratio were measured. The lesion conspicuity was analyzed using generalized estimating equation analysis. Lesion detection was evaluated using the jackknife free-response receiver operating characteristic figures-of-merit. Compared with SDCT, a significantly lower effective dose (16.4 ± 7.2 mSv vs. 10.4 ± 6.0 mSv, 36.6% reduction) and iodine amount (350 mg iodine/mL vs. 270 mg iodine/mL, 22.9% reduction) were utilized in DLDCT. The mean overall arterial and portal phase image quality scores of DLDCT were significantly higher than SDCT (arterial phase, 4.77 ± 0.45 vs. 4.93 ± 0.24, AUCVGA 0.572 [95% CI, 0.507-0.638]; portal phase, 4.83 ± 0.38 vs. 4.92 ± 0.26, AUCVGA 0.535 [95% CI, 0.469-0.601]). Furthermore, DLDCT showed significantly superior quantitative results for the lesion contrast-to-noise ratio (7.55 ± 4.55 vs. 3.70 ± 2.64, p < 0.001) and lesion detectability (0.97 vs. 0.86, p = 0.003). In patients with chronic liver disease, DLDCT using DL-CB can provide acceptable image quality without impairing the detection and evaluation of hepatic focal lesions compared to SDCT.

An authentication scheme for color images with grayscale invariance and recoverability using image demosaicing

This paper proposes a novel recoverable authentication scheme for color images that ensures the consistency of grayscale values after embedment. In the proposed scheme, the color image is sub-sampled based on the Bayer pattern, and the most significant bits (MSBs) of the sub-sampled image are used as the recovery codes. Image blocks are divided into two categories, feasible and infeasible. Feasible blocks are utilized for embedding both authentication and recovery codes, whereas infeasible ones are only used for embedding authentication codes. In case of tampering, the authentication codes are employed to detect the tampered regions, and the undamaged recovery codes can be utilized to reconstruct the sub-sampled image. The demosaicing algorithm is applied to the sub-sampled image to obtain the demosaicked image, which is then used to recover the tampered regions. Experiments demonstrate that the proposed scheme yields an acceptable image quality even when the tampering rate is as high as 40 %.

Dose assessment in petrous bone CT: Effect of tube current modulation and scan length

The aim of this study is to compare the dosimetric performance of two CT acquisition protocols for the petrous bone: a standard fixed mAs protocol and a protocol using tube current modulation (TCM). By analyzing the scan length, dose-length product (DLP) and effective dose (ED) obtained with each protocol, we investigate whether TCM offers an advantage in terms of dose reduction while maintaining comparable image quality. To this end, 249 patients (women: 52%; men: 48%) were included in this study. The first center used TCM (120 kVp, 150 mA reference) for image acquisition, while the other centers opted for fixed mAs values. Acceptability of image quality was assessed using a standardized phantom (model 610 AAPM) by measuring noise uniformity and accuracy, contrast-to-noise ratio (CNR) and spatial resolution. The CTDIvol was not significantly different between center A and the other centers (p > 0.05), while DLP and ED were significantly different (p < 0.001), being reduced by a factor of 2 and 2.23 for scan length. We concluded that a scan with a 14.28% decrease in tube voltage from 140 kVp (136–170–210) mAs fixed to 120 kVp with TCM to 150 mA reference in helical scan mode can be recommended as an optimized protocol during CT examination of petrous bone while reducing DLP and ED to 50% without loss of diagnostic accuracy.

Quantitative and qualitative evaluation of high-quality hepatobiliary phase imaging with shortened timing and utility in patients with compromised liver function.

To compare high flip angle (FA) hepatobiliary-phase (hHBP) imaging with variable time intervals to conventional HBP (cHBP) to assess the impact of increased FA on image quality in shortened HBP imaging. Data from 218 patients, divided into normal liver group (n = 184) and decompensated liver group (n = 34), who underwent liver magnetic resonance imaging (MRI) including 10-min, 15-min, 20-min hHBP, and cHBP were analyzed. Signal-to-noise ratio (SNR), contrast-ratio (CR), contrast-to-noise ratio (CNR), signal intensity ratios (SIRs), and relative enhancement (RE) of the liver were calculated for quantitative analysis. Sharpness, noise, and artifacts of the image, contrast media visibility, overall image quality, and lesion conspicuity were evaluated by two abdominal radiologists. Quantitative analysis showed that SNR, RE, SIR for liver/muscle, liver/spleen, and CR of all hHBP images demonstrated a significantly higher value compared to cHBP images in the normal liver group (p < 0.001). These values were also superior in the normal liver group compared to the decompensated liver group (p < 0.01). In qualitative analysis, both normal and decompensated liver groups exhibited significantly superior image sharpness in all hHBP images compared to cHBP images and the overall image quality of the 15-min and 20-min hHBP did not show significant difference compared to cHBP. All values tended to be better in the normal liver group than the decompensated liver group with statistical significance except for lesion conspicuity (p < 0.01). High-FA HBP has proven to be a valuable image acquisition method, potentially shortening liver MR imaging time while maintaining acceptable image quality.

Investigations of injection strategies to use heparinized normal saline instead of contrast media for intracoronary optical coherence tomography imaging.

The benefits of intravascular imaging-guided percutaneous coronary interventions (PCI) are well established. Intravascular imaging guidance improves short- and long-term outcomes, especially in complex PCI. Optical coherence tomography (OCT) has a higher resolution than intravascular ultrasound. However, the usage of OCT is mainly limited by the need to use contrast for flushing injections, which increases the risk of contrast-induced acute kidney injury, especially in patients with underlying chronic kidney disease. The aim of this study was to prove that flushing techniques with normal saline instead of contrast can be used in OCT imaging and can generate high-quality images. This prospective single-center observational study included patients with indications for OCT-guided PCI. For OCT pullbacks, heparinized saline was injected by an automatic pump injector at different rates, and additional extension catheters for selective coronary artery engagement were used at the operator's discretion. Recordings were made using the Ilumien Optis OCT system (Abbott) and the Dragonfly (Abbott) catheter and were analyzed at 1-mm intervals by two operators. Pullbacks were categorized as having optimal, acceptable, or unacceptable imaging quality. A clinically usable run was determined if >75% of the region of interest length was described as having optimal or acceptable imaging quality. A total of 32 patients were enrolled in the study; 47 different lesions were assessed before and after PCI. In total, 91.5% of runs were described as clinically suitable for use. Heparinized saline injections for OCT imaging are effective in generating good-quality OCT images suitable for clinical use.

Advanced MRI techniques in abdominal imaging.

Magnetic resonance imaging (MRI) is a crucial modality for abdominal imaging evaluation of focal lesions and tissue properties. However, several obstacles, such as prolonged scan times, limitations in patients' breath-hold capacity, and contrast agent-associated artifacts, remain in abdominal MR images. Recent techniques, including parallel imaging, three-dimensional acquisition, compressed sensing, and deep learning, have been developed to reduce the scan time while ensuring acceptable image quality or to achieve higher resolution without extending the scan duration. Quantitative measurements using MRI techniques enable the noninvasive evaluation of specific materials. A comprehensive understanding of these advanced techniques is essential for accurate interpretation of MRI sequences. Herein, we therefore review advanced abdominal MRI techniques.

Nanoimprinted TiO2 Metasurfaces with Reduced Meta-Atom Aspect Ratio and Enhanced Performance for Holographic Imaging.

Metasurface holograms, with the capability to manipulate spatial light amplitudes and phases, are considered next-generation solutions for holographic imaging. However, conventional fabrication approaches for meta-atoms are heavily dependent on electron-beam lithography (EBL), a technique known for its expensive and time-consuming nature. In this paper, a polarization-insensitive metasurface hologram is proposed using a cost-effective and rapid nanoimprinting method with titanium dioxide (TiO2) nanoparticle loaded polymer (NLP). Based on a simulation, it has been found that, despite a reduction in the aspect ratio of meta-atoms of nearly 20%, which is beneficial to silicon master etching, NLP filling, and the mold release processes, imaging efficiency can go up to 54% at wavelength of 532 nm. In addition, it demonstrates acceptable imaging quality at wavelengths of 473 and 671 nm. Moreover, the influence of fabrication errors and nanoimprinting material degradation in terms of residual layer thickness, meta-atom loss or fracture, thermal-induced dimensional variation, non-uniform distribution of TiO2 particles, etc., on the performance is investigated. The simulation results indicate that the proposed device exhibits a high tolerance to these defects, proving its applicability and robustness in practice.

Clinical evaluation of isotropic MAVRIC-SL for symptomatic hip arthroplasties at 3 T MRI

Background3D multi-spectral imaging (MSI) of metal implants necessitates relatively long scan times. ObjectiveWe implemented a fast isotropic 3D MSI technique at 3 T and compared its image quality and clinical utility to non-isotropic MSI in the evaluation of hip implants. MethodsTwo musculoskeletal radiologists scored images from coronal proton density-weighted conventional MAVRIC-SL and an isotropic MAVRIC-SL sequence accelerated with robust-component-analysis on a 3-point scale (3: diagnostic, 2: moderately diagnostic, 1: non-diagnostic) for overall image quality, metal artifact, and visualization around femoral and acetabular components. Grades were compared using a signed Wilcoxon test. Images were evaluated for effusion, synovitis, osteolysis, loosening, pseudotumor, fracture, and gluteal tendon abnormalities. Reformatted axial and sagittal images for both sequences were subsequently generated and compared for image quality with the Wilcoxon test. Whether these reformats increased diagnostic confidence or revealed additional pathology, including findings unrelated to arthroplasty that may contribute to hip pain, was also compared using the McNemar test. Inter-rater agreement was measured by Cohen's kappa. Results39 symptomatic patients with a total of 59 hip prostheses were imaged (mean age, 70 years ±9, 14 males, 25 females). Comparison scores between coronal images showed no significant difference in image quality, metal artifact, or visualization of the femur and acetabulum. Except for loosening, reviewers identified more positive cases of pathology on the original coronally-acquired isotropic sequence. In comparison of reformatted axial and sagittal images, the isotropic sequence scored significantly (p < 0.01) higher for overall image quality (3.0 vs 2.0) and produced significantly (p < 0.01) more cases of increased diagnostic confidence (42.4% vs 7.6%) or additional diagnoses (50.8% vs 22.9%). Inter-rater agreement was substantial (k = 0.798) for image quality. Mean scan times were 4.2 mins (isotropic) and 7.1 mins (non-isotropic). ConclusionCompared to the non-isotropic sequence, isotropic 3D MSI was acquired in less time while maintaining diagnostically acceptable image quality. It identified more pathology, including postoperative complications and potential pain-generating pathology unrelated to arthroplasty. This fast isotropic 3D MSI sequence demonstrates promise for improving diagnostic evaluation of symptomatic hip prostheses at 3 T while simultaneously reducing scan time.

An Open-Source Face-Aware Capture System

Poor-quality facial images pose challenges in biometric authentication, especially in passport photo acquisition and recognition. This study proposes a novel and open-source solution to address these issues by introducing a real-time facial image quality analysis utilizing computer vision technology on a low-power single-board computer. We present an open-source complete hardware solution that consists of a Jetson processor, a 16 MP autofocus RGB camera, a custom enclosure, and a touch sensor LCD for user interaction. To ensure the integrity and confidentiality of captured facial data, Advanced Encryption Standard (AES) is used for secure image storage. Using the pilot data collection, the system demonstrated its ability to capture high-quality images, achieving 98.98% accuracy in storing images of acceptable quality. This open-source, readily deployable, secure system offers promising potential for diverse real-time applications such as passport verification, security systems, etc.

Facial mask-wearing prediction and adaptive gender classification using convolutional neural networks

The world has lived an exceptional time period caused by the Coronavirus pandemic. To limit Covid-19 propagation, governments required people to wear a facial mask outside. In facial data analysis, mask-wearing on the human face creates predominant occlusion hiding the important oral region and causing more challenges for human face recognition and categorisation. The appropriation of existing solutions by taking into consideration the masked context is indispensable for researchers. In this paper, we propose an approach for mask-wearing prediction and adaptive facial human-gender classification. The proposed approach is based on convolutional neural networks (CNNs). Both mask-wearing and gender information are crucial for various possible applications. Experimentation shows that mask-wearing is very well detectable by using CNNs and justifies its use as a prepossessing step. It also shows that retraining with masked faces is indispensable to keep up gender classification performances. In addition, experimentation proclaims that in a controlled face-pose with acceptable image quality' context, the gender attribute remains well detectable. Finally, we show empirically that the adaptive proposed approach improves global performance for gender prediction in a mixed context.

3-D ultrasonic imaging of bolt thread cracks using a linear array probe

The bolt is a widely-used standard component, and the cracking of its thread often leads to final fractures. Traditional 1-D (one-dimensional) A-scan detection and 2-D (two-dimensional) imaging methods only display the position and depth of bolt cracks, without providing information about the crack’s circumferential characteristics. While 1-D array linear scanning and 2-D array can achieve 3-D (three-dimensional) imaging in simple planar components, they have limitations when it comes to bolts. The former has a narrow scanning area, and the latter requires a long post-processing time. To overcome these limitations, the three-dimensional reconstruction algorithm (3-DR) and the half-matrix sparse total focus method (HMS-TFM) are proposed. Our approach involves adopting 360° circumferential scanning based on a 1-D linear array to achieve 3-D imaging of bolt cracks. At each scanning position, the 2-D image data are captured and are then fused to obtain the 3-D image. Different from the simple stacking of 2-D image data in linear scanning, our 3-DR algorithm reconstructs 3-D images based on discrete sampled 2-D images. To reduce the time required for 2-D imaging, the proposed HMS-TFM reduces the required data volume while maintaining acceptable image quality. Experiment results demonstrate that circumferential scanning based on the linear array can achieve 3-D imaging of bolt defects with lower time cost compared to using a matrix array for 3-D imaging. We obtained accurate circumferential quantification for radial-direction cracks with depths of 1 mm, 1.5 mm, and 2 mm with a relative error better than 6.23 %.

Feasibility of deep learning-reconstructed thin-slice single-breath-hold HASTE for detecting pancreatic lesions: A comparison with two conventional T2-weighted imaging sequences

ObjectiveThe objective of this study was to evaluate the clinical feasibility of deep learning reconstruction-accelerated thin-slice single-breath-hold half-Fourier single-shot turbo spin echo imaging (HASTEDL) for detecting pancreatic lesions, in comparison with two conventional T2-weighted imaging sequences: compressed-sensing HASTE (HASTECS) and BLADE. MethodsFrom March 2022 to January 2023, a total of 63 patients with suspected pancreatic-related disease underwent the HASTEDL, HASTECS, and BLADE sequences were enrolled in this retrospectively study. The acquisition time, the pancreatic lesion conspicuity (LCP), respiratory motion artifact (RMA), main pancreatic duct conspicuity (MPDC), overall image quality (OIQ), signal-to-noise ratio (SNR), and contrast-noise-ratio (CNR) of the pancreatic lesions were compared among the three sequences by two readers. ResultsThe acquisition time of both HASTEDL and HASTECS was 16 s, which was significantly shorter than that of 102 s for BLADE. In terms of qualitative parameters, Reader 1 and Reader 2 assigned significantly higher scores to the LCP, RMA, MPDC, and OIQ for HASTEDL compared to HASTECS and BLADE sequences; As for the quantitative parameters, the SNR values of the pancreatic head, body, tail, and lesions, the CNR of the pancreatic lesion measured by the two readers were also significantly higher for HASTEDL than for HASTECS and BLADE sequences. ConclusionsCompared to conventional T2WI sequences (HASTECS and BLADE), deep-learning reconstructed HASTE enables thin slice and single-breath-hold acquisition with clinical acceptable image quality for detection of pancreatic lesions.

A Vision for Global CT Radiation Dose Optimization

The topic of CT radiation dose management is receiving renewed attention since the recent approval by CMS for new CT dose measures. Widespread variation in CT dose persists in practices across the world, suggesting that current dose optimization techniques are lacking. The author outlines a proposed strategy for facilitating global CT radiation dose optimization. CT radiation dose optimization can be defined as the routine use of CT scan parameters that consistently produce images just above the minimum threshold of acceptable image quality for a given clinical indication, accounting for relevant patient characteristics, using the most dose-efficient techniques available on the scanner. To accomplish this, an image quality–based target dose must be established for every protocol; for nonhead CT applications, these target dose values must be expressed as a function of patient size. As variation in outcomes is reduced, the dose targets can be decreased to more closely approximate the minimum image quality threshold. Maintaining CT radiation dose optimization requires a process control program, including measurement, evaluation, feedback, and control. This is best accomplished by local teams made up of radiologists, medical physicists, and technologists, supported with protected time and needed tools, including analytics and protocol management applications. Other stakeholders critical to facilitating CT radiation dose management include researchers, funding agencies, industry, regulators, accreditors, payers, and the ACR. Analogous coordinated approaches have transformed quality in other industries and can be the mechanism for achieving the universal goal of CT radiation dose optimization.

BEAS-Net: a Shape-Prior-Based Deep Convolutional Neural Network for Robust Left Ventricular Segmentation in 2D Echocardiography.

Left ventricle (LV) segmentation of 2D echocardiography images is an essential step in the analysis of cardiac morphology and function and - more generally - diagnosis of cardiovascular diseases. Several deep learning (DL) algorithms have recently been proposed for the automatic segmentation of the LV, showing significant performance improvement over the traditional segmentation algorithms. However, unlike the traditional methods, prior information about the segmentation problem, e.g. anatomical shape information, is not usually incorporated for training the DL algorithms. This can degrade the generalization performance of the DL models on unseen images if their characteristics are somewhat different from those of the training images, e.g. low-quality testing images. In this study, a new shape-constrained deep convolutional neural network (CNN) - called BEAS-Net - is introduced for automatic LV segmentation. The BEAS-Net learns how to associate the image features, encoded by its convolutional layers, with anatomical shape-prior information derived by the B-spline explicit active surface (BEAS) algorithm to generate physiologically meaningful segmentation contours when dealing with artifactual or low-quality images. The performance of the proposed network was evaluated using three different in-vivo datasets and was compared a deep segmentation algorithm based on the U-Net model. Both networks yielded comparable results when tested on images of acceptable quality, but the BEAS-Net outperformed the benchmark DL model on artifactual and low-quality images.

Advanced Abdominal MRI Techniques and Problem-Solving Strategies.

MRI plays an important role in abdominal imaging because of its ability to detect and characterize focal lesions. However, MRI examinations have several challenges, such as comparatively long scan times and motion management through breath-holding maneuvers. Techniques for reducing scan time with acceptable image quality, such as parallel imaging, compressed sensing, and cutting-edge deep learning techniques, have been developed to enable problem-solving strategies. Additionally, free-breathing techniques for dynamic contrast-enhanced imaging, such as extra-dimensional-volumetric interpolated breath-hold examination, golden-angle radial sparse parallel, and liver acceleration volume acquisition Star, can help patients with severe dyspnea or those under sedation to undergo abdominal MRI. We aimed to present various advanced abdominal MRI techniques for reducing the scan time while maintaining image quality and free-breathing techniques for dynamic imaging and illustrate cases using the techniques mentioned above. A review of these advanced techniques can assist in the appropriate interpretation of sequences.

Deep learning-based projection synthesis for low-dose cone-beam computed tomography imaging in image-guided radiotherapy.

The imaging dose of cone-beam computed tomography (CBCT) in image-guided radiotherapy (IGRT) poses adverse effects on patient health. To improve the quality of sparse-view low-dose CBCT images, a projection synthesis convolutional neural network (SynCNN) model is proposed. Included in this retrospective, single-center study were 223 patients diagnosed with brain tumours from Beijing Cancer Hospital. The proposed SynCNN model estimated two pairs of orthogonally direction-separable spatial kernels to synthesize the missing projection in between the input neighboring sparse-view projections via local convolution operations. The SynCNN model was trained on 150 real patients to learn patterns for inter-view projection synthesis. CBCT data from 30 real patients were used to validate the SynCNN, while data from a phantom and 43 real patients were used to test the SynCNN externally. Sparse-view projection datasets with 1/2, 1/4, and 1/8 of the original sampling rate were simulated, and the corresponding full-view projection datasets were restored using the SynCNN model. The tomographic images were then reconstructed with the Feldkamp-Davis-Kress algorithm. The root-mean-square error (RMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) metrics were measured in both the projection and image domains. Five experts were invited to grade the image quality blindly for 40 randomly selected evaluation groups with a four-level rubric, where a score greater than or equal to 2 was considered acceptable image quality. The running time of the SynCNN model was recorded. The SynCNN model was directly compared with the three other methods on 1/4 sparse-view reconstructions. The phantom and patient studies showed that the missing projections were accurately synthesized. In the image domain, for the phantom study, compared with images reconstructed from sparse-view projections, images with SynCNN synthesis exhibited significantly improved qualities with decreased values in RMSE and increased values in PSNR and SSIM. For the patient study, between the results with and without the SynCNN synthesis, the averaged RMSE decreased by 3.4×10-4, 10.3×10-4, and 21.7×10-4, the averaged PSNR increased by 3.4, 6.6, and 9.4 dB, and the averaged SSIM increased by 5.2×10-2, 18.9×10-2 and 33.9×10-2, for the 1/2, 1/4, and 1/8 sparse-view reconstructions, respectively. In expert subjective evaluation, both the median scores and acceptance rates of the images with SynCNN synthesis were higher than those reconstructed from sparse-view projections. It took the model less than 0.01 s to synthesize an inter-view projection. Compared with the three other methods, the SynCNN model obtained the best scores in terms of the three metrics in both domains. The proposed SynCNN model effectively improves the quality of sparse-view CBCT images at a low time cost. With the SynCNN model, the CBCT imaging dose in IGRT could be reduced potentially.

Optimizing cone-beam computed tomography exposure for an effective radiation dose and image quality balance.

The aim of this study was to evaluate the influence of different cone-beam computed tomography (CBCT) acquisition protocols on reducing the effective radiation dose while maintaining image quality. The effective dose emitted by a CBCT device was calculated using thermoluminescent dosimeters placed in a Rando Alderson phantom. Image quality was assessed by 3 experienced evaluators. The relationship between image quality and confidence was evaluated using the Fisher exact test, and the agreement among raters was assessed using the kappa test. Multiple linear regression analysis was performed to investigate whether the technical parameters could predict the effective dose. P-values<0.05 were considered to indicate statistical significance. The optimized protocol (3 mA, 99 kVp, and 450 projection images) demonstrated good image quality and a lower effective dose for radiation-sensitive organs. Image quality and confidence had consistent values for all structures (P<0.05). Multiple linear regression analysis resulted in a statistically significant model. The milliamperage (b=0.504; t=3.406; P=0.027), kilovoltage peak (b=0.589; t=3.979; P=0.016) and number of projection images (b=0.557; t=3.762; P=0.020) were predictors of the effective dose. Optimized CBCT acquisition protocols can significantly reduce the effective radiation dose while maintaining acceptable image quality by adjusting the milliamperage and projection images.

Unsupervised Domain Adaptation for Low-Dose CT Reconstruction via Bayesian Uncertainty Alignment.

Low-dose computed tomography (LDCT) image reconstruction techniques can reduce patient radiation exposure while maintaining acceptable imaging quality. Deep learning (DL) is widely used in this problem, but the performance of testing data (also known as target domain) is often degraded in clinical scenarios due to the variations that were not encountered in training data (also known as source domain). Unsupervised domain adaptation (UDA) of LDCT reconstruction has been proposed to solve this problem through distribution alignment. However, existing UDA methods fail to explore the usage of uncertainty quantification, which is crucial for reliable intelligent medical systems in clinical scenarios with unexpected variations. Moreover, existing direct alignment for different patients would lead to content mismatch issues. To address these issues, we propose to leverage a probabilistic reconstruction framework to conduct a joint discrepancy minimization between source and target domains in both the latent and image spaces. In the latent space, we devise a Bayesian uncertainty alignment to reduce the epistemic gap between the two domains. This approach reduces the uncertainty level of target domain data, making it more likely to render well-reconstructed results on target domains. In the image space, we propose a sharpness-aware distribution alignment (SDA) to achieve a match of second-order information, which can ensure that the reconstructed images from the target domain have similar sharpness to normal-dose CT (NDCT) images from the source domain. Experimental results on two simulated datasets and one clinical low-dose imaging dataset show that our proposed method outperforms other methods in quantitative and visualized performance.