Contrast-to-noise Ratio Research Articles (Page 1)

Reduced contrast media (CM) dose at elevated heart rates (HRs) poses a challenge for coronary artery disease (CAD) assessment. Using a dynamic phantom, we evaluated the performance of a vendor-specific coronary motion-compensated reconstruction (MCR) at various simulated HRs and reduced CM dose in low virtual monoenergetic images (VMIs) from coronary computed tomography angiography (CCTA). A clinical CCTA protocol was used to image a 5-mm artificial coronary artery, filled with 100% (400 Hounsfield units (HU)) and 50% CM dose, using a robotic arm for translation at six velocities (0-50mm/s, 10mm/s steps). Conventional images and VMIs (40-70keV, 10keV steps) were reconstructed without and with MCR. The study evaluated the MCR influence on motion area and contrast-to-noise ratio (CNR) of the resulting segmented arteries (motion area), with the static conventional reconstructed artery at 100% CM dose as the reference, and non-overlapping 95% confidence intervals with the reference indicating significant differences. At 50% CM dose, motion area increased significantly (up to 50%) at elevated velocities (≥ 30mm/s) without MCR, while no significant variations were observed with MCR. Additionally, without MCR, VMIs exhibited significant CNR decreases (up to 65%) at velocities ≥ 30mm/s. Only the combination of MCR and 40keV VMI achieved CNR comparable to the reference, regardless of HR. The combination of MCR with low VMIs enables 50% CM dose reduction, with similar motion area and CNR when compared to conventional CCTA with 100% CM dose. These parameter settings can potentially be used to optimize low CM dose CCTA at higher HRs.

Deep Learning Denoising Algorithm for Improved Assessment of Coronary Arteries in Transcatheter Aortic Valve Implantation CT Imaging.

Approximately 30% of patients with schizophrenia do not respond to antipsychotics. While schizophrenia has been primarily explained by the dopamine dysfunction hypothesis, treatment-resistant schizophrenia (TRS) may involve a different pathophysiology. Neuromelanin (NM), a product of dopamine metabolism in the substantia nigra (SN), indirectly measures long-term dopamine synthesis capacity. Few studies have examined SN NM levels in TRS. Therefore, we investigated the relationship between SN NM levels and treatment responsiveness in schizophrenia. We included age- and sex-matched TRS, patients with schizophrenia in remission of positive symptoms (SZ-R), and healthy controls (HCs). Neuromelanin-sensitive magnetic resonance imaging was used to measure SN NM signals. We also evaluated clinical symptoms and cognitive impairment. We conducted voxel-wise analyses of NM contrast-to-noise ratio (CNR) to compare groups pairwise. Correlation analyses examined relationships between NM signals and symptom severity. Seventy-two participants (n = 24 per group) completed the study. The TRS group had higher dorsal SN CNR than the HC group (510 out of 1948 voxels at p < 0.05, corrected p = 0.005, permutation test). In contrast, no significant differences were observed in the other comparisons. No significant correlations were found between NM CNR and clinical severity. Our findings contrast with previous positron emission tomography studies on dorsal striatal dopamine function. Since the dorsal SN contributes to both the mesolimbic and nigrostriatal pathways, with a relatively greater role in the former, dopamine functions in these pathways may play different roles for treatment responsiveness. Further research with multimodal imaging is needed to examine dopamine function and antipsychotic treatment responsiveness in schizophrenia.

Energy window selection is a critical parameter for optimizing planar gamma image quality in nuclear medicine. In this study, we developed dedicated nuclear medicine phantoms using 3D printing technology to evaluate the impact of varying energy window levels on image quality. Three types of phantoms—a Derenzo phantom with six different sphere diameters, a modified Hoffman phantom incorporating lead for attenuation, and a quadrant bar phantom with four bar thicknesses constructed from bronze filament—were fabricated using Fusion 360 and an Ultimaker S5 3D printer with PLA and bronze-based materials. Planar images were acquired using 37 MBq of Tc-99m for 60 s at energy windows centered at 122, 140, and 159 keV. Quantitative assessments included contrast-to-noise ratio (CNR), coefficient of variation (COV), peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM), comparing all images with the 140 keV image as the reference. The results showed a consistent decline in image quality at 122 keV and 159 keV, with the highest CNR, lowest COV, and optimal PSNR/SSIM values obtained at 140 keV. In visual analysis using the quadrant bar phantom, thinner bars were more clearly discernible at 140 keV than at other energy levels. These findings demonstrate that the application of an appropriate energy window—particularly 140 keV for Tc-99m—substantially improves image quality in planar gamma imaging. The use of customized, material-specific 3D-printed phantoms also enables flexible, reproducible evaluation protocols for energy-dependent imaging optimization and quality assurance in clinical nuclear medicine.

Abstract X-ray phase contrast imaging (XPCI) shows promising potential for analyzing blood vessels affected by a cardiovascular disease and may become a valuable diagnostic tool in angiography. Identifying the most suitable XPCI method for providing optimal visibility at clinically acceptable doses is critical. This study presents a two-dimensional computational simulation in Geant4 to quantitatively compare two XPCI techniques—Inline XPCI and Edge Illumination Single Mask (EISM)—in visualizing atherosclerotic plaques within the projected thickness image of the lumen. The angiographic specimen analyzed was modeled as three concentric cylinders, with half of the inner cylinder simulating the lumen of the blood vessel and the other half simulating an atherosclerotic plaque, the intermediate cylinder simulating the vessel tissue, and the outer cylinder simulating the tissue around the blood vessel. The computational setup consisted of a W-anode polychromatic x-ray source and a photon counting detector, with six spectra and three sample-to-detector distances (magnifications) considered. To enable a direct comparison with Inline XPCI, a multi-material phase retrieval procedure was adapted to work within the EISM framework. Estimates of absorbed dose were also performed for each XPCI method. The results demonstrate that Inline XPCI achieves a superior contrast-to-noise ratio (CNR) as compared to EISM in both analyzed regions of the recovered image, across all spectra and magnifications, while also exhibiting lower absorbed doses. The difference in CNR between the two XPCI methods decreases as the voltage and magnification increases. While Inline XPCI shows superior contrast in the plaque region for all voltages and magnifications studied, EISM excels in recovering the projected thickness in the free-plaque region, achieving better contrast at low voltages for any magnification.

To evaluate whether deep learning-based combined noise reduction and contrast enhancement reconstruction (DLR) improves image quality and resectability prediction accuracy compared to conventional iterative reconstruction (IR) in post-neoadjuvant pancreatic cancer CT assessment. This retrospective study included 114 patients with pancreatic cancer following neoadjuvant therapy. Contrast-enhanced CT images were reconstructed using conventional IR and vendor-neutral ClariACE. Three abdominal radiologists independently assessed image quality (based on 8 parameters: tumor conspicuity, tumor margin, image noise, sharpness of the main pancreatic duct, arterial depiction, venous depiction, plasticity and overall image quality) and determined tumor resectability with confidence levels. Quantitative analysis included aortic and portal venous attenuation measurements and pancreas-to-tumor contrast-to-noise ratio (CNR). Diagnostic performance was evaluated using the area under the receiver operating characteristic curve (AUC) with DeLong's test. Sensitivity, specificity, accuracy, and reader confidence were compared using McNemar's test. DLR demonstrated significantly superior vessel enhancement (p < 0.001) and improved CNR (p < 0.001) versus conventional IR. Two readers consistently rated DLR images higher across all qualitative categories (p < 0.001) except plasticity, while the third reader favored DLR in five of eight parameters (p < 0.001 to 0.020). However, all readers noted increased artificial appearance in DLR images (p < 0.001). Despite image quality improvements, no significant differences were observed in resectability assessment accuracy (62.3%-65.8%), AUC values (0.485-0.520), or high-confidence diagnosis rates between reconstruction methods. Although deep learning-based combined noise reduction and contrast enhancement reconstruction significantly improved quantitative and subjective image quality metrics, it did not enhance diagnostic accuracy for predicting R0 resectability in post-neoadjuvant pancreatic cancer patients.

Radiography is a medical imaging technique that utilizes X-ray radiation to obtain images of organs in the body, including the abdomen. Image quality is very important in supporting the accuracy of diagnosis and can be measured objectively through the Signal-to-Noise Ratio (SNR) and Contrast-to-Noise Ratio (CNR) parameters. As digital technology advances, Python-based image processing offers significant potential in improving the visual and diagnostic quality of radiographic images. This study aims to analyze the effectiveness of digital image processing techniques in improving the quality of computed radiography (CR) radiography, especially in terms of increasing SNR and CNR values. This study uses an experimental approach with CR radiographic image data obtained from dr. Gunawan Mangunkusumo Ambarawa Hospital. The image was processed using the Histogram Equalization and Contrast Limited Adaptive Histogram Equalization (CLAHE) algorithms in the Python platform. The results of the analysis showed that both methods were able to increase the SNR and CNR values, with the Equalization Histogram resulting in the highest CNR of 24.09, while the CLAHE achieved a maximum value of 16.34. Although Histogram Equalization improves global contrast, this method tends to reduce local details. In contrast, CLAHE shows excellence in maintaining anatomical structure and providing a more even contrast increase. Thus, Python-based digital image processing has proven to be effective in improving the quality of abdominal radiographic images and has the potential to be a reliable diagnostic tool in modern radiology practice.

Objective This study aimed to evaluate the diagnostic efficacy of magnetic resonance imaging (MRI) combined with ultrasound and mammography for breast cancer (BC) using array spatial sensitivity encoding technique (ASSET). Methods MRI images are processed using parallel imaging (PI) and ASSET techniques. The signal-to-noise ratio (SNR) of ASSET-diffusion-weighted imaging (DWI) and PI-DWI, as well as the contrast-to-noise ratio (CNR) between lesions and normal breast tissue, were compared. Image quality was also assessed. Using 70 cases of BC as the observation group (OG) and 70 non-BC cases as the control group (CG), the imaging characteristics of MRI, ultrasound, and mammography in both groups were compared. The Accuracy ( Acc ), Sensitivity ( Sen ), Specificity ( Spe ), and consistency of single and combined diagnosis using the three methodologies were evaluated. Results Relative to the PI-DWI sequence, the ASSET-DWI sequence demonstrated notably shorter scanning time, higher CNR between lesions and normal breast tissue, better lesion visualization, clearer lesion margins, fewer image artifacts, and higher overall image quality ( P &amp;lt; 0.05). In contrast to the CG, patients in the OG exhibited a higher proportion of irregular lesion morphology, non-smooth margins, and uneven enhancement on MRI, as well as a higher proportion of low echoic lesions, unclear boundaries, irregular morphology, irregular margins, posterior echo attenuation, and visible blood flow signals on ultrasound. Additionally, a higher proportion of irregular tumor margins, irregular morphology, spiculated signs, calcifications, and absence of capsule were observed on mammography ( P &amp;lt; 0.05). Relative to MRI, ultrasound, and mammography alone, the combined diagnostic method showed significantly higher Acc , Sen , Spe , and Kappa values ( P &amp;lt; 0.05). Conclusion The combined use of MRI, ultrasound, and mammography based on ASSET for BC diagnosis offers significant advantages, providing clinicians with more reliable diagnostic tools.

Assessment of choroidal melanoma and nevus lesions using ultrasound vibro-elastography and parametric imaging approach.

Improved image quality using a whole-heart motion correction algorithm in coronary computed tomography angiography.

Lung cancer screening CT acquisition protocols for three generations of CT systems conforming to German legislation.

Optimization strategy for fat-suppressed T2-weighted images in liver imaging: The combined application of AI-assisted compressed sensing and respiratory triggering.

Effect of gadolinium-based contrast media on intravoxel incoherent motion (IVIM) MRI parameters in brain imaging.

Preliminary Experience of Protocol for Digital Subtraction Angiography Using Diluted Contrast Medium during Endovascular Aneurysm Repair.

VMARC: A general framework for ultrasound imaging enhancement.

Deep Learning-Accelerated Prostate MRI: Improving Speed, Accuracy, and Sustainability.

Metal artifact reduction from surgical clips for intracranial aneurysms in photon-counting detector CT angiography.

Comparison of precise imaging and iterative reconstruction techniques at low doses using the dose right index in 100-kVp cerebral CT angiography.

Deep learning (DL) methods are increasingly applied to address the low signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of low-field MRI (LFMRI). This study evaluates the potential of diffusion models for LFMRI enhancement, comparing the Super-resolution via Repeated Refinement (SR3), a generative diffusion model, to traditional architectures such as CycleGAN and UNet for translating LFMRI to high-field MRI (HFMRI). Using synthetic LFMRI (64mT) FLAIR brain images generated from the BraTS 2019 dataset (3T), the models were assessed with traditional metrics, including structural similarity index (SSIM) and normalized root-mean-squared error (nRMSE), alongside specialized structural error measurements such as gradient entropy (gEn), gradient error (GE), and perception-based image quality evaluator (PIQE). SR3 significantly outperformed (p-value < < 0.05) the other models across all metrics, achieving SSIM scores over 0.97 and excelling in preserving pathological structures such as necrotic core and edema, with lower gEn and GE values. These findings suggest diffusion models are a robust alternative to conventional DL approaches for LF-to-HF MRI translation. By preserving structural details and enhancing image quality, SR3 could improve the clinical utility of LFMRI systems, making high-quality MRI more accessible. This work demonstrates the potential of diffusion models in advancing medical image enhancement and translation.

Abstract Background Pulmonary embolism (PE) is a serious condition requiring accurate imaging for timely diagnosis. While computed tomography pulmonary angiography (CTPA) is the standard tool, its quality may be limited by contrast timing or noise. Dual-energy CT enables virtual monochromatic imaging (VMI), which enhances contrast but suffers from high noise at low keV levels. Advanced VMI (AVMI) addresses this by optimizing noise while preserving contrast. The objective of this study was to evaluate the quantitative image quality of an advanced virtual monoenergetic imaging (AVMI) algorithm in comparison with the equivalent 120-kVp acquisition at dual-energy CT pulmonary angiography (DE-CTPA) in patients with suspected pulmonary embolism (PE). Methods We evaluated dual-source dual-energy CTPA examinations from 123 patients (58 men and 65 women; mean age, 54.2 ± 17.7 years; age range, 20–87 years) who had suspected pulmonary embolism (PE). Images were generated as AVMI series with energies ranging from 40- to 190-keV in 1-keV steps. Attenuation, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were measured in five different pulmonary arteries. AVMIs at 75-keV are considered equivalent to standard 120-kVp single energy CT (SECT) acquisition. The CNR and SNR were compared in AVMIs and 120-kVp equivalent images for each artery by repeated-measures analysis of variance (ANOVA) with post-hoc Bonferroni correction. Result The highest mean attenuation and CNR value of pulmonary arteries were observed in the 40-keV AVMI series, with significant CNR and SNR values compared to 120-kVp equivalent images ( p &lt; 0.05). At 40-keV, energy level improvements of attenuation, SNR, and CNR by 225.8%, 31.0%, 51.2% were observed compared to 120-kVp equivalent images, respectively. Conclusion The advanced noise-optimized VMI series showed significantly higher SNR and CNR in comparison with 120-kVp equivalent images.