Recent trends in AI applications for pelvic MRI: a comprehensive review.

Magnetic resonance imaging (MRI) is an essential tool for evaluating pelvic disorders affecting the prostate, bladder, uterus, ovaries, and/or rectum. Since the diagnostic pathway of pelvic MRI can involve various complex procedures depending on the affected organ, the Reporting and Data System (RADS) is used to standardize image acquisition and interpretation. Artificial intelligence (AI), which encompasses machine learning and deep learning algorithms, has been integrated into both pelvic MRI and the RADS, particularly for prostate MRI. This review outlines recent developments in the use of AI in various stages of the pelvic MRI diagnostic pathway, including image acquisition, image reconstruction, organ and lesion segmentation, lesion detection and classification, and risk stratification, with special emphasis on recent trends in multi-center studies, which can help to improve the generalizability of AI.

A longitudinal pilot study to assess temporal changes in coronary arterial 18F-sodium fluoride uptake.

PURPOSE: How coronary arterial 18F-sodium fluoride (18F-NaF) uptake on positron emission tomography changes over the long term and what clinical factors impact the changes remain unclear. We sought to investigate the topics in this study. METHODS: We retrospectively studied 15 patients with ≥1 coronary atherosclerotic lesion/s detected on cardiac computed tomography who underwent baseline and follow-up (interval of >3 years) 18F-NaF positron emission tomography/computed tomography scans. Focal 18F-NaF uptake in each lesion was quantified using maximum tissue-to-background ratio (TBRmax). The temporal change in TBRmax was assessed using a ratio of follow-up to baseline TBRmax (R-TBRmax). RESULTS: A total of 51 lesions were analyzed. Mean R-TBRmax was 0.96 ± 0.21. CT-based lesion features (location, obstructive stenosis, plaque types, features of high-risk plaque) did not correlate with an increase in R-TBRmax. In multivariate analysis, baseline TBRmax significantly correlated with higher follow-up TBRmax (ß = 0.57, P < 0.0001), and the presence of diabetes mellitus significantly correlated with both higher follow-up TBRmax (ß = 0.34, P = 0.001) and elevated R-TBRmax (ß = 0.40, P = 0.003). CONCLUSION: Higher coronary arterial 18F-NaF uptake is likely to remain continuously high. Diabetes mellitus affects the long-term increase in coronary arterial 18F-NaF uptake.

18F-sodium fluoride positron emission tomography following coronary computed tomography angiography in predicting long-term coronary events: a 5-year follow-up study.

PURPOSE: The predictive value of 18F-sodium fluoride (18F-NaF) positron emission tomography (PET) in combination with coronary computed tomography (CT) angiography (CCTA) for future coronary events has attracted interest. We evaluated the potential of 18F-NaF PET/CT following CCTA to predict major coronary events (MACE) during a 5-year follow-up period. METHODS: Forty patients with coronary atherosclerotic lesions detected on CCTA underwent 18F-NaF PET/CT examination. Each lesion was evaluated for luminal stenosis and high-risk plaque (HRP) with < 30 Hounsfield units and a > 1.1 remodeling index on CCTA. Focal 18F-NaF uptake in each lesion was quantified using the maximum tissue-to-background ratio (TBRmax), and the maximum TBRmax per patient (M-TBRmax) was determined. We followed MACE (cardiac death, acute coronary syndrome, and/or coronary revascularization > 6 months after 18F-NaF PET/CT) for 5 years. RESULTS: In total, 142 coronary lesions were analyzed. Eleven patients experienced any MACE. Patients with MACE showed a higher M-TBRmax than those without (1.40 ± .19 vs. 1.18 ± .18, P = .0011), and the optimal M-TBRmax cutoff to predict MACE was 1.29. Patients with M-TBRmax of ≥ 1.29 had a higher risk of MACE than those with lower values (P = .012, log-rank test), whereas patients with obstructive stenosis and those with HRP did not. Multivariate Cox proportional analysis adjusted for age, sex, coronary risk factors, and CCTA findings showed that M-TBRmax of ≥ 1.29 remained an independent predictor of 5-year MACE (hazard ratio, 5.4; 95% confidence interval, 1.1-25.4; P = .034). CONCLUSION: 18F-NaF PET/CT following CCTA provides useful strategies to predict 5-year MACE.

The reliability and utility of on-site CT-derived fractional flow reserve (FFR) based on fluid structure interactions: comparison with FFRCT based on computational fluid dynamics, invasive FFR, and resting full-cycle ratio.

Fractional flow reserve (FFR) derived off-site by coronary computed tomography angiography (CCTA) (FFRCT) is obtained by applying the principles of computational fluid dynamics. This study aimed to validate the overall reliability of on-site CCTA-derived FFR based on fluid structure interactions (CT-FFR) and assess its clinical utility compared with FFRCT, invasive FFR, and resting full-cycle ratio (RFR). We calculated the CT-FFR for 924 coronary vessels in 308 patients who underwent CCTA for clinically suspected coronary artery disease. Of these patients, 35 patients with at least one obstructive stenosis (> 50%) detected on CCTA underwent both CT-FFR and FFRCT for further investigation. Furthermore, 24 and 20 patients underwent invasive FFR and RFR in addition to CT-FFR, respectively. The inter-observer correlation (r) of CT-FFR was 0.93 (95% confidence interval [CI] 0.85-0.97, P < 0.0001) with a mean absolute difference of - 0.0042 (limits of agreement - 0.073, 0.064); 97.3% of coronary arteries without obstructive lesions on CCTA had negative results for ischemia on CT-FFR (> 0.80). The correlation coefficient between CT-FFR and FFRCT for 105 coronary vessels was 0.87 (95% CI 0.82-0.91, P < 0.0001) with a mean absolute difference of - 0.012 (limits of agreement - 0.12, 0.10). CT-FFR correlated well with both invasive FFR (r = 0.66, 95% CI 0.36-0.84, P = 0.0003) and RFR (r = 0.78, 95% CI 0.51-0.91, P < 0.0001). These data suggest that CT-FFR can potentially substitute for FFRCT and correlates closely with invasive FFR and RFR with high reproducibility. Our findings should be proven by further clinical investigation in a larger cohort.

Clinical applications of artificial intelligence in liver imaging.

This review outlines the current status and challenges of the clinical applications of artificial intelligence in liver imaging using computed tomography or magnetic resonance imaging based on a topic analysis of PubMed search results using latent Dirichlet allocation. LDA revealed that "segmentation," "hepatocellular carcinoma and radiomics," "metastasis," "fibrosis," and "reconstruction" were current main topic keywords. Automatic liver segmentation technology using deep learning is beginning to assume new clinical significance as part of whole-body composition analysis. It has also been applied to the screening of large populations and the acquisition of training data for machine learning models and has resulted in the development of imaging biomarkers that have a significant impact on important clinical issues, such as the estimation of liver fibrosis, recurrence, and prognosis of malignant tumors. Deep learning reconstruction is expanding as a new technological clinical application of artificial intelligence and has shown results in reducing contrast and radiation doses. However, there is much missing evidence, such as external validation of machine learning models and the evaluation of the diagnostic performance of specific diseases using deep learning reconstruction, suggesting that the clinical application of these technologies is still in development.

New trend in artificial intelligence-based assistive technology for thoracic imaging.

Although there is no solid agreement for artificial intelligence (AI), it refers to a computer system with intelligence similar to that of humans. Deep learning appeared in 2006, and more than 10 years have passed since the third AI boom was triggered by improvements in computing power, algorithm development, and the use of big data. In recent years, the application and development of AI technology in the medical field have intensified internationally. There is no doubt that AI will be used in clinical practice to assist in diagnostic imaging in the future. In qualitative diagnosis, it is desirable to develop an explainable AI that at least represents the basis of the diagnostic process. However, it must be kept in mind that AI is a physician-assistant system, and the final decision should be made by the physician while understanding the limitations of AI. The aim of this article is to review the application of AI technology in diagnostic imaging from PubMed database while particularly focusing on diagnostic imaging in thorax such as lesion detection and qualitative diagnosis in order to help radiologists and clinicians to become more familiar with AI in thorax.

Triaging of COVID-19 patients using low dose chest CT: Incidence and factor analysis of lung involvement on CT images.

INTRODUCTION: Despite an increase in CT studies to evaluate patients with coronavirus disease 2019 (COVID-19), their indication in triage is not well-established. The purpose was to investigate the incidence of lung involvement and analyzed factors related to lung involvement on CT images for establishment of the indication for CT scans in the triaging of COVID-19 patients. METHODS: Included were 192 COVID-19 patients who had undergone CT scans and blood tests for triaging. Two radiologists reviewed the CT images and recorded the incidence of lung involvement. The prediction model for lung involvement on CT images using clinico-laboratory variables [age, gender, body mass index, oxygen saturation of the peripheral artery (SpO2), comorbidities, symptoms, and blood data] were developed by multivariate logistic regression with cross-validation. RESULTS: In 120 of the 192 patients (62.5%), CT revealed lung involvement. The patient age (odds ratio [OR]; 4.95, 95% confidence interval [CI]; 0.93-26.49), albumin (OR; 4.66, 95%CI; 1.37-15.84), lactate dehydrogenase (OR; 5.79, 95%CI; 1.43-23.38) and C-reactive protein (OR; 8.93, 95%CI; 4.13-19.29) were selected for the final prediction model for lung involvement on CT images. The cross-validated area under the receiver operating characteristics (ROC) curve was 0.83. CONCLUSIONS: The high incidence of lung involvement (62.5%) was confirmed on CT images. The proposed prediction model that includes the patient age, albumin, lactate dehydrogenase, and C-reactive protein may be useful for predicting lung involvement on CT images and may assist in deciding whether triaged COVID-19 patients should undergo CT.

Optimal Phosphate Control Related to Coronary Artery Calcification in Dialysis Patients.

BACKGROUND: In patients on maintenance dialysis, cardiovascular mortality risk is remarkably high, which can be partly explained by severe coronary artery calcification (CAC). Hyperphosphatemia has been reported to be associated with the severity of CAC. However, the optimal phosphate range in patients on dialysis remains unknown. This study was planned to compare the effects on CAC progression of two types of noncalcium-based phosphate binders and of two different phosphate target ranges. METHODS: We conducted a randomized, open-label, multicenter, interventional trial with a two by two factorial design. A total of 160 adults on dialysis were enrolled and randomized to the sucroferric oxyhydroxide or lanthanum carbonate group, with the aim of reducing serum phosphate to two target levels (3.5-4.5 mg/dl in the strict group and 5.0-6.0 mg/dl in the standard group). The primary end point was percentage change in CAC scores during the 12-month treatment. RESULTS: The full analysis set included 115 patients. We observed no significant difference in percentage change in CAC scores between the lanthanum carbonate group and the sucroferric oxyhydroxide group. On the other hand, percentage change in CAC scores in the strict group (median of 8.52; interquartile range, -1.0-23.9) was significantly lower than that in the standard group (median of 21.8; interquartile range, 10.0-36.1; P=0.006). This effect was pronounced in older (aged 65-74 years) versus younger (aged 20-64 years) participants (P value for interaction =0.003). We observed a similar finding for the absolute change in CAC scores. CONCLUSIONS: Further study with a larger sample size is needed, but strict phosphate control shows promise for delaying progression of CAC in patients undergoing maintenance hemodialysis. CLINICAL TRIAL REGISTRY NAME AND REGISTRATION NUMBER: Evaluate the New Phosphate Iron-Based Binder Sucroferric Oxyhydroxide in Dialysis Patients with the Goal of Advancing the Practice of EBM (EPISODE), jRCTs051180048.

Native Myocardial T1 Value in Predicting 1-Year Outcomes in Patients with Nonischemic Dilated Cardiomyopathy Experiencing Recent Heart Failure.

The evidence for the clinical implications, especially the short-term utility, of native myocardial T1 value (T1native) on cardiac magnetic resonance (CMR) in nonischemic dilated cardiomyopathy (NIDCM) is scant. We investigated the potential of T1native to assess left ventricular (LV) myocardial characteristics and predict 1-year outcomes in patient with NIDCM experiencing recent heart failure (HF).Forty-five patients with NIDCM and HF symptoms within 3 months underwent CMR with cine, non-contrast T1 mapping, and late gadolinium enhancement (LGE). T1native per patient was defined as an averaged T1 value of 5 short-axis slices of base-to-apex LV myocardium. The appearance of LGE was visually examined. T1native correlated with the LV end-diastolic dimension normalized to height (LVEDD) (r = 0.38, P = 0.0103), ejection fraction (r = -0.39, P = 0.009), and serum N-terminal pro-brain natriuretic peptide levels (r = 0.48, P = 0.001), whereas the presence and segmental extent of LGE correlated only with LVEDD. In the 1-year follow-up cohort, the optimal cutoffs of T1native for predicting LV reverse remodeling (LVRR) and combined cardiac events (cardiac death, ventricular tachycardia/fibrillation, heart failure hospitalization) were 1366 ms and 1377 ms, respectively. In multivariate analysis, T1native < 1366 ms and T1native > 1377 ms remained significant predictors of LVRR (odds ratio, 11.3) and cardiac events (hazard ratio, 15.3), respectively, whereas the presence and segmental extent of LGE did not.T1native in patients with NIDCM experiencing recent HF may offer a promising strategy for assessing LV myocardial characteristics and predicting 1-year LVRR and cardiac events.

Incidence and factor analysis of laryngohyoid fractures in hanging individuals-computed tomography study.

OBJECTIVES: Although laryngohyoid fracture indicates the applied neck pressure and is an important finding in hanging individuals, the reported rate varies widely and its true incidence remains controversial. We used computed tomography (CT) studies to investigate the incidence of laryngohyoid fracture in hanging individuals and identify factors contributing to such fractures. METHODS: Considered for inclusion in this study were 107 attempted or successful hanging individuals subjected to CT studies between 2005 and 2019. After excluding 19 whose images were inadequate for evaluation, 88 subjects were included. Body suspension was complete in 20, partial in 49, and unknown in 19; 54 (61.4%) individuals died. Two radiologists performed image analysis and recorded the presence and site of laryngohyoid fractures. Multiple logistic regression analysis was used for factor analysis of laryngohyoid fractures; it included the gender, the age (< or ≧ 40 years), the type of suspension (complete or incomplete), and the outcome (death or survival). RESULTS: Of the 88 subjects, 35 (39.8%) presented with laryngohyoid fractures on CT images; the superior horn of the thyroid cartilage was fractured in 32 (91.4%) of the 35. Age was the only factor significantly related to laryngohyoid fracture (odds ratio = 2.85, 95% confidence interval = 1.08-7.52). CONCLUSIONS: In hanging individuals, the incidence of laryngohyoid fracture on CT images was 39.8%. The superior horn of the thyroid cartilage was the most frequent fracture site. KEY POINTS: • The incidence of laryngohyoid fracture on CT images of hanging individuals was almost 40%; the superior horn of the thyroid cartilage was the most frequent fracture site. • In older hanging individuals, attention must be paid to laryngohyoid fractures on CT images.

Clinical implications of 18F-sodium fluoride uptake in subclinical aortic valve calcification: Its relation to coronary atherosclerosis and its predictive value.

BACKGROUND: Uptake of 18F-sodium fluoride (18F-NaF) on positron emission tomography (PET) reflects active calcification. Application of this technique in the early phase of aortic valve calcification (AVC) is of clinical interest. We investigated clinical implications of 18F-NaF uptake in subclinical AVC evaluated simultaneously with coronary atherosclerosis, and the utility of 18F-NaF uptake in predicting AVC progression. METHODS: We studied 25 patients with subclinical AVC and coronary plaques detected on computed tomography (CT) who underwent 18F-NaF PET/CT. AVC score, volume, mean density, and the presence of high-risk coronary plaque were evaluated on CT in each patient. Focal 18F-NaF uptake in AVC and in coronary plaques was quantified with the maximum tissue-to-background ratio (TBRmax). RESULTS: There were positive correlations between AVC TBRmax (A-TBRmax) and AVC parameters on CT. The 14 patients with high-risk coronary plaque had significantly higher A-TBRmax than those without such plaque (1.60 ± 0.18 vs 1.42 ± 0.13, respectively; P = 0.012). A-TBRmax positively correlated with maximum TBRmax of coronary plaque per patient (r = 0.55, P = 0.0043). In the 11 patients who underwent follow-up CT scan, A-TBRmax positively correlated with subsequent increase in AVC score (r = 0.74, P = 0.0091). CONCLUSION: Our 18F-NaF PET- and CT-based data indicate relationships between calcification activity in subclinical AVC and characteristics of coronary atherosclerosis. 18F-NaF PET may provide new information regarding molecular conditions and future progression of subclinical AVC.

Accuracy of thin-slice model-based iterative reconstruction designed for brain CT to diagnose acute ischemic stroke in the middle cerebral artery territory: a multicenter study.

PURPOSE: Model-based iterative reconstruction (MBIR) yields higher spatial resolution and a lower image noise than conventional reconstruction methods. We hypothesized that thin-slice MBIR designed for brain CT could improve the detectability of acute ischemic stroke in the middle cerebral artery (MCA) territory. METHODS: Included were 41 patients with acute ischemic stroke in the MCA territory; they were seen at 4 medical centers. The controls were 39 subjects without acute stroke. Images were reconstructed with hybrid IR and with MBIR designed for brain CT at slice thickness of 2 mm. We measured the image noise in the ventricle and compared the contrast-to-noise ratio (CNR) in the ischemic lesion. We analyzed the ability of reconstructed images to detect ischemic lesions using receiver operating characteristics (ROC) analysis; 8 observers read the routine clinical hybrid IR with 5 mm-thick images, while referring to 2 mm-thick hybrid IR images or MBIR images. RESULTS: The image noise was significantly lower on MBIR- than hybrid IR images (1.2 vs. 3.4, p < 0.001). The CNR was significantly higher with MBIR than hybrid IR (6.3 vs. 1.6, p < 0.001). The mean area under the ROC curve was also significantly higher on hybrid IR plus MBIR than hybrid IR (0.55 vs. 0.48, p < 0.036). Sensitivity, specificity, and accuracy were 41.2%, 88.8%, and 65.7%, respectively, for hybrid IR; they were 58.8%, 86.1%, and 72.9%, respectively, for hybrid IR plus MBIR. CONCLUSION: The additional thin-slice MBIR designed for brain CT may improve the detection of acute MCA stroke.

Advanced CT techniques for assessing hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) is the sixth-most common cancer in the world, and hepatic dynamic CT studies are routinely performed for its evaluation. Ongoing studies are examining advanced imaging techniques that may yield better findings than are obtained with conventional hepatic dynamic CT scanning. Dual-energy CT-, perfusion CT-, and artificial intelligence-based methods can be used for the precise characterization of liver tumors, the quantification of treatment responses, and for predicting the overall survival rate of patients. In this review, the advantages and disadvantages of conventional hepatic dynamic CT imaging are reviewed and the general principles of dual-energy- and perfusion CT, and the clinical applications and limitations of these technologies are discussed with respect to HCC. Finally, we address the utility of artificial intelligence-based methods for diagnosing HCC.

Relationship between coronary arterial 18F-sodium fluoride uptake and epicardial adipose tissue analyzed using computed tomography.

PURPOSE: 18F-Sodium fluoride (18F-NaF) positron emission tomography (PET) has the potential to detect high-risk coronary plaques. Epicardial adipose tissue (EAT) reportedly correlates with coronary atherosclerosis progression. We evaluated the relationship between coronary arterial 18F-NaF uptake and EAT findings using computed tomography (CT). METHODS: We studied 40 patients with ≥ 1 coronary plaque detected on cardiac CT who underwent 18F-NaF PET/CT. EAT volume was measured using CT and indexed to body surface area in each patient. Each plaque was evaluated for CT-based luminal stenosis and high-risk features. The mean EAT density surrounding each plaque was calculated as perilesional EAT density (PLED) using non-contrast CT images. Focal 18F-NaF uptake in each plaque was quantified using the maximum tissue-to-background ratio (TBRmax). RESULTS: EAT volume index was similar between patients with TBRmax ≥ 1.28 (previously reported optimal cutoff to predict coronary events) and those with lower TBRmax, but patients with TBRmax ≥ 1.28 showed higher maximum PLED per patient (- 86 ± 12 Hounsfield units (HU) versus - 98 ± 11 HU, P = 0.0044). In the lesion-based analysis (n = 92), PLED was positively correlated with TBRmax, and the optimal PLED cutoff to identify TBRmax ≥ 1.28 was - 97 HU. On multivariate analysis adjusted for lesion location, obstructive stenosis, and high-risk plaque on CT, PLED ≥ - 97 HU remained a significant predictor of TBRmax ≥ 1.28. CONCLUSIONS: Increased PLED was associated with significant coronary arterial 18F-NaF uptake. Step-by-step analyses of EAT density on CT and coronary arterial 18F-NaF uptake on PET may offer novel strategies for risk prediction in coronary artery disease.

Possibility of Deep Learning in Medical Imaging Focusing Improvement of Computed Tomography Image Quality.

Deep learning (DL), part of a broader family of machine learning methods, is based on learning data representations rather than task-specific algorithms. Deep learning can be used to improve the image quality of clinical scans with image noise reduction. We review the ability of DL to reduce the image noise, present the advantages and disadvantages of computed tomography image reconstruction, and examine the potential value of new DL-based computed tomography image reconstruction.

Individual Optimization of Contrast Media Injection Protocol at Hepatic Dynamic Computed Tomography Using Patient-Specific Contrast Enhancement Optimizer.

OBJECTIVE: We developed a patient-specific contrast enhancement optimizer (p-COP) that can exploratorily calculate the contrast injection protocol required to obtain optimal enhancement at target organs using a computer simulator. Appropriate contrast media dose calculated by the p-COP may minimize interpatient enhancement variability. Our study sought to investigate the clinical utility of p-COP in hepatic dynamic computed tomography (CT). METHODS: One hundred thirty patients (74 men, 56 women; median age, 65 years) undergoing hepatic dynamic CT were randomly assigned to 1 of 2 contrast media injection protocols using a random number table. Group A (n = 65) was injected with a p-COP-determined iodine dose (developed by Higaki and Awai, Hiroshima University, Japan). In group B (n = 65), a standard protocol was used. The variability of measured CT number (SD) between the 2 groups of aortic and hepatic enhancement was compared using the F test. In the equivalence test, the equivalence margins for aortic and hepatic enhancement were set at 50 and 10 Hounsfield units (HU), respectively. The rate of patients with an acceptable aortic enhancement (250-350 HU) for the diagnosis of hypervascular liver tumors was compared using the χ test. RESULTS: The mean ± SD values of aortic and hepatic enhancement were 311.0 ± 39.9 versus 318.7 ± 56.5 and 59.0 ± 11.5 versus 58.6 ± 11.8 HU in groups A and B, respectively. Although the SD for aortic enhancement was significantly lower in group A (P = 0.006), the SD for hepatic enhancement was not significantly different (P = 0.871). The 95% confidence interval for the difference in aortic and hepatic enhancement between the 2 groups was within the range of the equivalence margins. The number of patients with acceptable aortic enhancement was significantly greater in group A than in group B (P < 0.01). CONCLUSIONS: The p-COP software reduced interpatient variability in aortic enhancement and obtained acceptable aortic enhancement at a significantly higher rate compared with the standard injection protocol for hepatic dynamic CT.

Minimizing individual variations in arterial enhancement on coronary CT angiographs using "contrast enhancement optimizer": a prospective randomized single-center study.

OBJECTIVES: To investigate the clinical utility of our newly developed contrast enhancement optimizer (CEO) software for coronary CT angiography (CCTA). METHODS: We randomly assigned 295 patients (168 males, 127 females, median age 71 years) undergoing CCTA to one of two contrast media injection protocols. Group A (n = 150) was injected with a CEO-selected iodine dose based on patient factors. In group B (n = 145), we used our standard protocol (245 mg I/kg). We recorded the CT number in the ascending aorta and determined whether the CT number was equivalent in groups A and B. For the equivalence test, we adopted 75 Hounsfield units (HU) as the equivalence margin. The standard deviation in the CT number and the rate of patients with an acceptable CT number were compared using the F test and the chi-square test, respectively. RESULTS: The iodine dose in group A was significantly smaller than that in group B (235.7 vs. 253.6 mg I/kg, p < 0.001). The CT number of the ascending aorta was 428.6 ± 55.5 HU in group A and 436.1 ± 68.7 HU in group B; the 95% confidence interval for the difference between the groups was -4.3 HU to 16.9 HU and within the range of the predetermined equivalence margins. In group A, the variance was significantly smaller than that in group B (p = 0.009). The number of patients with an acceptable CT number was significantly higher in group A than in group B (84.7% vs. 71.7%, p = 0.007). CONCLUSIONS: The use of our CEO for CCTA studies yielded optimal aortic contrast enhancement in significantly more patients than the standard protocol based on the body weight. KEY POINTS: • With our contrast enhancement optimizer (CEO) software, optimal and stable aortic enhancement can be obtained on coronary CT angiography scans irrespective of patient factors. • Management of contrast media becomes more appropriate by the CEO software. • The CEO software can control contrast enhancement at different tube voltage levels.

Neointimal formation after carotid artery stenting: phantom and clinical evaluation of model-based iterative reconstruction (MBIR).

OBJECTIVES: The objective of this study was to investigate the usefulness of model-based iterative reconstruction (IR) for detecting neointimal formations after carotid artery stenting. METHODS: In a cervical phantom harbouring carotid artery stents, we placed simulated neointimal formations measuring 0.40, 0.60, 0.80 and 1.00 mm along the stent wall. The thickness of in-stent neointimal formations was measured on images reconstructed with filtered-back projection (FBP), hybrid IR (AIDR 3D), and model-based IR (FIRST). The clinical study included 43 patients with carotid stents. Cervical computed tomography (CT) images obtained on a 320-slice scanner were reconstructed with AIDR 3D and FIRST. Five blinded observers visually graded the likelihood of neointimal formations on AIDR 3D and AIDR 3D plus FIRST images. Carotid ultrasound images were the reference standard. We analysed results of visual grading by using a Jack-knife type receiver observer characteristics analysis software. RESULTS: In the phantom study, the difference between the measured and the true diameter of the neointimal formations was smaller on FIRST than FBP or AIDR 3D images. In the clinical study, the sensitivity, specificity, positive predictive value, negative predictive value and accuracy of AIDR 3D were 58%, 88%, 83%, 67% and 73%, respectively. For AIDR 3D plus FIRST images they were 84%, 78%, 80%, 82% and 81%, respectively. The mean area under the curve was significantly higher on AIDR 3D plus FIRST than AIDR 3D images (0.82 vs 0.72; p < 0.01). CONCLUSIONS: The model-based IR algorithm helped to improve diagnostic performance for the detection of neointimal formations after carotid artery stenting. KEY POINTS: • Neointimal formations can be visualised more accurately with model-based IR. • Model-based IR improves the detection of neointimal formations after carotid artery stenting. • Model-based IR is suitable for follow up after carotid artery stenting.

Deep learning-based image restoration algorithm for coronary CT angiography.

OBJECTIVES: The purpose of this study was to compare the image quality of coronary computed tomography angiography (CTA) subjected to deep learning-based image restoration (DLR) method with images subjected to hybrid iterative reconstruction (IR). METHODS: We enrolled 30 patients (22 men, 8 women) who underwent coronary CTA on a 320-slice CT scanner. The images were reconstructed with hybrid IR and with DLR. The image noise in the ascending aorta, left atrium, and septal wall of the ventricle was measured on all images and the contrast-to-noise ratio (CNR) in the proximal coronary arteries was calculated. We also generated CT attenuation profiles across the proximal coronary arteries and measured the width of the edge rise distance (ERD) and the edge rise slope (ERS). Two observers visually evaluated the overall image quality using a 4-point scale (1 = poor, 4 = excellent). RESULTS: On DLR images, the mean image noise was lower than that on hybrid IR images (18.5 ± 2.8 HU vs. 23.0 ± 4.6 HU, p < 0.01) and the CNR was significantly higher (p < 0.01). The mean ERD was significantly shorter on DLR than on hybrid IR images, whereas the mean ERS was steeper on DLR than on hybrid IR images. The mean image quality score for hybrid IR and DLR images was 2.96 and 3.58, respectively (p < 0.01). CONCLUSIONS: DLR reduces the image noise and improves the image quality at coronary CTA. KEY POINTS: • Deep learning-based image restoration is a new technique that employs the deep convolutional neural network for image quality improvement. • Deep learning-based restoration reduces the image noise and improves image quality at coronary CT angiography. • This method may allow for a reduction in radiation exposure.

Contrast Material Injection Protocol With the Dose Determined According to Lean Body Weight at Hepatic Dynamic Computed Tomography: Comparison Among Patients With Different Body Mass Indices.

OBJECTIVE: The objective of this study was to compare enhancement of the aorta and liver on hepatic dynamic computed tomography scans acquired with contrast material doses based on the lean body weight (LBW) or the total body weight (TBW). METHODS: We randomly divided 529 patients (279 men, 250 women; median age, 66 years) scheduled for hepatic dynamic computed tomography into 2 groups. The LBW patients (n = 278) were injected with 679 mg iodine/kg (men) or 762 mg iodine/kg (women). The TBW group (n = 251) was injected with 600 mg iodine/kg TBW. Each group was subdivided into the 3 classes based on the body mass index (BMI; low, normal, high). Aortic enhancement during the hepatic arterial phase and hepatic enhancement during the portal venous phase was compared. The aortic and hepatic equivalence margins were 100 and 20 Hounsfield units, respectively. RESULTS: Comparison of the median iodine dose in patients with a normal or high BMI showed that it was significantly lower under the LBW protocol than the TBW protocol (558.2 and 507.0 mg iodine/kg, P < 0.001, respectively). However, in patients with a low BMI, the LBW protocol delivered a significantly higher dose than the TBW protocol (620.7 vs 600.0 mg iodine/kg, P < 0.001). The 95% confidence interval for the difference in aortic and hepatic enhancement between the 2 protocols was within the range of the predetermined equivalence margins in all BMI subgroups. CONCLUSIONS: Contrast enhancement was equivalent under both protocols. The LBW protocol can avoid iodine overdosing, especially in patients with a high BMI.