Insufficient Image Quality Research Articles

Atrial and ventricular mitral regurgitation in the general population

Abstract Background Functional mitral valve regurgitation (FMR) is the result of two concurrent pathomechanisms: (1) the dilatation of mitral valve annulus (atrial) or (2) remodelling of the left ventricle (ventricular) in the absence of relevant leaflet degeneration¹. Even though FMR has been reported to be the most common cause of relevant MR², there is no general accepted definition of the two subtypes, nor are there population-based data regarding their prevalence, distribution and risk profiles¹. Purpose We aimed to evaluate the prevalence of FMR and to provide an in-depth characterisation of the subtypes aFMR and vFMR in the general population. Methods The study included the first 10,000 participants of the population-based Hamburg City Health Study enrolled between 2016 and 2019. Of these, 8.543 echocardiographic samples were available for analysis. Mitral valve was assessed by two independent investigators. Isolated degenerative MRs as determined by valve mobility (prolapse or restrictive) were excluded. FMR was classified as ventricular when fulfilling one of the following criteria: history of myocardial infarction or coronary artery disease, left ventricular ejection fraction (LVEF) ≤ 40%, restrictive leaflet, tenting area ≥ 150 mm² or coaptation depth ≥ 10 mm. All FMR with LVEF > 55% were classified as atrial. Results After excluding the samples with an insufficient image quality or missing data and re-assessment by the second investigator, 216 of 282 participants with a relevant MR (i.e. moderate or severe) remained (2,6% of the total cohort, see figure 1). In the next step, 153 participants were classified as pure FMR (68.9%), the remaining were mainly functional with minor degenerative features. A total of 31 (14.4%) MRs were classified as vFMR, the remaining 185 participants had an aFMR. Female gender was significantly more common in the aFMR cohort. The remaining demographic variables (age, body mass index) as well as established cardiovascular risk factors (i.e., diabetes, smoking, hypertension) were comparable between the vFMR and aFMR cohort. However, vFMR patients had significantly higher levels of NT-proBNP or NHYA state and presented more frequently with an atrial fibrillation (see figure 2). LVEF and LV stroke volume, TAPSE, left atrial ejection fraction and E/A ratio as a parameter of diastolic dysfunction were lower in the vFMR cohort whereas tenting area and left atrial endsystolic volume was significantly increased. Conclusion In our population-based study, aFMR was found to be the predominant subtype of FMR. The established cardiovascular risk factors were comparable in the vFMR and aFMR cohort. Echocardiographic variables differed significantly and indicate more advanced heart failure and LV remodelling in the vFMR cohort. Further categorization of FMR, in particular regarding the prognostic impact and long-term outcomes, is strongly warranted to allow a clinically relevant definition of aFMR and vFMR.Figure 1:Overview of sample sizeFigure 2:Baseline characteristics

Comparison of dimensions and functional features of mitral and tricuspid annuli in the same healthy adults: insights from the three-dimensional speckle-tracking echocardiographic MAGYAR-Healthy Study.

Evaluation of mitral (MA) and tricuspid annuli (TA) in the same healthy subject in a non-invasive way in real-life clinical settings makes an opportunity to compare their dimensions and derived functional properties. The purpose of the present cohort study was to investigate whether there are any differences in the three-dimensional speckle-tracking echocardiography- (3DSTE-) measured size and derived functional characteristics of the MA and TA in the same healthy adults. The study comprised 248 healthy adults, in which 3DSTE was performed to determine MA and TA dimensions and functional properties. Due to insufficient image quality, 89 cases were excluded, therefore the remaining population consisted of 159 subjects (age: 35.6±12.9 years, 76 males). Subjects were enrolled on a voluntary basis consecutively between January 2011 and November 2017 in the outpatient clinic of the tertiary cardiology center at the Department of Medicine, University of Szeged, Hungary. Data were analyzed by Student's t-test, analysis of variance (ANOVA) test, Fischer's exact test, Pearsons' correlations, interclass correlations and Bland-Altman tests. Same-side MA/TA end-diastolic annular dilation is associated with simultaneous MA/TA end-systolic dilation and vice versa. MA dilation in end-diastole and end-systole results in MA functional improvement/deterioration. Dilation of end-diastolic TA dimensions does not obviously entail differences in TA function. However, similar to MA, more dilated TA in end-systole is associated with impaired TA function. Dilated MA dimensions (end-diastolic MA area: 4.31±0.62 vs. 10.89±1.18 cm2, P<0.05) are not obviously associated with dilated end-diastolic TA dimensions (area: 7.05±1.42 vs. 7.81±1.48 cm2, P=ns) and functional improvement/impairment (fractional area change: 27.5%±10.8% vs. 25.2%±10.6%, P=ns). Dilation of MA and TA is associated with different contralateral responses in morphology and function.

The Role of Transformer-cGANs Fusion in Digital Marketing

Advertising images play a key role in digital marketing and are an indispensable tool to attract users, convey messages and promote products. However, current advertising image generation faces many challenges, including insufficient image quality and diversity, and high production costs. In this context, the rise of deep learning technology provides a powerful tool to solve these problems. Deep learning is increasingly used in the field of computer vision, bringing new hope for advertising image generation. This article introduces an innovative Transformer-cGANs model that combines the sequence modeling capabilities of Transformer with the image generation technology of Generative Adversarial Networks (GANs). This model shows excellent performance in advertising image generation tasks in digital marketing. Through extensive experimental verification, our model successfully overcomes the insufficient image quality and diversity issues present in traditional methods.

Improving Rebar Twist Prediction Exploiting Unified-Channel Attention-Based Image Restoration and Regression Techniques.

Recent research has made significant progress in automated unmanned systems utilizing Artificial Intelligence (AI)-based image processing to optimize the rebar manufacturing process and minimize defects such as twisting during production. Despite various studies, including those employing data augmentation through Generative Adversarial Networks (GANs), the performance of rebar twist prediction has been limited due to image quality degradation caused by environmental noise, such as insufficient image quality and inconsistent lighting conditions in rebar processing environments. To address these challenges, we propose a novel approach for real-time rebar twist prediction in manufacturing processes. Our method involves restoring low-quality grayscale images to high resolution and employing an object detection model to identify and track rebar endpoints. We then apply regression analysis to the coordinates obtained from the bounding boxes to estimate the error rate of the rebar endpoint positions, thereby determining the occurrence of twisting. To achieve this, we first developed a Unified-Channel Attention (UCA) module that is robust to changes in intensity and contrast for grayscale images. The UCA can be integrated into image restoration models to more accurately detect rebar endpoint characteristics in object detection models. Furthermore, we introduce a method for predicting the future positions of rebar endpoints using various linear and non-linear regression models. The predicted positions are used to calculate the error rate in rebar endpoint locations, determined by the distance between the actual and predicted positions, which is then used to classify the presence of rebar twisting. Our experimental results demonstrate that integrating the UCA module with our image restoration model significantly improved existing models in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) metrics. Moreover, employing regression models to predict future rebar endpoint positions enhances the F1 score for twist prediction. As a result, our approach offers a practical solution for rapid defect detection in rebar manufacturing processes.

RGBD Synergetic Model for Image Enhancement in Animation Advertisements

This paper proposes a depth image symbiosis model to solve the problem of insufficient depth image quality in animated advertising. The model uses image surface information and image edge cues as the main guidance information to obtain image symbiosis information. Research data show that the model designed in this paper performs well in convergence, and enters a stable convergence process when the number of iterations is less than 5. Its PSNR data curve has the highest position and best performance, while a composite model structure has been adopted. Compared with the unitary model, the PSNR of this model reaches 41 dB when the number of iterations reaches 5, and the convergence effect of the three-step training is also better. Finally, in practical applications, the average PSNR value of the model mentioned in this article is the highest, 37.1 dB. From this comprehensive perspective, the depth image enhancement model in this study has better comprehensive performance and can provide better image enhancement effects for animation advertising depth images.

Improving the Classification of Unexposed Potsherd Cavities by Means of Preprocessing

The preparation of raw images for subsequent analysis, known as image preprocessing, is a crucial step that can boost the performance of an image classification model. Although deep learning has succeeded in image classification without handcrafted features, certain studies underscore the continued significance of image preprocessing for enhanced performance during the training process. Nonetheless, this task is often demanding and requires high-quality images to effectively train a classification model. The quality of training images, along with other factors, impacts the classification model’s performance and insufficient image quality can lead to suboptimal classification performance. On the other hand, achieving high-quality training images requires effective image preprocessing techniques. In this study, we perform exploratory experiments aimed at improving a classification model of unexposed potsherd cavities images via image preprocessing pipelines. These pipelines are evaluated on two distinct image sets: a laboratory-made, experimental image set that contains archaeological images with controlled lighting and background conditions, and a Jōmon–Yayoi image set that contains images of real-world potteries from the Jōmon period through the Yayoi period with varying conditions. The best accuracy performances obtained on the experimental images and the more challenging Jōmon–Yayoi images are 90.48% and 78.13%, respectively. The comprehensive analysis and experimentation conducted in this study demonstrate a noteworthy enhancement in performance metrics compared to the established baseline benchmark.

Intra- and interobserver reliability in measuring aortic root rotation with cardiac magnetic resonance imaging.

The rotational position of the aortic root (AoR) is of substantial clinical interest as it has been associated with severe aortic complications, such as aortic dissections. We described a cardiac magnetic resonance (CMR)-based method for measuring AoR rotation and evaluated the reliability of measurements. CMR was used for measuring AoR rotation in 50 consecutive healthy subjects. Intra- and interobserver reliability were assessed by comparing repeated measurements by the same analyst and by three independent analysts. The angles of the non-coronary sinus (NCS), right coronary sinus (RCS), and left coronary sinus (LCS) were measured relative to the interatrial septum (IAS). The angle between IAS and posterior atrial wall was measured to examine the reliability of the IAS as a key anatomical landmark. Intra- and interobserver agreement were determined using intraclass correlation coefficients (ICCs). Images of 47/50 (94%) subjects were analyzed; three were excluded due to insufficient image quality. The mean ± standard deviation (SD) AoR rotation angles of NCS, RCS, and LCS were 25.9°±12.9°, 37.5°±15.2°, and 97.0°±13.1°, respectively. For measurements of AoR rotation, both intraobserver [NCS: ICC =0.94, 95% confidence interval (CI): 0.88-0.96; RCS: ICC =0.93, 95% CI: 0.87-0.96; LCS: ICC =0.91, 95% CI: 0.84-0.95] and interobserver agreement (NCS: ICC =0.90, 95% CI: 0.84-0.94; RCS: ICC =0.86, 95% CI: 0.77-0.92; LCS: ICC =0.87, 95% CI: 0.80-0.92) were excellent. The IAS angle was 79.2°±8.9°; its intraobserver agreement was somewhat higher (ICC =0.94, 95% CI: 0.88-0.96) than the interobserver agreement (ICC =0.76, 95% CI: 0.63-0.85). The present study in healthy subjects reports a CMR-based approach for measuring AoR rotation. CMR allows to quantify AoR rotation with excellent intra- and interobserver agreement.

Current capabilities of anterior segment optical coherence tomography

Optical coherence tomography of the anterior segment of the eye (AS-OCT) is a non-invasive method based on the principles of optical reflectometry (measurement of the degree of backscattering of light passing through transparent or translucent media). Limitations of the first devices of this type were associated with insufficient image quality of the details of the anterior chamber angle and the posterior parts of the lens, primarily due to the "working" level of the scanning wavelength (within 800 μm). Fundamentally new possibilities in the structural and functional assessment of the anterior segment of the eye are associated with the introduction into clinical practice of swept-source AS-OCT device - the CASIA2 anterior optical coherence tomograph (Tomey Corporation, Japan). Its high scanning speed (50 000 A-scans per second) with a wavelength of 1310 μm allows high-quality visualization by building a scan at a depth of 13 mm. The previous model (CASIA SS-1000, Tomey Corporation, Japan) supported scan depth of only 6 mm. This review summarizes the results of research on the clinical use of CASIA2 tomograph.

Aortic valve sclerosis, stenosis, and prognosis in the general population

Abstract Background The presence of aortic valve (AV) sclerosis and non-severe aortic stenosis has been a suggested marker of cardiovascular risk, but whether AV sclerosis and non-severe AV stenosis is independently associated with increased cardiovascular risk or a marker of coexistent risk factors remains to be clarified. Objective We examined clinical characteristics and prognosis associated with the presence of AV sclerosis and stenosis in the general population. Methods A total of 4,466 participants from a prospective cohort study of the general population were included. Patients with insufficient image quality (n=119) and known heart failure (n=130) were excluded, resulting in a population of 4,217 participants. Evaluation of AV sclerosis/stenosis was based on AV Doppler measurements, i.e., peak and mean AV velocity, and peak and mean AV gradient. The primary composite outcome was defined as major adverse cardiovascular events (MACE), including incident heart failure, myocardial infarction, or cardiovascular death. Multivariable Cox regression models were used to assess the association between AV sclerosis/stenosis by AV Doppler measurements and risk of the outcome. Results Mean age was 56.8 ± 17.5 years, and 56% were women. In total, 18.7% were smokers, 51% had hypertension (HTN), 5% had diabetes (DM), and 7% had ischemic heart disease (IHD). Mean left ventricular ejection fraction (LVEF) was 56.4 ± 6.4% with a median left ventricular mass index (LVMi) of 82.6 g/m2 (IQR 71.1; 96.8) and a mean AVA of 2.72 ± 0.72 cm2. Median AV peak velocity was 1.3 m/s (IQR 1.1; 1.4) and median AV peak gradient was 6.0 mmHg (IQR 5.0; 8.0). Median AV mean velocity was 0.9 m/s (IQR 0.8; 1.0) and median AV mean gradient was 3.0 mmHg (IQR 3.0; 4.0). During a median follow-up of 3.5 years (IQR 2.6; 4.3), 190 patients (4.5%) reached the composite outcome. The incidence rate per 1,000 person years in the fourth quartile of AV peak velocity (AV peak velocity &amp;gt;1.4 m/s) was more than double that of the first quartile (AV peak velocity ≤1.1 m/s) (26.6 [95% CI 21.1 – 33.6] versus 12.9 [95% CI 9.9 – 16.9], p&amp;lt;0.001). In univariable analysis, all AV measurements of sclerosis/stenosis were associated with the composite outcome (p&amp;lt;0.001 for all) (Figure 1). All AV measurements remained independent predictors of the outcome after adjustment for demographic, clinical and echocardiographic risk factors (age, gender, smoking, HTN, DM, IHD, eGFR, LVMi, LAVi, LVEF and E/e’); AV peak PG (HR 1.19 per 5 mmHg increase, 95% CI 1.06 – 1.33, p=0.002), AV peak velocity (HR 1.05 per 0.1 m/s increase [95% CI 1.01 – 1.20], p = 0.013), AV mean PG (HR 1.30 per 5 mmHg increase [95% CI 1.09 – 1.56], p=0.002), AV mean velocity (HR 1.08 per 0.1 m/s increase [95% CI 1.02 – 1.23], p=0.005). Conclusion A greater degree of AV sclerosis/stenosis was independently associated with a higher risk of MACE.Figure 1:Restricted cubic spline model

Safety and procedural success of left appendage closure interventions with transesophageal echocardiography microprobes

Abstract Introduction Transesophageal echocardiography (TOE) is the most frequently used imaging technique for intraprocedural guidance in left appendage closure interventions. However, it usually requires the use general anaesthesia and implies an increased risk of bleeding, and of esophageal and upper airway tract lesions. Intracardiac echocardiography and TOE with microprobes (micro-TOE) have been proposed as alternative techniques for procedural imaging. Purpose The aim of this study is to compare the results of LAAC interventions in terms of success, safety and hospital stay depending of the imaging technique used for procedural guidance. Methods This is a single-center, prospective study, which included patients in which LAAC was performed between the years 2009 and 2021. In the group with conventional TOE guidance, procedure was performed under general anesthesia using the probe Philips X8-2t. In the group with micro-TOE guidance, the procedure was performed under conscious sedation with midazolam and fentanyl with Philips S8-3t microprobe. Image quality was assessed independently by two echocardiographers. Basal characteristics, procedure length, complications and hospital length of stay were analyzed. Results Between the years 2009 and 2021, 511 LAAC procedures were performed, of which 357 were guided with TOE and 154 with microTOE. Mean age was of 77.2 ± 8 years, and 63.6% of patients were male. Successful device implantation was achieved in all patients of both groups. In 10 patients in which the scheduled imaging technique was micro-TOE were eventually perfomed with conventional TOE probe due to insufficient image quality. Procedural length was significantly reduced in patients in the micro-TOE (73 vs 51 min, p &amp;lt; 0.001). There were no significant differences in major complications between group. In-hospital length of stay was significantly reduced in patient with micro-TOE guidance (1.15 ± 1.9 days versus 2.2 ± 2.3 days, p &amp;lt; 0.001). Conclusions Transesophageal echocardiography with microprobes is a safe and efective imaging technique for procedural guidance for LAAC interventions, with procedural success and complication rates similar to conventional TOE. Furthermore, its use reduces in-hospital length of stay, enabling outpatient procedures.Table 1.Comparison between groups

Organ-based tube current modulation versus spectral shaping via tin prefiltration: What does bismuth breast shielding add in low-dose lung CT?

PurposeSince organ-based tube current modulation (OBTCM) and tin prefiltration are limited on their own in lowering the dose of lung CT examinations, this experimental study was designed to investigate whether combinations with anterior patient shielding can increase the dose reduction potential. Material and methodsThree pairs of scan protocols without/with breast shield (P1/P2: standard 120kVp, P3/P4: OBTCM at 100 kVp, P5/P6: Sn 100 kVp) were employed for radiation exposure and image quality comparisons on an anthropomorphic Alderson-Rando phantom. Equivalent doses were measured in eleven sites via thermoluminescent dosimetry and the effective dose was obtained by summation of the weighted organ doses. Dose-weighted contrast-to-noise ratios (CNRD) were calculated and four radiologists independently assessed the quality of images generated with each protocol. ResultsWhile no significant difference was determined between standard and OBTCM protocols regardless of breast shield (p ≥ 0.068), equivalent doses with spectral shaping were substantially lower (p ≤ 0.003). The highest effective dose was ascertained for standard scans (P1/P2: 7.3/6.8 mSv) with a dose reduction of 8.0 % via breast shielding. The use of a bismuth shield was more beneficial in OBTCM (P3/P4: 6.6/5.3 mSv) and spectral shaping (P5/P6: 0.7/0.6 mSv), reducing the effective dose by 19.8 % and 13.9 %, respectively. Subjective assessment favoured standard protocol P1 over tin prefiltration low-dose scans (p ≤ 0.032), however, no scan protocol entailed diagnostically insufficient image quality. ConclusionsWhereas breast shielding is particularly beneficial in combination with OBTCM, spectral shaping via tin prefiltration facilitates the most pronounced dose reduction in lung CT imaging with acceptable image quality.

Automated F18-FDG PET/CT image quality assessment using deep neural networks on a latest 6-ring digital detector system

To evaluate whether a machine learning classifier can evaluate image quality of maximum intensity projection (MIP) images from F18-FDG-PET scans. A total of 400 MIP images from F18-FDG-PET with simulated decreasing acquisition time (120 s, 90 s, 60 s, 30 s and 15 s per bed-position) using block sequential regularized expectation maximization (BSREM) with a beta-value of 450 and 600 were created. A machine learning classifier was fed with 283 images rated “sufficient image quality” and 117 images rated “insufficient image quality”. The classification performance of the machine learning classifier was assessed by calculating sensitivity, specificity, and area under the receiver operating characteristics curve (AUC) using reader-based classification as the target. Classification performance of the machine learning classifier was AUC 0.978 for BSREM beta 450 and 0.967 for BSREM beta 600. The algorithm showed a sensitivity of 89% and 94% and a specificity of 94% and 94% for the reconstruction BSREM 450 and 600, respectively. Automated assessment of image quality from F18-FDG-PET images using a machine learning classifier provides equivalent performance to manual assessment by experienced radiologists.

Left ventricular longitudinal strain: a useful tool in the evaluation of patients with Takotsubo syndrome

Abstract Funding Acknowledgements Type of funding sources: None. Introduction In contrast with the usual type of Takotsubo syndrome (TS), where apical wall motion abnormality occurs, the less common patterns (mid-ventricular, basal, and focal) are often misdiagnosed. Purpose In this retrospective study we examined the role of global (GLS) and territorial longitudinal strain with speckle tracking in the evaluation of patients with TS. Methods Patients with TS were examined at the echocardiography laboratory of our general hospital, from August 2019 until August 2022. Left ventricular ejection fraction (LVEF) and GLS were calculated and territorial longitudinal strain was assessed. Results Out of 21 patients with TS (19 females), of median age 67.5 years (range 33.0 – 87.0 years), 11 (52.4%) had apical distribution, 6 (28.6%) mid-ventricular, 2 (9.5%) basal and 2 (9.5%) focal. Of those, 3 were excluded from the study due to insufficient echocardiographic image quality. Patients with apical pattern had median LVEF 37% (range 23–51%) and GLS −7.6% (−5.8 to −12.5%), patients with mid-ventricular pattern had median LVEF 40% (35–50%) and GLS −14.2% (−9.3 to −16.5%) and patients with basal pattern had median LVEF 52% (47–56%) and GLS −16.5% (−14.6 to −18.4%). In the first case of focal distribution, wall motion abnormality was found at the anterior septal wall, and LVEF was calculated 46% and GLS −12.8%. In the second case, abnormality was found at the inferoposterior wall and LVEF was calculated 56% and GLS −17.3%. Patients with apical TS had median strain of basal territory −11.7%, mid territory −3.7% and apical territory +3.6%. Patients with mid ventricular TS had median strain of basal territory −10.2%, mid territory −6.5% and apical territory −19.7%. Patients with basal TS had median strain of basal territory −4.3%, mid territory −16.4% and apical territory −29.1%. Conclusion GLS and territorial longitudinal strain reflect the reduction in left ventricular systolic function and the distribution of wall motion abnormalities in patients with TS, which further supports the initial diagnosis.

Coronary artery assessment on pre transcatheter aortic valve implantation computed tomography may avoid the need for additional coronary angiography

PurposeThe purpose of this study was to evaluate the percentage of coronary angiography that can be securely avoided by the interpretation of coronary arteries on pre transcatheter aortic valve implantation CT (TAVI-CT), using CT images obtained with deep-learning reconstruction and motion correction algorithms. Material and methodAll consecutive patients who underwent TAVI-CT and coronary angiography, from December 2021 to July 2022 were screened for inclusion in the study. Patients who had previous coronary artery revascularization or who did not undergo TAVI were excluded. All TAVI-CT examinations were obtained using deep-learning reconstruction and motion correction algorithms. On TAVI-CT examinations, quality and stenosis of coronary artery were analyzed retrospectively. When insufficient image quality and/or when diagnosis or doubt of one significant coronary artery stenosis, patients were considered as having possible coronary artery stenosis. The results of coronary angiography were used as the standard of reference for significant CAS. ResultsA total of 206 patients (92 men; mean age, 80.6 years) were included; of these 27/206 (13%) had significant coronary artery stenosis on coronary angiography and were referred for potential revascularization. Sensitivity, specificity, negative predictive value, positive predictive value, and accuracy of TAVI-CT to identify patients requiring coronary artery revascularization was 100% (95% confidence interval [CI]: 87.2–100%), 100% (95% CI: 96.3–100%), 54% (95% CI: 46.6–61.6), 25% (95% CI: 17.0–34.0%) and 60% (95% CI: 53.1–66.9%) respectively. Intra- and inter observer variability was substantial agreement for quality and decision to recommend coronary angiography. Mean reading time was 2 ± 1.2 (standard deviation) min (range: 1–5 min). Overall, TAVI-CT could potentially rule out indication for revascularization for 97 patients (47%). ConclusionAnalysis of coronary artery on TAVI-CT using deep-learning reconstruction and motion correction algorithms can potentially safely avoid coronary angiography in 47% of patients.

3D Synthetic Brain MRI with Compressed Sensing: Can It Be a Promising Way Forward for Daily Neuroimaging?

Synthetic MRI can provide multiple contrast-weighted brain images with high resolution from a single scan via a 3D sequence using an interleaved Look-Locker acquisition sequence with a T2 preparation pulse (3D-QALAS). This study aimed to assess the diagnostic image quality of 3D synthetic MRI using compressed sensing (CS) in clinical practice. We retrospectively reviewed the imaging data of 47 patients who underwent brain MRI, including 3D synthetic MRI using CS in a single session, between December 2020 and February 2021. Two neuroradiologists independently evaluated the overall image quality, anatomic demarcation, and artifacts for synthetic 3D T1-weighted, T2-weighted, FLAIR, phase-sensitive inversion recovery (PSIR), and double inversion recovery images, using a 5-point Likert scale. The interobserver agreement between the two readers was assessed using percent agreement and weighted κ statistics. The overall image quality of 3D synthetic T1WI and PSIR was good to excellent, with easy or excellent anatomic demarcation and mild or no visible artifact. However, other 3D synthetic MRI-derived images showed insufficient image quality and anatomic demarcation with marked CSF pulsation artifacts. In particular, 3D synthetic FLAIR showed high-signal artifacts on the brain surface. 3D synthetic MRI, at its current status, cannot completely replace conventional brain MRI in daily clinical practice. However, 3D synthetic MRI can achieve scan-time reduction using CS and parallel imaging and may be useful for motion-prone or pediatric patients requiring 3D images where time-efficiency is important.

Prognostic role of three-dimensional speckle-tracking echocardiography-derived left ventricular global longitudinal strain in cardiac amyloidosis: Insights from the MAGYAR-Path Study.

Systemic amyloidosis is an uncommon disorder in which amyloid fibrils deposit extracellularly. Three-dimensional speckle-tracking echocardiography (3DSTE) is a novel method able to assess left ventricular (LV) global longitudinal strain (GLS). Our aim was to evaluate the prognostic impact of 3DSTE-derived LV-GLS in patients with cardiac amyloidosis (CA). A total of 35 patients suffering from light-chain (AL) CA or transthyretin (TTR) CA were selected, but 7 patients had to be excluded due to insufficient image quality or were lost for follow-up. With AL-CA 23 cases, while for TTR-CA 5 patients were diagnosed. Complete two-dimensional Doppler and 3DSTE were performed in all subjects. The median follow-up was 201 days (ranging from 36 to 632 days) during which cardiovascular event was detected in 17 CA patients, including 8 cardiac deaths. Six patients were diagnosed with acute heart failure, two patients needed invasive interventions (percutaneous coronary intervention with stent-implantation, implantable cardioverter defibrillator implantation) and in one patient new higher grade atrioventricular block was registered. Using ROC analysis, 3DSTE-derived LV-GLS ≥11.8% (absolute value) was found to be a significant predictor for cardiovascular event-free survival (sensitivity 65%, specificity 64%, area under the curve 0.71, p=.05). Lower LV ejection fraction was confirmed in patients with LV-GLS <11.8% as compared to cases with LV-GLS ≥11.8%. In case of a cardiovascular event, LV-GLS was lower as compared to that of subjects with no events. Multivariable regression analysis confirmed that LV-GLS and LV end-diastolic diameter were independent predictors of cardiovascular survival. 3DSTE-derived LV-GLS is an independent predictor for future cardiovascular events in CA patients.

A one-day focused cardiac ultrasound training has a limited impact on residents' abilities to visually assess right ventricular dimensions and function.

Focused cardiac ultrasound (FCU) is a helpful tool to rapidly identify right ventricular (RV) causes of hemodynamic instability and facilitate the initiation of therapy. The clinical value of existing course models often remains unclear. This study investigated the effects of a one-day FCU training on the visual estimation skills of RV characteristics. Four residents were included as the study group after completing a standardized one-day FCU training. Four gender-matched controls did not take part in the training. All residents graded image quality, RV systolic function, and RV dimensions in a test comprising 35 ultrasound clips. The study and control group did not differ in ICU or ultrasound experience. Overall, training participants were able to distinguish between good and insufficient image quality significantly better than the control group (agreement 80.0%vs 61.4%, p=0.04). The agreement for the estimation of RV function and RV dimensions was not different between the groups (63.2%vs 60.5%, p=0.66 and 64.3%vs 67.1%, p=0.18, respectively). Descriptively, only small differences were found between the groups for the estimation of RV function and RV dimensions in subgroups of patients with normal versus reduced systolic RV function or normal versus enlarged RV dimensions, respectively. Both groups struggled in identifying RV enlargement (34.6%vs 46.2%). In this study, a single one-day FCU training had no impact on residents' skills to visually assess systolic RV function or RV dimensions. Improvements of current training modalities or continuous teaching models are needed to optimize residency programs and patient care.

Four-dimensional fetal cardiac imaging in a cohort of fetuses with suspected congenital heart disease.

Fetal cardiac magnetic resonance imaging (MRI) requires high spatial and temporal resolution and robustness to random fetal motion to capture the dynamics of the beating fetal heart. Slice-to-volume reconstruction techniques can produce high-resolution isotropic images while compensating for random fetal motion. The objective of this study was to evaluate image quality for slice-to-volume reconstruction of four-dimensional balanced steady-state free precession (bSSFP) imaging of the fetal heart. A cohort of 13 women carrying fetuses with congenital heart disease were imaged with real-time bSSFP sequences. Real-time bSSFP sequences were post-processed using a slice-to-volume reconstruction algorithm to produce retrospectively gated 4-D sequences with isotropic spatial resolution. Two radiologists evaluated slice-to-volume reconstruction image quality on a scale from 0 to 4 using 11 categories based on a segmental approach to defining cardiac anatomy and pathology. A score of 0 corresponded to cardiac structures not visualized at all and four corresponded to high quality and distinct appearance of structures. In 11 out of 13 cases, the average radiologist score of image quality across all categories was 3.0 or greater. In the remaining two cases, slice-to-volume reconstruction was not possible due to insufficient image quality in the acquisition. Slice-to-volume reconstruction has the potential to produce isotropic images with high spatial and temporal resolution that can display the anatomy of the fetal heart in arbitrary imaging planes retrospectively. More rapid, motion-robust acquisitions may be necessary to successfully reconstruct the fetal heart in all patients.

The degradation of performance of a state-of-the-art skin image classifier when applied to patient-driven internet search

Model Dermatology (https://modelderm.com; Build2021) is a publicly testable neural network that can classify 184 skin disorders. We aimed to investigate whether our algorithm can classify clinical images of an Internet community along with tertiary care center datasets. Consecutive images from an Internet skin cancer community (‘RD’ dataset, 1,282 images posted between 25 January 2020 to 30 July 2021; https://reddit.com/r/melanoma) were analyzed retrospectively, along with hospital datasets (Edinburgh dataset, 1,300 images; SNU dataset, 2,101 images; TeleDerm dataset, 340 consecutive images). The algorithm’s performance was equivalent to that of dermatologists in the curated clinical datasets (Edinburgh and SNU datasets). However, its performance deteriorated in the RD and TeleDerm datasets because of insufficient image quality and the presence of out-of-distribution disorders, respectively. For the RD dataset, the algorithm’s Top-1/3 accuracy (39.2%/67.2%) and AUC (0.800) were equivalent to that of general physicians (36.8%/52.9%). It was more accurate than that of the laypersons using random Internet searches (19.2%/24.4%). The Top-1/3 accuracy was affected by inadequate image quality (adequate = 43.2%/71.3% versus inadequate = 32.9%/60.8%), whereas participant performance did not deteriorate (adequate = 35.8%/52.7% vs. inadequate = 38.4%/53.3%). In this report, the algorithm performance was significantly affected by the change of the intended settings, which implies that AI algorithms at dermatologist-level, in-distribution setting, may not be able to show the same level of performance in with out-of-distribution settings.

Fractal dimension of retinal vasculature as an image quality metric for automated fundus image analysis systems

Automated fundus screening is becoming a significant programme of telemedicine in ophthalmology. Instant quality evaluation of uploaded retinal images could decrease unreliable diagnosis. In this work, we propose fractal dimension of retinal vasculature as an easy, effective and explainable indicator of retinal image quality. The pipeline of our approach is as follows: utilize image pre-processing technique to standardize input retinal images from possibly different sources to a uniform style; then, an improved deep learning empowered vessel segmentation model is employed to extract retinal vessels from the pre-processed images; finally, a box counting module is used to measure the fractal dimension of segmented vessel images. A small fractal threshold (could be a value between 1.45 and 1.50) indicates insufficient image quality. Our approach has been validated on 30,644 images from four public database.