Poor Image Quality Research Articles

Parasitic Egg Detection and Classification in Low-Cost Microscopic Images Using Transfer Learning

Intestinal parasitic infection leads to several morbidities in humans worldwide, especially in tropical countries. The traditional diagnosis usually relies on manual analysis from microscopic images which is prone to human error due to morphological similarity of different parasitic eggs and abundance of impurities in a sample. Many studies have developed automatic systems for parasite egg detection to reduce human workload. However, they work with high-quality microscopes, which unfortunately remain unaffordable in some rural areas. Our work thus exploits a benefit of a low-cost USB microscope. This instrument however provides poor quality images due to the limitation of magnification (10×\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\ imes$$\\end{document}), causing difficulty in parasite detection and species classification. In this paper, we propose a CNN-based technique using transfer learning strategy to enhance the efficiency of automatic parasite classification in poor-quality microscopic images. The patch-based technique with a sliding window is employed to search for the location of the eggs. Two networks, AlexNet and ResNet50, are examined with a trade-off between architecture size and classification performance. The results show that our proposed framework outperforms the state-of-the-art object recognition methods. Our system combined with the final decision from an expert may improve the real faecal examination with low-cost microscopes.

The effect of non-contrast enhanced MRA on patients with renal insufficiency and foot pain.

To investigate the feasibility of non-contrast magnetic resonance angiography of arteries and veins (NATIVE) sampling perfection with application optimized contrasts by using different flip angle evolution (SPACE) and quiescent interval single shot (QISS) in assessing foot arteries of patients with renal insufficiency and foot pain. Fifty-three patients (mean age = 44.2 ± 11.4 years, male: female = 27:26) underwent QISS and NATIVE-SPACE. The source images were reconstructed to maximum intensity projection and volume render. The image quality of QISS and NATIVE-SPACE was rated (0-3, poor to excellent), and was compared using Wilcoxon test. True or false positive was determined by comparing the findings of QISS and NATIVE-SPACE. The relative signal intensity of artery was obtained for each case, and was compared between QISS and NATIVE-SPACE using Mann Whitney test. The acquisition time of NATIVE-SPACE was significantly longer than that of QISS (178.4 ± 35.7 seconds vs 45.4 ± 8.9 seconds, P < .001). QISS had significantly lower image quality score versus NATIVE-SPACE (1.4 ± 0.5 vs 2.4 ± 0.6, P = .02). Fifteen percentage (8/53) NATIVE-SPACE cases had poor image quality due to the similarity of peak flow and minimum flow. The relative signal intensity was significantly lower in QISS versus NATIVE-SPACE (9.7 ± 1.3 vs 68.2 ± 12.4, P < .001). NATIVE-SPACE is valuable in evaluating foot arteries of patients with renal insufficiency. QISS can serve as an alternative test to NATIVE-SPACE.

Dental bitewing radiographs segmentation using deep learning-based convolutional neural network algorithms.

Dental radiographs, particularly bitewing radiographs, are widely used in dental diagnosis and treatment Dentalimage segmentation is difficult for various reasons, such as intricate structures, low contrast, noise, roughness, andunclear borders, resulting in poor image quality. Recent developments in deep learning models have improvedperformance in analyzing dental images. In this research, our primary objective is to determine the most effectivesegmentation technique for bitewing radiographs based on different metrics: accuracy, training time, and thenumber of training parameters as a reflection of architectural cost. In this research, we employ several deep learning models, namely Resnet-18, Resnet-50, Xception, InceptionResnet v2, and Mobilenetv2, to segment bitewing radiographs. The process begins by importing the radiographsinto MATLAB®(MathWorks Inc), where the images are first improved, then segmented using the graph cut methodbased on regions to produce a binary mask that distinguishes the background from the original X-ray. The deep learning models were trained on 298 and 99 radiograph training and validation sets and were evaluatedusing 99 images from the testing set. We also compare the segmentation model using several criteria, includingaccuracy, speed, and size, to determine which network is superior. Furthermore, we compare our findings with priorresearch to provide a comprehensive understanding of the advancements made in dental image segmentation. Theaccurate segmentation achieved was 93.67% and 94.42% by the Resnet-18 and Resnet-50 models, respectively. This research advances dental image analysis and facilitates more accurate diagnoses and treatment planning bydetermining the best segmentation technique. The outcomes of this study can guide researchers and practitionersin selecting appropriate segmentation methods for practical dental image analysis.

Improving portable low-field MRI image quality through image-to-image translation using paired low- and high-field images

Low-field portable magnetic resonance imaging (MRI) scanners are more accessible, cost-effective, sustainable with lower carbon emissions than superconducting high-field MRI scanners. However, the images produced have relatively poor image quality, lower signal-to-noise ratio, and limited spatial resolution. This study develops and investigates an image-to-image translation deep learning model, LoHiResGAN, to enhance the quality of low-field (64mT) MRI scans and generate synthetic high-field (3T) MRI scans. We employed a paired dataset comprising T1- and T2-weighted MRI sequences from the 64mT and 3T and compared the performance of the LoHiResGAN model with other state-of-the-art models, including GANs, CycleGAN, U-Net, and cGAN. Our proposed method demonstrates superior performance in terms of image quality metrics, such as normalized root-mean-squared error, structural similarity index measure, peak signal-to-noise ratio, and perception-based image quality evaluator. Additionally, we evaluated the accuracy of brain morphometry measurements for 33 brain regions across the original 3T, 64mT, and synthetic 3T images. The results indicate that the synthetic 3T images created using our proposed LoHiResGAN model significantly improve the image quality of low-field MRI data compared to other methods (GANs, CycleGAN, U-Net, cGAN) and provide more consistent brain morphometry measurements across various brain regions in reference to 3T. Synthetic images generated by our method demonstrated high quality both quantitatively and qualitatively. However, additional research, involving diverse datasets and clinical validation, is necessary to fully understand its applicability for clinical diagnostics, especially in settings where high-field MRI scanners are less accessible.

Variations of arterial compliance and vascular resistance due to plaque or infarct in a single vascular territory of the middle cerebral artery.

Arterial compliance (AC) and vascular resistance (VR) are crucial for the regulation capacity of the vascular system. However, alterations of these features and hemodynamics due to atherosclerosis in a single intracranial artery territory have not been extensively investigated. Thus this study aimed to examine the AC, VR, and hemodynamic variations due to plaque and infarction in the middle cerebral artery (MCA). Patients with symptomatic MCA atherosclerosis were recruited. Both sides of the MCA were assessed and then classified according to the following scheme: group 0, without plaque; group 1, with plaque but without infarct; group 2, with plaque and infarct in the supplying territories. Data on AC, VR, blood flow, and pulsatility index (PI) were obtained based on 4D flow magnetic resonance imaging (MRI) and the Windkessel model. A total of 63 patients were recruited. After 17 MCAs were excluded (occlusion, n=6; poor image quality, n=11), datasets on 109 MCAs were finally collected and classified into group 0 (n=39), group 1 (n=40), and group 2 (n=30). From groups 0 to 2, there was a decrease in AC (0.0060±0.0031 vs. 0.0052±0.0029 vs. 0.0026±0.0020 mL/mmHg) and an increase in VR [28.65±16.11 vs. 42.59±27.53 vs. 63.21±40.37 mmHg/(mL/s)]. Compared to group 1, group 2 had significantly decreased AC (0.0052±0.0029 vs. 0.0026±0.0020 mL/mmHg; P=0.003) and increased VR [42.59±27.53 vs. 63.21±40.37 mmHg/(mL/s); P=0.021]. From group 0 to group 2, there was a decrease in blood flow (179.29±73.57 vs. 125.11±59.04 vs. 92.05±48.79 mL/min; P<0.001). The PI varied significantly among the 3 groups (0.86±0.20 vs. 1.12±0.50 vs. 0.79±0.16; P<0.001), with group 1 having the highest PI. With the occurrence of plaque and infarct, AC and blood flow progressively decrease while VR increases. The PI was the highest in the group with plaque and without infarct. Assessments of vascular function and hemodynamics in a single artery territory can clarify comprehensive alterations in the cerebral vascular system (CVS).

Volumetric and Simultaneous Photoacoustic and Ultrasound Imaging with a Conventional Linear Array in a Multiview Scanning Scheme.

Volumetric, multimodal imaging with precise spatial and temporal co-registration can provide valuable and complementary information for diagnosis and monitoring. Considerable research has sought to combine 3D photoacoustic (PA) and ultrasound (US) imaging in clinically translatable configurations. However, technical compromises currently result in poor image quality either for photoacoustic or ultrasonic modes. This work aims to provide translatable, high quality, simultaneously co-registered dual-mode PA/US 3D tomography. Volumetric imaging based on a synthetic aperture approach was implemented by interlacing PA and US acquisitions during a rotate-translate scan with a 5-MHz linear array (12 angles and 30-mm translation to image a 21-mm diameter, 19 mm long cylindrical volume within 21 seconds). For co-registration, an original calibration method using a specifically designed thread phantom was developed to estimate 6 geometrical parameters and 1 temporal off-set through global optimization of the reconstructed sharpness and superposition of calibration phantom structures. Phantom design and cost function metrics were selected based on analysis of a numerical phantom, and resulted in a high estimation accuracy for the 7 parameters. Experimental estimations validated the calibration repeatability. The estimated parameters were used for bimodal reconstruction of additional phantoms with either identical or distinct spatial distributions of US and PA contrasts. Superposition distance of the two modes was within < 10% of the acoustic wavelength and a wavelength-order uniform spatial resolution was obtained. This dual-mode PA/US tomography should contribute to more sensitive and robust detection and follow-up of biological changes or the monitoring slower-kinetic phenomena in living systems such as the accumulation of nanoagents.

Cross-Level Multi-Modal Features Learning With Transformer for RGB-D Object Recognition

Object recognition, one of the main goals of robot vision, is a vital prerequisite for service robots to perform domestic tasks. Thanks to the rich sense of information provided by RGB-D sensors, RGB-D-based object recognition has received increasing attention. However, the existing works focus on collaborative RGB and depth data for object recognition, while ignoring the influence of depth image quality on recognition performance. Moreover, in real-world scenarios, there are many objects with strong similarity from certain observation angles, which poses a challenge for the service robot to recognize objects accurately. In this paper, we propose CNN-TransNet, a novel end-to-end Transformer-based architecture with convolutional neural networks (CNNs) for RGB-D object recognition. In order to deal with the effect of high inter-class similarity, discriminative multi-modal feature representations are generated by learning and relating multi-modal features at multiple levels. Besides, we employ a multi-modal fusion and projection (MMFP) module to reweight the contribution of each modality to address the problem of poor-quality depth image. Our proposed approach achieves state-of-the-art performance on three datasets (including Washington RGB-D Object Dataset, JHUIT-50, and Object Clutter Indoor Dataset), with accuracy of 95.4%, 98.1%, and 94.7%, respectively. The results demonstrate the effectiveness and superiority of the proposed model in RGB-D object recognition task.

Fish Recognition in the Underwater Environment Using an Improved ArcFace Loss for Precision Aquaculture

Accurate fish individual recognition is one of the critical technologies for large-scale fishery farming when trying to achieve accurate, green farming and sustainable development. It is an essential link for aquaculture to move toward automation and intelligence. However, existing fish individual data collection methods cannot cope with the interference of light, blur, and pose in the natural underwater environment, which makes the captured fish individual images of poor quality. These low-quality images can cause significant interference with the training of recognition networks. In order to solve the above problems, this paper proposes an underwater fish individual recognition method (FishFace) that combines data quality assessment and loss weighting. First, we introduce the Gem pooing and quality evaluation module, which is based on EfficientNet. This module is an improved fish recognition network that can evaluate the quality of fish images well, and it does not need additional labels; second, we propose a new loss function, FishFace Loss, which will weigh the loss according to the quality of the image so that the model focuses more on recognizable fish images, and less on images that are difficult to recognize. Finally, we collect a dataset for fish individual recognition (WideFish), which contains and annotates 5000 images of 300 fish. The experimental results show that, compared with the state-of-the-art individual recognition methods, Rank1 accuracy is improved by 2.60% and 3.12% on the public dataset DlouFish and the proposed WideFish dataset, respectively.

Single photon compressive imaging with enhanced quality using an untrained neural network

Traditional single photon compressive imaging has poor imaging quality. Although the method of deep learning can alleviate the problem, the harsh training sets have become a problem. In this paper, an untrained neural network is used to address this problem. A whole imaging system was established, and simulation studies based on the Monte Carlo method have been undertaken. The results show that the proposed method has improved the image quality and solved the troublesome training sets problem while ensuring imaging speed. At the same time, the discussion of input pictures, imaging type, and anti-noise capability provide a way to prove CNN’s tendency to natural images. It is also found that the network changes the sensitivity of the system to the photon numbers. The research work will provide some basis for subsequent study on single compressive photon imaging and untrained neural networks.

Joint learning of sparse and limited-view guided waves signals for feature reconstruction and imaging

Sparse and limited-view ultrasonic guided wave imaging has become a research hotspot in the field. Studies have shown that traditional under-sampling ultrasonic imaging methods either require a significant amount of time to recover the full data or produce poor quality imaging results. To address these issues, this paper proposes an end-to-end ultrasonic guided wave joint learning imaging method for sparse and limited-view transducer arrays, which integrates sparse feature reconstruction and deep learning imaging methods. Numerical and experimental studies demonstrate that this approach significantly improves the quality of imaging results. The quality of imaging results for sparse and limited-view transducer arrays is evaluated and quantified using average correlation coefficients on the testing set. The feasibility and effectiveness of the proposed method have been verified.

Proper Timing and Manipulation of Self – Processing Dental Radiographs in Baghdad Weather (Clinical Comparative study)

Poor image quality in self processed films is not an uncommon problem despite following the manufacturer's instructions, this matter has its influence on the ultimate diagnosis when the radiographic examination is the backbone in this diagnosis. The present study was carried out to investigate this problem and to provide a guideline measure to overcome it. One hundred fifty films were processed using three different manipulations: message, shaking and a combination of both, the resultant film quality was measured and comparison was made the results showed that combination of message and shaking procedure yield significantly better results regarding image quality in comparison to the other two procedures.

Runtime optimization of acquisition trajectories for x-ray computed tomography with a robotic sample holder

Tomographic imaging systems are expected to work with a wide range of samples that house complex structures and challenging material compositions, which can influence image quality in a bad way. Complex samples increase total measurement duration and may introduce beam-hardening artifacts that lead to poor reconstruction image quality. This work presents an online trajectory optimization method for an x-ray computed tomography system with a robotic sample holder. The proposed method reduces measurement time and increases reconstruction image quality by generating an optimized spherical trajectory for the given sample without prior knowledge. The trajectory is generated successively at runtime based on intermediate sample measurements. We present experimental results with the robotic sample holder where two sample measurements using an optimized spherical trajectory achieve improved reconstruction quality compared to a conventional spherical trajectory. Our results demonstrate the ability of our system to increase reconstruction image quality and avoid artifacts at runtime when no prior information about the sample is provided.

Comparative Analysis of Vision Transformer Models for Facial Emotion Recognition Using Augmented Balanced Datasets

Facial emotion recognition (FER) has a huge importance in the field of human–machine interface. Given the intricacies of human facial expressions and the inherent variations in images, which are characterized by diverse facial poses and lighting conditions, the task of FER remains a challenging endeavour for computer-based models. Recent advancements have seen vision transformer (ViT) models attain state-of-the-art results across various computer vision tasks, encompassing image classification, object detection, and segmentation. Moreover, one of the most important aspects of creating strong machine learning models is correcting data imbalances. To avoid biased predictions and guarantee reliable findings, it is essential to maintain the distribution equilibrium of the training dataset. In this work, we have chosen two widely used open-source datasets, RAF-DB and FER2013. As well as resolving the imbalance problem, we present a new, balanced dataset, applying data augmentation techniques and cleaning poor-quality images from the FER2013 dataset. We then conduct a comprehensive evaluation of thirteen different ViT models with these three datasets. Our investigation concludes that ViT models present a promising approach for FER tasks. Among these ViT models, Mobile ViT and Tokens-to-Token ViT models appear to be the most effective, followed by PiT and Cross Former models.

Echocardiographic features predict the performance of transthoracic echocardiography in ruling out infective endocarditis: a 5-year study

Abstract Background Transoesophageal echocardiography (TOE) remains the gold standard imaging modality to aid in the diagnosis of infective endocarditis (IE), but is resource intensive and confers patient risk. Transthoracic echocardiography (TTE) is a more accessible alternative, but previous studies report varying ability of TTE to effectively rule out disease. As a result, negative TTE is often still followed with TOE when clinical suspicion remains. However, these studies did not account for TTE quality and other echocardiographic features within the TTE that may decrease its diagnostic utility and thereby reduce its ability to effectively rule out disease. Understanding the risk of missing IE for a given negative TTE could avoid the need for subsequent TOE. Purpose Our aim is to devise and test the validity of a scoring system which can stratify TTEs by quality to predict TTE performance and inform the decision for whether follow-up TOE is necessary. Methods In this retrospective single-centre study, the Predictor of TTE Performance (PTP) scoring system was developed and applied to patients undergoing both TTE and TOE for suspected IE at our tertiary centre between May 2017 and May 2022. Within the scoring system, a point was allocated for the presence of each of the following on TTE: a mobile echodensity; poor image quality; more than mild regurgitation; a prosthetic valve or prosthetic intracardiac material; and other valvular abnormalities. Poor image quality was defined as any TTE that was unable to achieve good endocardial visualisation from at least two views. Valvular abnormalities warranting a point included congenital abnormalities such as bicuspid aortic valve and acquired abnormalities such as rheumatic valvular disease, mitral valve prolapse and valvular stenosis. The negative predictive value (NPV) for TTEs in each score category was analysed using TOE as the gold standard. Results A total of 362 patients with paired TTE/TOE were included for analysis. 60 (16.6%) had positive TOE, with the overall NPV of TTE being 88.5% (95% CI: 84.6–91.5%). Using the PTP scoring system, patients were stratified into low-risk (score of 0, n = 116), moderate-risk (score of 1, n = 140) and high-risk (score greater than 1, n = 106) groups. Increasing PTP score was associated with poorer TTE performance, with the low-risk, moderate-risk and high-risk categories having NPVs of 98.3% (95% CI: 93.9–99.7%), 90.2% (95% CI: 83.9–94.2%) and 72.3% (95% CI: 61.8–80.8%) respectively. Conclusion The ability of TTE to exclude IE is excellent when patients have no echocardiographic features that reduce TTE quality or predict increased risk for endocarditis. In patients with a low PTP score, negative TTE may render follow-up TOE unnecessary.

Calcification of different cardiac structures shown by cardiac CT: how important are lipids?

Abstract Background The relationship between dyslipidemia and atherosclerotic plaque as a risk factor for coronary artery disease is well-established. However, the involvement of lipids in the development of valve and aortic calcification beyond coronary disease is yet to be understood. Objective To evaluate the involvement of lipids in the calcification of various cardiac structures as measured by cardiac CT. Methods In this single-center study, we analyzed data from 316 consecutive patients who underwent cardiac CT scans between January 2018 and December 2019. Patients with poor imaging quality, constrictive pericarditis, prosthetic valves and/or devices were excluded. The calcium score of various cardiac structures, including coronary arteries (CA), mitral valve (MV), aortic valve (AoV), ascending aorta (AAo), and aortic arch (AAc), was calculated using the Agatston method from non-contrast ECG-gated CT. These scores were then combined to derive the valvular (VA=MV+AoV), total cardiac (TC=CA+VA), total vascular (TV=AAo+AAc), and total cardiovascular (TCV=TC+TV) calcium scores. We then collected data regarding lipid values (total cholesterol, LDL and HDL cholesterol and triglycerides). Results 275 CT scans were suitable for analysis, 142 were male (52%), with a mean age of 60±12 years (range 26–93 years). A total of 183 (66.7%) patients presented calcification in at least one location. Patients with calcification on any of the prespecified locations had higher prevalence of hypertension, dyslipidemia and type 2 diabetes mellitus (DM) than those without any calcium (p&amp;lt;0.05). The best regression models (using backwards conditional input method) showed that age was a significant predictor of calcification in all locations (and combinations of locations, p&amp;lt;0.05 for all) and HDL cholesterol was a part of these models across all calcifications types, being significant for TC score (p=0.010) and TCV score (p=0.024, see Figure). Patients above the fourth quartile (Q4) of TCV had significantly lower HDL cholesterol levels (47±12 vs 53±16mg/dL, p=0.007). Of note, total cholesterol, LDL cholesterol and triglycerides were not significant predictors of calcification in any location. Conclusion Calcification across the cardiovascular system is consistently associated with low HDL cholesterol levels, which extends beyond the traditional association with coronary disease. The calcification effect of HDL is found to be more significant than other lipids, however, the specific contribution of HDL subtypes is not yet fully understood.

Left atrial strain measured by three-dimensional echocardiography predicts atrial fibrillation in the general population

Abstract Background Left atrial (LA) strain is prognostic of adverse cardiovascular events. LA strain by three-dimensional echocardiography (3DE) has been proposed as a more accurate measure of LA function and provide incremental prognostic benefits than traditional two-dimensional approaches. Our aim was to evaluate the prognostic value of LA strain by 3DE in predicting incident atrial fibrillation (AF) in the general population. Methods The study included 4,466 participants from a prospective longitudinal cohort study in the general population, among these 2,082 participants underwent 3DE of the LA. After excluding participants with poor image quality (n=70), low volume rate (n=8) and AF at baseline (n=51) 1,956 participants were included in the final study. The endpoint was incident AF. Results Mean age was 54±17 years, 44% were male. During a median follow-up time of 4.8 years (interquartile range 4.3-5.5 years) 59 participants (3.0%) developed AF. At baseline, mean left atrial reservoir strain (LASr) was 29.3±6.7%, mean left atrial conduit strain (LAScd) was 16.5±6.7%, and mean left atrial contractile strain (LASct) was 12.8±4.8. The incidence of AF increased significantly with decreasing LA strain measures (p&amp;lt;0.001) (figure 1). In univariable analysis, all three parameters were associated with incident AF (p value for all &amp;lt;0.01). After multivariable adjustment for CHARGE-AF risk score (age, height, weight, systolic blood pressure, diastolic blood pressure, smoking, antihypertensive medication, diabetes, congestive heart failure and myocardial infarction), only LASr and LASct were associated with incident AF (LASr: HR 1.11 (1.07-1.17), p&amp;lt;0.001, per 1% decrease; LASct: HR 1.16 (1.09-1.24), p&amp;lt;0.001, per 1% decrease), whereas LAScd was not (HR 1.04 (0.98-1.10), p=0.17, per 1% decrease). Conclusion LA reservoir and contraction strain by 3DE are independently associated incident AF in the general population.

Evaluating the safe zone for lumbar pedicle screws: are midline crossing screws indicative of pedicle breach?

BACKGROUND CONTEXTPedicle screw breach (PSB) is not uncommon following lumbar instrumentation, and in some instances, it may lead to vascular and/or neurologic complications. Previous literature suggested that screws crossing the vertebral midline on an anterior-posterior (AP) radiograph (or midsagittal on CT) are concerning for medial pedicle breach. OBJECTIVEOur primary aim was to map out the safe zones (SZ) of bilateral pedicle instrumentation and their relationship at each lumbar vertebral level. Our secondary aim was to evaluate the presence of SZs’ intersection at each lumbar level, denoting safe midline pedicle screw crossing not otherwise associated with medial pedicle breach. STUDY DESIGN/SETTINGRetrospective Anatomical Study. PATIENT SAMPLEAdult patients in the from “The Cancer Imaging Archive” (TCIA) database who have not had thoraco-lumbo-sacral fusion. OUTCOME MEASURESPhysiologic measures obtained through 3D analysis of CT images and virtual pedicle screws. METHODCT scans of 51 patients were randomly selected from “The Cancer Imaging Archive” (TCIA) online database for analysis. The Sectra 3D Spine software was used to create 3D renderings, place virtual screws, and make measurements. At each lumbar vertebra, the right and left pedicle corridors were mapped. At each pedicle, two screw positions were templated, the “medial limit screw” (MLS) and the “lateral limit screw” (LLS). Each limit screw was the most extreme position that the screw could exist in without causing a medial or lateral breach. The safe zone was defined as the zone between MLS and LLS. Measurements were taken for each level (between L1 and L5) and side (Left, Right). RESULTSA total of 253 lumbar vertebrae from 51 patients (mean age 53.1, 56.9% male) were included. Two vertebrae from two patients were removed for poor image quality. Out of the 506 screw positions analyzed in our study, 97.4% had overlapping SZ and crossed the midplane without medial pedicle breach. The significant factors (p<.01) for safe midplane-crossing screws included: the screw length (L1–L5); the laterality of the screw entry point (L1–L4); and the pedicle diameter (L2 and L5). CONCLUSIONSA midline crossing pedicle screw on a lumbar AP radiograph is not necessarily indicative of a medial pedicle screw breach. Anatomical (ie, larger pedicle diameter) and technical (ie, longer screws, and lateral entry points) factors allow for safety zone intersections and indicate safe midline crossing by pedicle screws.

Analysis of The Utilization of The Automatic Exposure Control (AEC) Feature in The Use of Deep Learning Breast Image Technology in Women's Mammogram Screening Examinations at Dharmais Cancer Hospital

Deep learning technology is useful for radiology specialists as double reading to help increase the accuracy of image interpretation results. One of the preparations for maximizing the use of this technology is using good-quality images as the source. The Automatic Exposure Control (AEC) feature, which functions to determine exposure factors automatically, is expected to help produce images with good and consistent quality so that deep learning technology can work more effectively. This research aims to determine the quality results of mammogram images taken using the AEC feature and to analyze the use of deep learning technology in evaluating mammogram images. This research method is retrospective by collecting 800 mammogram images randomly and anonymously. Three hundred images were tested, 500 were evaluated, and 250 were analyzed for image quality based on references related to applying AEC and assessing the contrast-to-noise ratio (CNR). Deep learning technology was analyzed by comparing the results of mammogram image evaluation using deep learning and the evaluation results of a radiology specialist. Deep learning technology analysis shows that 98% of mammograms have the same results as the radiology doctor's evaluation, and 2% have different results from the radiology doctor's evaluation where the image has a dense breast type. The image quality results in this research showed that 97.6% of the 250 samples taken using the AEC feature had good image quality, and 2.4% had poor image quality due to inappropriate breast positioning during the examination.

WSRGAN: A wavelet-based GAN for super-resolution of plane-wave ultrasound images without sampling loss

Plane-wave ultrasound imaging has ultra-fast temporal resolution and can image at frame rates exceeding 1 kHz, enabling breakthrough technologies like shear wave elastography. However, the imaging process of this technology lacks focus and poor imaging quality and usually requires beamforming technology to assist imaging, such as Coherent Plane-Wave Compounding (CPWC). The imaging quality of this method is often directly related to the number of beams, sacrificing the time advantage of single-beam plane wave imaging. This study proposes a Wavelet-based Generative Adversarial Network for Super-Resolution (WSRGAN) to meet the growing demand for high-quality imaging of single-beam plane waves. WSRGAN uses the encoding and decoding network as the generator and uses wavelet transform to replace the sampling process. Furthermore, we further designed adaptive wavelet attention to enable the model to focus attention on different levels of features at different stages. Wavelet GAN loss was proposed as GAN loss, and a new combined loss function was designed for the generator. The experiment was conducted on the Plane-wave Imaging Challenge in Medical UltraSound (PICMUS) 2016 dataset. On the point target, the Full Width at Half Maximum (FWHM) of WSRGAN reached 0.268 mm to 0.502 mm. On the cyst target, the contrast of WSRGAN reaches 26.156 dB to 37.223 dB. Among invivo targets, the Peak Signal-to-Noise Ratio (PSNR) of WSRGAN is 41.518 dB, and the Structural Similarity Index Measurement (SSIM) is 0.984. Experiments show that each module proposed by the model has a total contribution and is helpful for various experiments. All targets performed well.

Enhanced independent component analysis and fuzzy C-mean clustering based on novel bat algorithm for noisy image segmentation

Fuzzy c-means clustering is widely recognized as one of the most effective methods for image segmentation and achieving accurate classification. However, this method has two significant drawbacks: its sensitivity to noise and its convergence to local minimum clusters’ centroids. In this paper, we proposed a novel model called EIFCMNB, which incorporates enhanced independent component analysis (EICA), fuzzy c-means clustering (FCMC) and novel bat algorithm (NBA) for noise image segmentation. The suggested model consists of two main phases: image denoising and extraction of the regions of interest (ROIs). In the first phase, the enhanced independent component analysis (EICA) algorithm is used for recovering a good quality image, from a noisy image of poor quality. Several noisy images, with noise variances ranging from 5 to 20, were filtered. The resulting images were then evaluated based on several criteria viz: Peak Signal to Noise Ratio (PSNR), Relative Norm Error (RE), Normalized Cross-Correlation (NCC), and Structural Similarity index measure (SSIM). In the second phase, the fuzzy c-means clustering based on a novel bat algorithm is adopted to calculate optimal clusters’ centroids and extract the ROIs. By incorporating the new bat algorithm, we aim to overcome the problem of converging to local minimums and achieve improved segmentation accuracy. Promising experimental results have been obtained by applying the proposed model to MRI brain images and x-ray welding images. Two criteria viz: VPE end VPC have been employed to evaluate the suggested approach. The experiments clearly demonstrate that our suggested model effectively addresses the sensitivity to noise problem and provides optimal clusters’ centroids. Moreover, it outperforms several FCMC-based algorithms, exhibiting superior performance in terms of image segmentation and classification.