Set Of Images Research Articles

Oral CT Contrast Agents: What's New and Why, From the AJR Special Series on Contrast Media.

Current CT oral contrast agents improve the conspicuity of and confidence in bowel and peritoneal findings in many clinical scenarios, particularly for outpatient and oncologic abdominopelvic imaging. Yet, existing positive and neutral oral contrast agents may diminish the detectability of certain radiologic findings, frequently in the same scans in which the oral contrast agent improves the detectability of other findings. With ongoing improvements in CT technology, particularly multienergy CT, opportunities are opening for new types of oral contrast agents to further improve anatomic delineation and disease detection using CT. The CT signal of new dark oral contrast agents and of new high-Z oral contrast agents promises to combine the strengths of both positive and neutral oral CT contrast agents by providing distinct CT appearances in comparison with bodily tissues, iodinated IV contrast agents, and other classes of new CT contrast agents. High-Z oral contrast agents will unlock previously inaccessible capabilities of multienergy CT, particularly photon-counting detector CT, for differentiating simultaneously administered IV and oral contrast agents; this technique will allow generation of rich 3D, intuitive, perfectly coregistered, high-resolution image sets with individual contrast agent "colors" that provide compelling clarity for intertwined intraabdominal anatomy and disease processes.

Harnessing the power of longitudinal medical imaging for eye disease prognosis using Transformer-based sequence modeling

Deep learning has enabled breakthroughs in automated diagnosis from medical imaging, with many successful applications in ophthalmology. However, standard medical image classification approaches only assess disease presence at the time of acquisition, neglecting the common clinical setting of longitudinal imaging. For slow, progressive eye diseases like age-related macular degeneration (AMD) and primary open-angle glaucoma (POAG), patients undergo repeated imaging over time to track disease progression and forecasting the future risk of developing a disease is critical to properly plan treatment. Our proposed Longitudinal Transformer for Survival Analysis (LTSA) enables dynamic disease prognosis from longitudinal medical imaging, modeling the time to disease from sequences of fundus photography images captured over long, irregular time periods. Using longitudinal imaging data from the Age-Related Eye Disease Study (AREDS) and Ocular Hypertension Treatment Study (OHTS), LTSA significantly outperformed a single-image baseline in 19/20 head-to-head comparisons on late AMD prognosis and 18/20 comparisons on POAG prognosis. A temporal attention analysis also suggested that, while the most recent image is typically the most influential, prior imaging still provides additional prognostic value.

SPD Manifold Deep Metric Learning for Image Set Classification.

By characterizing each image set as a nonsingular covariance matrix on the symmetric positive definite (SPD) manifold, the approaches of visual content classification with image sets have made impressive progress. However, the key challenge of unhelpfully large intraclass variability and interclass similarity of representations remains open to date. Although, several recent studies have mitigated the two problems by jointly learning the embedding mapping and the similarity metric on the original SPD manifold, their inherent shallow and linear feature transformation mechanism are not powerful enough to capture useful geometric features, especially in complex scenarios. To this end, this article explores a novel approach, termed SPD manifold deep metric learning (SMDML), for image set classification. Specifically, SMDML first selects a prevailing SPD manifold neural network (SPDNet) as the backbone (encoder) to derive an SPD matrix nonlinear representation. To counteract the degradation of structural information during multistage feature embedding, we construct a Riemannian decoder at the end of the encoder, trained by a reconstruction error term (RT), to induce the generated low-dimensional feature manifold of the hidden layer to capture the pivotal information about the visual data describing the imaged scene. We demonstrate through theory and experiments that it is feasible to replace the Riemannian metric with Euclidean distance in RT. Then, the ReCov layer is introduced into the established Riemannian network to regularize the local statistical information within each input feature matrix, which enhances the effectiveness of the learning process. The theoretical analysis of the activation function used in the ReCov layer in terms of continuity and conditions for generating positive definite matrices is beneficial for network design. Inspired by the fact that the single cross-entropy loss used for training is unable to effectively parse the geometric distribution of the deep representations, we finally endow the suggested model with a novel metric learning regularization term. By explicitly incorporating the encoding and processing of the data variations into the network learning process, this term can not only derive a powerful Riemannian representation but also train an effective classifier. The experimental results show the superiority of the proposed approach on three typical visual classification tasks.

Comprehensive Analysis of Adverse Events Associated With Hypoglossal Nerve Stimulators: Insights From the MAUDE Database

AbstractObjectiveThis study aims to examine the adverse events associated with hypoglossal nerve stimulator (HNS) implantation for treating obstructive sleep apnea (OSA), drawing data from the Manufacturer and User Facility Device Experience (MAUDE) database. We aim to provide a comprehensive and updated account of these adverse events.Study DesignRetrospective analysis.SettingMAUDE Database review.MethodsA retrospective analysis was performed on the MAUDE database to collect all HNS‐related reports from May 2014 to December 2023. Variables collected included date of event, event description, nature of event, iatrogenic injuries, required interventions, and, if available, root causes. Each event description was analyzed to classify the adverse event, the postevent intervention, and device model number.ResultsOut of 1178 reports fulfilling the inclusion criteria, 1312 adverse events were identified. Common adverse events included infection (24.0%), pain (19.7%), and hematoma/seroma (10.2%). Approximately 83.1% of these adverse events necessitated medical and/or surgical intervention. The most frequent procedures included explantation (29.4%) and device repositioning (15.8%). Pneumothorax was reported in 50 cases, with 41 (82.0%) requiring a chest tube to be inserted. Three adverse events described overstimulation in the setting of magnetic resonance imaging (MRI) despite the implantation of MRI‐compatible second‐generation internal pulse generators.ConclusionWhile HNS implantation has been established as a reliable intervention for OSA in cases of continuous positive airway pressure failure or intolerance, this study highlights several perioperative and postoperative difficulties and complications. Understanding these challenges is essential for refining surgical practices and enhancing patient consent processes, ultimately aiming to improve therapeutic outcomes.

Examination of the PET in vivo generator 134Ce as a theranostic match for 225Ac.

The radionuclide pair cerium-134/lanthanum-134 (134Ce/134La) was recently proposed as a suitable diagnostic counterpart for the therapeutic alpha-emitter actinium-225 (225Ac). The unique properties of 134Ce offer perspectives for developing innovative in vivo investigations that are not possible with 225Ac. In this work, 225Ac- and 134Ce-labelled tracers were directly compared using internalizing and slow-internalizing cancer models to evaluate their in vivo comparability, progeny meandering, and potential as a matched theranostic pair for clinical translation. Despite being an excellent chemical match, 134Ce/134La has limitations to the setting of quantitative positron emission tomography imaging. The precursor PSMA-617 and a macropa-based tetrazine-conjugate (mcp-PEG8-Tz) were radiolabelled with 225Ac or 134Ce and compared in vitro and in vivo using standard (radio)chemical methods. Employing biodistribution studies and positron emission tomography (PET) imaging in athymic nude mice, the radiolabelled PSMA-617 tracers were evaluated in a PC3/PIP (PC3 engineered to express a high level of prostate-specific membrane antigen) prostate cancer mouse model. The 225Ac and 134Ce-labelled mcp-PEG8-Tz were investigated in a BxPC-3 pancreatic tumour model harnessing the pretargeting strategy based on a trans-cyclooctene-modified 5B1 monoclonal antibody. In vitro and in vivo studies with both 225Ac and 134Ce-labelled tracers led to comparable results, confirming the matching pharmacokinetics of this theranostic pair. However, PET imaging of the 134Ce-labelled precursors indicated that quantification is highly dependent on tracer internalization due to the redistribution of 134Ce's PET-compatible daughter 134La. Consequently, radiotracers based on internalizing vectors like PSMA-617 are suited for this theranostic pair, while slow-internalizing 225Ac-labelled tracers are not quantitatively represented by 134Ce PET imaging. When employing slow-internalizing vectors, 134Ce might not be an ideal match for 225Ac due to the underestimation of tumour uptake caused by the in vivo redistribution of 134La. However, this same characteristic makes it possible to estimate the redistribution of 225Ac's progeny noninvasively. In future studies, this unique PET in vivo generator will further be harnessed to study tracer internalization, trafficking of receptors, and the progression of the tumour microenvironment.

Artwork Recommendations based on User Preferences: Integrating Clustering Analysis with Visual Features

Recently, recommendation systems have become one of the important elements for sales and marketing, and their application is almost essential in the shopping and cultural industries. Despite the increase in online exhibitions and the growing audience engaging with artworks in digital spaces, the utilization of artwork recommendation systems remains inadequate. Thus, this study proposes an artwork recommendation system, which provides artwork groups based on a visual clustering technique and user preferences with WikiArt datasets. The visual attributes of artworks were extracted using VGG16, and k -means clustering was utilized to group a set of images according to their feature similarities. To generate recommendations, new artworks were randomly selected from particular clusters, taking into account users’ preferences. Then, an experiment was conducted to investigate whether the recommended artworks satisfied the users. The statistical results indicate that users’ perceived satisfaction with the recommended artworks is notably more positive compared to their satisfaction with traditional suggested artworks. Based on this study’s findings, we present implications and limitations for future research.

On the relationship between viscoelasticity and water diffusion in soft biological tissues

Magnetic resonance elastography (MRE) and diffusion-weighted imaging (DWI) are complementary imaging techniques that detect disease based on viscoelasticity and water mobility, respectively. However, the relationship between viscoelasticity and water diffusion is still poorly understood, hindering the clinical translation of combined DWI-MRE markers.We used DWI-MRE to study 129 biomaterial samples including native and cross-linked collagen, glycosaminoglycans (GAGs) with different sulfation levels, and decellularized specimens of pancreas and liver, all with different proportions of solid tissue, or solid fractions. We developed a theoretical framework of the relationship between mechanical loss and tissue-water mobility based on two parameters, solid and fluid viscosity. These parameters revealed distinct DWI-MRE property clusters characterizing weak, moderate, and strong water-network interactions. Sparse networks interacting weakly with water, such as collagen or diluted decellularized tissue, resulted in marginal changes in water diffusion over increasing solid viscosity. In contrast, dense networks with larger solid fractions exhibited both free and hindered water diffusion depending on the polarity of the solid components. For example, polar and highly sulfated GAGs as well as native soft tissues hindered water diffusion despite relatively low solid viscosity.Our results suggest that two fundamental properties of tissue networks, solid fraction and network polarity, critically influence solid and fluid viscosity in biological tissues. Since clinical DWI and MRE are sensitive to these viscosity parameters, the framework we present here can be used to detect tissue remodeling and architectural changes in the setting of diagnostic imaging. Statement of significanceThe viscoelastic properties of biological tissues provide a wealth of information on the vital state of cells and host matrix. Combined measurement of viscoelasticity and water diffusion by medical imaging is sensitive to tissue microarchitecture. However, the relationship between viscoelasticity and water diffusion is still poorly understood, hindering full exploitation of these properties as a combined clinical biomarker. Therefore, we analyzed the parameter space accessible by diffusion-weighted imaging (DWI) and magnetic resonance elastography (MRE) and developed a theoretical framework for the relationship between water mobility and mechanical parameters in biomaterials. Our theory of solid material properties related to particle motion can be translated to clinical radiology using clinically established MRE and DWI.

Development of a mobile, high-throughput, and low-cost image-based plant growth phenotyping system.

Nondestructive plant phenotyping forms a key technique for unraveling molecular processes underlying plant development and response to the environment. While the emergence of high-throughput phenotyping facilities can further our understanding of plant development and stress responses, their high costs greatly hinder scientific progress. To democratize high-throughput plant phenotyping, we developed sets of low-cost image- and weight-based devices to monitor plant shoot growth and evapotranspiration. We paired these devices to a suite of computational pipelines for integrated and straightforward data analysis. The developed tools were validated for their suitability for large genetic screens by evaluating a cowpea (Vigna unguiculata) diversity panel for responses to drought stress. The observed natural variation was used as an input for a genome-wide association study, from which we identified nine genetic loci that might contribute to cowpea drought resilience during early vegetative development. The homologs of the candidate genes were identified in Arabidopsis (Arabidopsis thaliana) and subsequently evaluated for their involvement in drought stress by using available T-DNA insertion mutant lines. These results demonstrate the varied applicability of this low-cost phenotyping system. In the future, we foresee these setups facilitating the identification of genetic components of growth, plant architecture, and stress tolerance across a wide variety of plant species.

Joint Metric Learning-Based Class-Specific Representation for Image Set Classification.

With the rapid advances in digital imaging and communication technologies, recently image set classification has attracted significant attention and has been widely used in many real-world scenarios. As an effective technology, the class-specific representation theory-based methods have demonstrated their superior performances. However, this type of methods either only uses one gallery set to measure the gallery-to-probe set distance or ignores the inner connection between different metrics, leading to the learned distance metric lacking robustness, and is sensitive to the size of image sets. In this article, we propose a novel joint metric learning-based class-specific representation framework (JMLC), which can jointly learn the related and unrelated metrics. By iteratively modeling probe set and related or unrelated gallery sets as affine hull, we reconstruct this hull sparsely or collaboratively over another image set. With the obtained representation coefficients, the combined metric between the query set and the gallery set can then be calculated. In addition, we also derive the kernel extension of JMLC and propose two new unrelated set constituting strategies. Specifically, kernelized JMLC (KJMLC) embeds the gallery sets and probe sets into the high-dimensional Hilbert space, and in the kernel space, the data become approximately linear separable. Extensive experiments on seven benchmark databases show the superiority of the proposed methods to the state-of-the-art image set classifiers.

The use of ferumoxytol for high-resolution vascular imaging and troubleshooting for abdominal allografts.

Ferumoxytol is an ultrasmall superparamagnetic iron oxide which has been used as an off-label intravenous contrast agent for MRI. Unlike gadolinium-based contrast agents, ferumoxytol remains in the intravascular space with a long half-life of 14-21h. During the first several hours, it acts as a blood-pool agent and has minimal parenchymal enhancement. Studies have shown adequate intravascular signal for up to 72h after initial contrast bolus. Ferumoxytol has been shown to be safe, even in patients with renal failure. Ferumoxytol has shown promise in a variety of clinical settings. The exquisite resolution enabled by the long intravascular times and lack of background parenchymal enhancement is of particular interest in the vascular imaging of solid organ allografts. Ferumoxytol magnetic resonance angiography (MRA) may identify clinically actionable findings months before ultrasound, CT angiography, or Gadolinium-enhanced MRA. Ferumoxytol MRA is of particular benefit as a troubleshooting tool in the setting of equivocal ultrasound and CT imaging. In the following review, we highlight the use of ferumoxytol for high-resolution MR vascular imaging for abdominal solid organ allografts, with representative cases.

MagiCapture: High-Resolution Multi-Concept Portrait Customization

Large-scale text-to-image models including Stable Diffusion are capable of generating high-fidelity photorealistic portrait images. There is an active research area dedicated to personalizing these models, aiming to synthesize specific subjects or styles using provided sets of reference images. However, despite the plausible results from these personalization methods, they tend to produce images that often fall short of realism and are not yet on a commercially viable level. This is particularly noticeable in portrait image generation, where any unnatural artifact in human faces is easily discernible due to our inherent human bias. To address this, we introduce MagiCapture, a personalization method for integrating subject and style concepts to generate high-resolution portrait images using just a few subject and style references. For instance, given a handful of random selfies, our fine-tuned model can generate high-quality portrait images in specific styles, such as passport or profile photos. The main challenge with this task is the absence of ground truth for the composed concepts, leading to a reduction in the quality of the final output and an identity shift of the source subject. To address these issues, we present a novel Attention Refocusing loss coupled with auxiliary priors, both of which facilitate robust learning within this weakly supervised learning setting. Our pipeline also includes additional post-processing steps to ensure the creation of highly realistic outputs. MagiCapture outperforms other baselines in both quantitative and qualitative evaluations and can also be generalized to other non-human objects.

A3C: An Image-Association-Based Computing Device Authentication Framework for People with Upper Extremity Impairments

Current computing device authentication often presents accessibility barriers for people with upper extremity impairments (UEI). In this paper, we present a framework called Accessible image-Association-based Authentication for Computing-devices (A3C) , a novel recognition-based graphical authentication framework specifically designed for people with UEI to authenticate to their computing devices. A3C requires users to provide a set of primary images that the user knows that are recognizable to them and subsequently associate each primary image with a secondary image. To evaluate the efficacy of the A3C framework, we instantiated the framework by implementing a version of A3C called A3C-FA , which uses images of faces of people the user knows as the primary image and animal images as the secondary image. We then performed three studies to evaluate A3C-FA: a shoulder-surfing attack study ( N=319 ), a close-adversary attack study ( N=268 ), and a usability study with people with UEI ( N=14 ). We found that A3C was robust against both shoulder-surfing and close-adversary attacks. We also performed a detailed study to evaluate the accessibility of A3C-FA. Our participants reported that A3C-FA was more usable and more secure than the authentication approaches with which they were familiar. Based on these findings, we suggest four areas of future research to further improve the design of A3C framework.

Nonunion scaphoid bone shape prediction using iterative kernel principal polynomial shape analysis.

The scaphoid is an important mechanical stabilizer for both the proximal and distal carpal columns. The precise estimation of the complete scaphoid bone based on partial bone geometric information is a crucial factor in the effective management of scaphoid nonunion. Statistical shape model (SSM) could be utilized to predict the complete scaphoid shape based on the defective scaphoid. However, traditional principal component analysis (PCA) based SSM is limited by its linearity and the inability to adjust the number of modes used for prediction. This study proposes an iterative kernel principal polynomial shape analysis (iKPPSA)-based SSM to predict the pre-morbid shape of the scaphoid, aiming at enhancing the accuracy as well as the robustness of the model. Sixty-five sets of scaphoid images were used to train SSM and nine sets of scaphoid images were used for validation. For each validation image set, three defect types (tubercle, proximal pole, and avascular necrosis) were virtually created. The predicted shapes of the scaphoid by PCA, PPSA, KPCA, and iKPPSA-based SSM were evaluated against the original shape in terms of mean error, Hausdorff distance error, and Dice coefficient. The proposed iKPPSA-based scaphoid SSM demonstrates significant robustness, with a generality of 0.264mm and a specificity of 0.260mm. It accounts for 99% of variability with the first seven principal modes of variation. Compared to the traditional PCA-based model, the iKPPSA-based scaphoid model prediction demonstrated superior performance for the proximal pole type fracture, with significant reductions of 25.2%, 24.7%, and 24.6% in mean error, Hausdorff distance, and root mean square error (RMSE), respectively, and a 0.35% improvement in Dice coefficient. This study showed that the iKPPSA-based SSM exploits the nonlinearity of data features and delivers high reconstruction accuracy. It can be effectively integrated into preoperative planning for scaphoid fracture management or morphology-based biomechanical modeling of the scaphoid.

Online electronic signature recognition using sparse classification techniques that support neural models

With the rapid development of information technology, electronic signature plays an increasingly important role in people’s production practice. However, there are a large number of hackers maliciously stealing information in the network. In order to avoid this phenomenon, we urgently need to strengthen the research on online electronic signature recognition technology. Based on the sparse classification technology of neural model, this paper constructs an online electronic signature recognition model by using convolutional neural network and sparse classification technology. We first extract the local features of online electronic signatures, construct feature vectors and perform sparse representation. Sub-model we construct a scheme for online electronic signature recognition based on neural models and sparse classification techniques using a combination of algorithms. We first extract the local features of online electronic signatures, construct feature vectors and perform sparse representation. At the same time, the features in the training image set are extracted, local feature sets are constructed, feature dictionaries are created, and the vectors in the feature dictionaries are matched with the global sparse vectors constructed by the electronic signatures to be detected, and the matching results are finally obtained. At the same time, the features in the training image set are extracted, the local feature set is constructed, the feature dictionary is created, and the vector in the feature dictionary is matched with the global sparse vector constructed by the electronic signature to be detected, and finally the matching result is obtained. In order to verify the accuracy of the model, we first extracted 1000 respondents for online e-signature recognition experimental results show that the recognition accuracy of online e-signature has been significantly improved. Finally, in order to determine the optimal number of training sets for the model constructed in this experiment, we analyzed the correlation between training and sample size and recognition accuracy. Finally, it was concluded that the recognition accuracy increased with the increase of the number of training samples. Electronic signatures can quickly examine the signature results, and electronic signature recognition can be used to fix and tamper-proof evidence to enhance the security and trustworthiness of signatures, and it is imperative to improve the security of electronic signatures. In this paper, we study online electronic signature recognition technology, using neural model and sparse classification to construct an efficient and accurate recognition model. Experiments show that the model is effective and the number of training samples affects the recognition accuracy. This paper provides a new approach for the development of this technique. When the training samples are greater than 1300, the recognition accuracy is stable at 95%. This research has certain theoretical and practical significance, and promotes the rapid development of online electronic signature recognition.

Automatic hepatic tumor segmentation in intra-operative ultrasound: a supervised deep-learning approach.

Training and evaluation of the performance of a supervised deep-learning model for the segmentation of hepatic tumors from intraoperative US (iUS) images, with the purpose of improving the accuracy of tumor margin assessment during liver surgeries and the detection of lesions during colorectal surgeries. In this retrospective study, a U-Net network was trained with the nnU-Net framework in different configurations for the segmentation of CRLM from iUS. The model was trained on B-mode intraoperative hepatic US images, hand-labeled by an expert clinician. The model was tested on an independent set of similar images. The average age of the study population was 61.9 ± 9.9 years. Ground truth for the test set was provided by a radiologist, and three extra delineation sets were used for the computation of inter-observer variability. The presented model achieved a DSC of 0.84 (), which is comparable to the expert human raters scores. The model segmented hypoechoic and mixed lesions more accurately (DSC of 0.89 and 0.88, respectively) than hyper- and isoechoic ones (DSC of 0.70 and 0.60, respectively) only missing isoechoic or >20mm in diameter (8% of the tumors) lesions. The inclusion of extra margins of probable tumor tissue around the lesions in the training ground truth resulted in lower DSCs of 0.75 (). The model can accurately segment hepatic tumors from iUS images and has the potential to speed up the resection margin definition during surgeries and the detection of lesion in screenings by automating iUS assessment.

18F-FDG PET/CT and radiolabeled leukocyte SPECT/CT imaging for the evaluation of cardiovascular infection in the multimodality context: ASNC Imaging Indications (ASNC I2) Series Expert Consensus Recommendations from ASNC, AATS, ACC, AHA, ASE, EANM, HRS, IDSA, SCCT, SNMMI, and STS.

This document on cardiovascular infection, including infective endocarditis, is the first in the American Society of Nuclear Cardiology Imaging Indications (ASNC I2) series to assess the role of radionuclide imaging in the multimodality context for the evaluation of complex systemic diseases with multi-societal involvement including pertinent disciplines. A rigorous modified Delphi approach was used to determine consensus clinical indications, diagnostic criteria, and an algorithmic approach to diagnosis of cardiovascular infection including infective endocarditis. Cardiovascular infection incidence is increasing and is associated with high morbidity and mortality. Current strategies based on clinical criteria and an initial echocardiographic imaging approach are effective but often insufficient in complicated cardiovascular infection. Radionuclide imaging with 18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) and single photon emission computed tomography/CT leukocyte scintigraphy can enhance the evaluation of suspected cardiovascular infection by increasing diagnostic accuracy, identifying extracardiac involvement, and assessing cardiac implanted device pockets, leads, and all portions of ventricular assist devices. This advanced imaging can aid in key medical and surgical considerations. Consensus diagnostic features include focal/multi-focal or diffuse heterogenous intense 18F-FDG uptake on valvular and prosthetic material, perivalvular areas, device pockets and leads, and ventricular assist device hardware persisting on non-attenuation corrected images. There are numerous clinical indications with a larger role in prosthetic valves, and cardiac devices particularly with possible infective endocarditis or in the setting of prior equivocal or non-diagnostic imaging. Illustrative cases incorporating these consensus recommendations provide additional clarification. Future research is necessary to refine application of these advanced imaging tools for surgical planning, to identify treatment response, and more.

An improved nonlinear anisotropic model with p(x)‐growth conditions applied to image restoration and enhancement

This work proposes a novel nonlinear parabolic equation with ‐growth conditions for image restoration and enhancement. Based on the generalized Lebesgue and Sobolev spaces with variable exponent, we demonstrate the well‐posedness of the proposed model. As a first result, we prove the existence of a weak solution to our model when the reaction term is bounded by a suitable function. Secondly, we use the approximations method to establish the existence of a nonnegative weak SOLA (Solution Obtained as Limit of Approximations) solution to the proposed model. We illustrate our theoretical results in the context of image denoising and enhancement. Specifically, we present various numerical implementations on a set of grayscale images. To further enrich these simulations, we assess the efficiency of the proposed model on several color images. The obtained numerical results strongly suggest that our model outperforms existing state‐of‐the‐art methods in terms of both visual and quantitative comparison, demonstrating superior efficiency and robustness in noise removal and contrast enhancement.

Evaluation of Effectiveness of Self-Supervised Learning in Chest X-Ray Imaging to Reduce Annotated Images.

A significant challenge in machine learning-based medical image analysis is the scarcity of medical images. Obtaining a large number of labeled medical images is difficult because annotating medical images is a time-consuming process that requires specialized knowledge. In addition, inappropriate annotation processes can increase model bias. Self-supervised learning (SSL) is a type of unsupervised learning method that extracts image representations. Thus, SSL can be an effective method to reduce the number of labeled images. In this study, we investigated the feasibility of reducing the number of labeled images in a limited set of unlabeled medical images. The unlabeled chest X-ray (CXR) images were pretrained using the SimCLR framework, and then the representations were fine-tuned as supervised learning for the target task. A total of 2000 task-specific CXR images were used to perform binary classification of coronavirus disease 2019 (COVID-19) and normal cases. The results demonstrate that the performance of pretraining on task-specific unlabeled CXR images can be maintained when the number of labeled CXR images is reduced by approximately 40%. In addition, the performance was significantly better than that obtained without pretraining. In contrast, a large number of pretrained unlabeled images are required to maintain performance regardless of task specificity among a small number of labeled CXR images. In summary, to reduce the number of labeled images using SimCLR, we must consider both the number of images and the task-specific characteristics of the target images.

Few‐shot learning for satellite characterisation from synthetic inverse synthetic aperture radar images

AbstractSpace situational awareness systems primarily focus on detecting and tracking space objects, providing crucial positional data. However, understanding the complex space domain requires characterising satellites, often involving estimation of bus and solar panel sizes. While inverse synthetic aperture radar allows satellite visualisation, developing deep learning models for substructure segmentation in inverse synthetic aperture radar images is challenging due to the high costs and hardware requirements. The authors present a framework addressing the scarcity of inverse synthetic aperture radar data through synthetic training data. The authors approach utilises a few‐shot domain adaptation technique, leveraging thousands of rapidly simulated low‐fidelity inverse synthetic aperture radar images and a small set of inverse synthetic aperture radar images from the target domain. The authors validate their framework by simulating a real‐case scenario, fine‐tuning a deep learning‐based segmentation model using four inverse synthetic aperture radar images generated through the backprojection algorithm from simulated raw radar data (simulated at the analogue‐to‐digital converter level) as the target domain. The authors results demonstrate the effectiveness of the proposed framework, significantly improving inverse synthetic aperture radar image segmentation across diverse domains. This enhancement enables accurate characterisation of satellite bus and solar panel sizes as well as their orientation, even when the images are sourced from different domains.

Noise level estimation using locality preserving natural image statistics

Natural images are known to have certain regular statistical properties. These properties get modified under any artificial change or distortion in natural images. Most common form of image degradation occurs in the form of noise. The amount of degradation in noisy images is measured by estimating the noise level. Many image processing applications such as denoising, restoration, segmentation, compression etc. use noise level information as a prior; inaccurate estimate of which may impact their performance. In this article, we explore natural image statistics in locality preserving transform domain. This property groups structurally similar images/image patches when projected in the transform domain. Image patches corrupted with similar noise level get projected close by in the locality preserving domain and show consistent coefficient behaviour. In particular, we use Two Dimensional Orthogonal Locality Preserving Projection (2DOLPP) as the domain transformation technique. 2DOLPP basis, representing natural images, are learnt in advance from a set of clean images, thereby reducing the computational time significantly. Features based on natural image statistics are extracted from 2DOLPP domain representation of input image patches. Mapping from feature space to noise level is carried out using support vector regression. The proposed noise estimation approach is at par with or surpasses the state-of-the-art techniques with much less computational time. Performance of this approach is stable across a wide range of noise levels and independent of the image structure.