RESUMEN

Retinal vessel segmentation is crucial for diagnosing and monitoring various eye diseases such as diabetic retinopathy, glaucoma, and hypertension. In this study, we examine how sharpness-aware minimization (SAM) can improve RF-UNet's generalization performance. RF-UNet is a novel model for retinal vessel segmentation. We focused our experiments on the digital retinal images for vessel extraction (DRIVE) dataset, which is a benchmark for retinal vessel segmentation, and our test results show that adding SAM to the training procedure leads to notable improvements. Compared to the non-SAM model (training loss of 0.45709 and validation loss of 0.40266), the SAM-trained RF-UNet model achieved a significant reduction in both training loss (0.094225) and validation loss (0.08053). Furthermore, compared to the non-SAM model (training accuracy of 0.90169 and validation accuracy of 0.93999), the SAM-trained model demonstrated higher training accuracy (0.96225) and validation accuracy (0.96821). Additionally, the model performed better in terms of sensitivity, specificity, AUC, and F1 score, indicating improved generalization to unseen data. Our results corroborate the notion that SAM facilitates the learning of flatter minima, thereby improving generalization, and are consistent with other research highlighting the advantages of advanced optimization methods. With wider implications for other medical imaging tasks, these results imply that SAM can successfully reduce overfitting and enhance the robustness of retinal vessel segmentation models. Prospective research avenues encompass verifying the model on vaster and more diverse datasets and investigating its practical implementation in real-world clinical situations.

Asunto(s)

RESUMEN

CoCrNi medium-entropy alloys (MEAs) have attracted extensive attention and research because of their superior mechanical properties, such as higher ductility, strength, and toughness. This study uses molecular dynamics (MD) simulations to investigate the cutting behavior of a gradient nanograined (GNG) CoCrNi MEA. Moreover, it explores the influence of relative tool sharpness and rake angle on the cutting process. The results show that an increase in the average grain size of the GNG samples leads to a decrease in the average resultant cutting force, as predicted by the Hall-Petch relationship. The deformation behavior shows that grain boundaries are crucial in inhibiting the propagation of strain and stress. As the average grain size of the GNG sample increases, the range of shear strain distribution and average von Mises stress decreases. Moreover, the cutting chips become thinner and longer. The subsurface damage is limited to a shallow layer at the surface. Since thermal energy is generated in the high grain boundary density, the temperature of the contact zone between the substrate and the cutting tool increases as the GNG size decreases. The cutting chips removed from the GNG CoCrNi MEA substrates will transform into a mixed structure of face-centered cubic and hexagonally close-packed phases. The sliding and twisting of grain boundaries and the merging of grains are essential mechanisms for polycrystalline deformation. Regarding the cutting parameters, the average resultant force, the material accumulation, and the chip volume increase significantly with the increase in cutting depth. In contrast to sharp tools, which mainly use shear deformation, blunt tools remove material by plowing, and the cutting force increases with the increase in cutting-edge radius and negative rake angle.

RESUMEN

High-resolution and wide-swath (HRWS) synthetic aperture radar (SAR) imaging with azimuth multi-channel always suffers from channel phase and amplitude errors. Compared with spatial-invariant error, the range-dependent channel phase error is intractable due to its spatial dependency characteristic. This paper proposes a novel parameterized channel equalization approach to reconstruct the unambiguous SAR imagery. First, a linear model is established for the range-dependent channel phase error, and the sharpness of the reconstructed Doppler spectrum is used to measure the unambiguity quality. Furthermore, the intrinsic relationship between the channel phase errors and the sharpness is revealed, which allows us to estimate the optimal parameters by maximizing the sharpness of the reconstructed Doppler spectrum. Finally, the results from real-measured data show that the suggested method performs exceptionally for ambiguity suppression in HRWS SAR imaging.

RESUMEN

In recent years, distributed stochastic algorithms have become increasingly useful in the field of machine learning. However, similar to traditional stochastic algorithms, they face a challenge where achieving high fitness on the training set does not necessarily result in good performance on the test set. To address this issue, we propose to use of a distributed network topology to improve the generalization ability of the algorithms. We specifically focus on the Sharpness-Aware Minimization (SAM) algorithm, which relies on perturbation weights to find the maximum point with better generalization ability. In this paper, we present the decentralized stochastic sharpness-aware minimization (D-SSAM) algorithm, which incorporates the distributed network topology. We also provide sublinear convergence results for non-convex targets, which is comparable to consequence of Decentralized Stochastic Gradient Descent (DSGD). Finally, we empirically demonstrate the effectiveness of these results in deep networks and discuss their relationship to the generalization behavior of SAM.

Asunto(s)

RESUMEN

Background. Thoracoabdominal MRI is limited by respiratory motion, especially in populations who cannot perform breath-holds. One approach for reducing motion blurring in radially-acquired MRI is respiratory gating. Straightforward 'hard-gating' uses only data from a specified respiratory window and suffers from reduced SNR. Proposed 'soft-gating' reconstructions may improve scan efficiency but reduce motion correction by incorporating data with nonzero weight acquired outside the specified window. However, previous studies report conflicting benefits, and importantly the choice of soft-gated weighting algorithm and effect on image quality has not previously been explored. The purpose of this study is to map how variable soft-gated weighting functions and parameters affect signal and motion blurring in respiratory-gated reconstructions of radial lung MRI, using neonates as a model population.Methods. Ten neonatal inpatients with respiratory abnormalities were imaged using a 1.5 T neonatal-sized scanner and 3D radial ultrashort echo-time (UTE) sequence. Images were reconstructed using ungated, hard-gated, and several soft-gating weighting algorithms (exponential, sigmoid, inverse, and linear weighting decay outside the period of interest), with %Nprojrepresenting the relative amount of data included. The apparent SNR (aSNR) and motion blurring (measured by the maximum derivative of image intensity at the diaphragm, MDD) were compared between reconstructions.Results. Soft-gating functions produced higher aSNR and lower MDD than hard-gated images using equivalent %Nproj, as expected. aSNR was not identical between different gating schemes for given %Nproj. While aSNR was approximately linear with %Nprojfor each algorithm, MDD performance diverged between functions as %Nprojdecreased. Algorithm performance was relatively consistent between subjects, except in images with high noise.Conclusion. The algorithm selection for soft-gating has a notable effect on image quality of respiratory-gated MRI; the timing of included data across the respiratory phase, and not simply the amount of data, plays an important role in aSNR. The specific soft-gating function and parameters should be considered for a given imaging application's requirements of signal and sharpness.

Asunto(s)

RESUMEN

BACKGROUND: Gliomas differ from meningiomas in their margins, most of which are not separated from the surrounding tissue by a distinct interface. PURPOSE: To characterize the margins of gliomas quantitatively based on the margin sharpness coefficient (MSC) is significant for clinical judgment and invasive analysis of gliomas. MATERIAL AND METHODS: The data for this study used magnetic resonance image (MRI) data from 67 local patients and 15 open patients to quantify the intensity of changes in the glioma margins of the brain using MSC. The accuracy of MSC was assessed by consistency analysis and Bland-Altman test analysis, as well as invasive correlations using receiver operating characteristic (ROC) and Spearman correlation coefficients for subjects. RESULTS: In grading the tumors, the mean MSC values were significantly lower for high-grade gliomas (HGG) than for low-grade gliomas (LGG). The concordance correlation between the measured gradient and the actual gradient was high (HGG: 0.981; LGG: 0.993), and the Bland-Altman mean difference at the 95% confidence interval (HGG: -0.576; LGG: 0.254) and the limits of concordance (HGG: 5.580; LGG: 5.436) indicated no statistical difference. The correlation between MSC and invasion based on the margins of gliomas showed an AUC of 0.903 and 0.911 for HGG and LGG, respectively. The mean Spearman correlation coefficient of the MSC versus the actual distance of invasion was -0.631 in gliomas. CONCLUSION: The relatively low MSC on the blurred margins and irregular shape of gliomas may help in benign-malignant differentiation and invasion prediction of gliomas and has potential application for clinical judgment.

Asunto(s)

RESUMEN

PURPOSE: To investigate the potential of combining Compressed Sensing (CS) and a newly developed AI-based super resolution reconstruction prototype consisting of a series of convolutional neural networks (CNN) for a complete five-minute 2D knee MRI protocol. METHODS: In this prospective study, 20 volunteers were examined using a 3T-MRI-scanner (Ingenia Elition X, Philips). Similar to clinical practice, the protocol consists of a fat-saturated 2D-proton-density-sequence in coronal, sagittal and transversal orientation as well as a sagittal T1-weighted sequence. The sequences were acquired with two different resolutions (standard and low resolution) and the raw data reconstructed with two different reconstruction algorithms: a conventional Compressed SENSE (CS) and a new CNN-based algorithm for denoising and subsequently to interpolate and therewith increase the sharpness of the image (CS-SuperRes). Subjective image quality was evaluated by two blinded radiologists reviewing 8 criteria on a 5-point Likert scale and signal-to-noise ratio calculated as an objective parameter. RESULTS: The protocol reconstructed with CS-SuperRes received higher ratings than the time-equivalent CS reconstructions, statistically significant especially for low resolution acquisitions (e.g., overall image impression: 4.3 ±â€¯0.4 vs. 3.4 ±â€¯0.4, p < 0.05). CS-SuperRes reconstructions for the low resolution acquisition were comparable to traditional CS reconstructions with standard resolution for all parameters, achieving a scan time reduction from 11:01 min to 4:46 min (57 %) for the complete protocol (e.g. overall image impression: 4.3 ±â€¯0.4 vs. 4.0 ±â€¯0.5, p < 0.05). CONCLUSION: The newly-developed AI-based reconstruction algorithm CS-SuperRes allows to reduce scan time by 57% while maintaining unchanged image quality compared to the conventional CS reconstruction.

Asunto(s)

RESUMEN

Deep learning models have gained prominence in human activity recognition using ambient sensors, particularly for telemonitoring older adults' daily activities in real-world scenarios. However, collecting large volumes of annotated sensor data presents a formidable challenge, given the time-consuming and costly nature of traditional manual annotation methods, especially for extensive projects. In response to this challenge, we propose a novel AttCLHAR model rooted in the self-supervised learning framework SimCLR and augmented with a self-attention mechanism. This model is designed for human activity recognition utilizing ambient sensor data, tailored explicitly for scenarios with limited or no annotations. AttCLHAR encompasses unsupervised pre-training and fine-tuning phases, sharing a common encoder module with two convolutional layers and a long short-term memory (LSTM) layer. The output is further connected to a self-attention layer, allowing the model to selectively focus on different input sequence segments. The incorporation of sharpness-aware minimization (SAM) aims to enhance model generalization by penalizing loss sharpness. The pre-training phase focuses on learning representative features from abundant unlabeled data, capturing both spatial and temporal dependencies in the sensor data. It facilitates the extraction of informative features for subsequent fine-tuning tasks. We extensively evaluated the AttCLHAR model using three CASAS smart home datasets (Aruba-1, Aruba-2, and Milan). We compared its performance against the SimCLR framework, SimCLR with SAM, and SimCLR with the self-attention layer. The experimental results demonstrate the superior performance of our approach, especially in semi-supervised and transfer learning scenarios. It outperforms existing models, marking a significant advancement in using self-supervised learning to extract valuable insights from unlabeled ambient sensor data in real-world environments.

Asunto(s)

RESUMEN

Sharpness aware minimization (SAM) optimizer has been extensively explored as it can generalize better for training deep neural networks via introducing extra perturbation steps to flatten the landscape of deep learning models. Integrating SAM with adaptive learning rate and momentum acceleration, dubbed AdaSAM, has already been explored empirically to train large-scale deep neural networks without theoretical guarantee due to the triple difficulties in analyzing the coupled perturbation step, adaptive learning rate and momentum step. In this paper, we try to analyze the convergence rate of AdaSAM in the stochastic non-convex setting. We theoretically show that AdaSAM admits a O(1/bT) convergence rate, which achieves linear speedup property with respect to mini-batch size b. Specifically, to decouple the stochastic gradient steps with the adaptive learning rate and perturbed gradient, we introduce the delayed second-order momentum term to decompose them to make them independent while taking an expectation during the analysis. Then we bound them by showing the adaptive learning rate has a limited range, which makes our analysis feasible. To the best of our knowledge, we are the first to provide the non-trivial convergence rate of SAM with an adaptive learning rate and momentum acceleration. At last, we conduct several experiments on several NLP tasks and the synthetic task, which show that AdaSAM could achieve superior performance compared with SGD, AMSGrad, and SAM optimizers.

Asunto(s)

RESUMEN

Halal and kosher slaughter have given the utmost importance to the sharpness of knives during the slaughter of animals. A sharp knife of appropriate dimension (blade length) makes slaughter less painful during neck severance and facilitates desirable bleeding. The role of knife sharpness has not been given due credit from an animal welfare perspective and is likely ignored by the people involved in slaughterhouses. A neat, clean, and efficient neck cut by an extremely sharp knife reduces the pain. It improves the bleeding out, thus making animals unconscious early without undergoing unnecessary pain and stress. It also helps in improving meat quality and food safety. A slight incremental improvement in knife sharpness could significantly improve the animal welfare, productivity, efficiency, and safety of meat plant workers. The present review critically analyzed the significance of knife sharpness in religious slaughter by reducing stress and pain and improving meat quality and food safety. The objective quantification of knife sharpness, proper regular training of slaughterers, and slow slaughter rate are the challenges faced by the meat industry.

RESUMEN

Differentiable Architecture Search (DARTS) is a simple yet efficient Neural Architecture Search (NAS) method. During the search stage, DARTS trains a supernet by jointly optimizing architecture parameters and network parameters. During the evaluation stage, DARTS discretizes the supernet to derive the optimal architecture based on architecture parameters. However, recent research has shown that during the training process, the supernet tends to converge towards sharp minima rather than flat minima. This is evidenced by the higher sharpness of the loss landscape of the supernet, which ultimately leads to a performance gap between the supernet and the optimal architecture. In this paper, we propose Self-Distillation Differentiable Neural Architecture Search (SD-DARTS) to alleviate the discretization gap. We utilize self-distillation to distill knowledge from previous steps of the supernet to guide its training in the current step, effectively reducing the sharpness of the supernet's loss and bridging the performance gap between the supernet and the optimal architecture. Furthermore, we introduce the concept of voting teachers, where multiple previous supernets are selected as teachers, and their output probabilities are aggregated through voting to obtain the final teacher prediction. Experimental results on real datasets demonstrate the advantages of our novel self-distillation-based NAS method compared to state-of-the-art alternatives.

Asunto(s)

RESUMEN

Fundus images are widely used in the screening and diagnosis of eye diseases. Current classification algorithms for computer-aided diagnosis in fundus images rely on large amounts of data with reliable labels. However, the appearance of noisy labels degrades the performance of data-dependent algorithms, such as supervised deep learning. A noisy label learning framework suitable for the multiclass classification of fundus diseases is presented in this paper, which combines data cleansing (DC), adaptive negative learning (ANL), and sharpness-aware minimization (SAM) modules. Firstly, the DC module filters the noisy labels in the training dataset based on the prediction confidence. Then, the ANL module modifies the loss function by choosing complementary labels, which are neither the given labels nor the labels with the highest confidence. Moreover, for better generalization, the SAM module is applied by simultaneously optimizing the loss and its sharpness. Extensive experiments on both private and public datasets show that our method greatly promotes the performance for classification of multiple fundus diseases with noisy labels.

Asunto(s)

RESUMEN

The cutting performance of steel blades is an eternal, attractive topic in the knife industry. It is a complicated process to cut up materials because it usually involves the contact mechanics of the material been cut, the geometry and roughness of the blade edge and the hardness and wear resistance of the blade steel. Therefore, a comprehensive analysis is required to evaluate the cutting performance of knife blades. In this study, such an analysis was conducted based on a quantitative model to describe the cutting depth of paper cards containing SiO2 particles by steel blades, and major contributing factors were summarized. The effect of the micro-geometries of blade edges was thoroughly discussed, and a geometry factor ξ for the micro-geometry of a blade edge was introduced into the model. The experimental results indicated that mechanical processing could produce a rough blade edge and a higher ξ value, accordingly. A similar effect was caused by the carbides in the martensitic steels for blades, and the ξ value was found to increase linearly with the volumetric fraction of the carbides. The extraordinary cutting behavior of the 3V blade implied that fine coherent carbides may result in an efficient improvement (40-50%) in the total cutting depth.

RESUMEN

New CMOS imaging sensor (CIS) techniques in smartphones have helped user-generated content dominate our lives over traditional DSLRs. However, tiny sensor sizes and fixed focal lengths also lead to more grainy details, especially for zoom photos. Moreover, multi-frame stacking and post-sharpening algorithms would produce zigzag textures and over-sharpened appearances, for which traditional image-quality metrics may over-estimate. To solve this problem, a real-world zoom photo database is first constructed in this paper, which includes 900 tele-photos from 20 different mobile sensors and ISPs. Then we propose a novel no-reference zoom quality metric which incorporates the traditional estimation of sharpness and the concept of image naturalness. More specifically, for the measurement of image sharpness, we are the first to combine the total energy of the predicted gradient image with the entropy of the residual term under the framework of free-energy theory. To further compensate for the influence of over-sharpening effect and other artifacts, a set of model parameters of mean subtracted contrast normalized (MSCN) coefficients are utilized as the natural statistics representatives. Finally, these two measures are combined linearly. Experimental results on the zoom photo database demonstrate that our quality metric can achieve SROCC and PLCC over 0.91, while the performance of single sharpness or naturalness index is around 0.85. Moreover, compared with the best tested general-purpose and sharpness models, our zoom metric outperforms them by 0.072 and 0.064 in SROCC, respectively.

RESUMEN

Cortical activity, as recorded via electroencephalography, has been linked to the refractive error of an individual. It is however unclear which optical metric modulates this response. Here, we measured simultaneously the brain activity and the retinal defocus of a visual stimulus perceived through several values of spherical blur. We found that, contrary to the existing literature on the topic, the cortical response as a function of the overcorrections follows a sigmoidal shape rather than the classical bell shape, with the inflection point corresponding to the subjective refraction and to the stimulus being in focus on the retina. However, surprisingly, the amplitude of the cortical response does not seem to be a good indicator of how much the stimulus is in or out of focus on the retina. Nonetheless, the defocus is not equivalent to the retinal image quality, nor is an absolute predictor of the visual performance of an individual. Simulations of the retinal image quality seem to be a powerful tool to predict the modulation of the cortical response with the refractive error.

RESUMEN

In this paper, a novel ultra-wideband UWB antenna element with triple-band notches is proposed. The proposed UWB radiator element operates from 2.03 GHz up to 15.04 GHz with triple rejected bands at the WiMAX band (3.28-3.8 GHz), WLAN band (5.05-5.9 GHz), and X-band (7.78-8.51 GHz). In addition, the radiator supports the Bluetooth band (2.4-2.483 GHz). Three different techniques were utilized to obtain the triple-band notches. An alpha-shaped coupled line with a stub-loaded resonator (SLR) band stop filter was inserted along the main feeding line before the radiator to obtain a WiMAX band notch characteristic. Two identical U-shaped slots were etched on the proposed UWB radiator to achieve WLAN band notch characteristics with a very high degree of selectivity. Two identical metallic frames of an octagon-shaped electromagnetic band gap structure (EBG) were placed along the main feeding line to achieve the notch characteristic with X-band satellite communication with high sharpness edges. A novel UWB multiple-input multiple-output (MIMO) radiator is proposed. The proposed UWB-MIMO radiator was fabricated on FR-4 substrate material and measured. The isolation between every two adjacent ports was below -20 dB over the FCC-UWB spectrum and the Bluetooth band for the four MIMO antennas. The envelope correlation coefficient (ECC) between the proposed antennas in MIMO does not exceed 0.05. The diversity gains (DG) for all the radiators are greater than 9.98 dB.

RESUMEN

Statistical effects of cortical metrics derived from standard T1- and T2-weighted magnetic resonance imaging (MRI) images, such as gray-white matter contrast (GWC), boundary sharpness coefficient (BSC), T1-weighted/T2-weighted ratio (T1w/T2w), and cortical thickness (CT), are often interpreted as representing or being influenced by intracortical myelin content with little empirical evidence to justify these interpretations. We first examined spatial correspondence with more biologically specific microstructural measures, and second compared between-marker age-related trends with the underlying hypothesis that different measures primarily driven by similar changes in myelo- and microstructural underpinnings should be highly related. Cortical MRI markers were derived from MRI images of 127 healthy subjects, aged 18-81, using cortical surfaces that were generated with the CIVET 2.1.0 pipeline. Their gross spatial distributions were compared with gene expression-derived cell-type densities, histology-derived cytoarchitecture, and quantitative R1 maps acquired on a subset of participants. We then compared between-marker age-related trends in their shape, direction, and spatial distribution of the linear age effect. The gross anatomical distributions of cortical MRI markers were, in general, more related to myelin and glial cells than neuronal indicators. Comparing MRI markers, our results revealed generally high overlap in spatial distribution (i.e., group means), but mostly divergent age trajectories in the shape, direction, and spatial distribution of the linear age effect. We conclude that the microstructural properties at the source of spatial distributions of MRI cortical markers can be different from microstructural changes that affect these markers in aging.

Asunto(s)

RESUMEN

In the digital heritage field, the accurate reproduction of hard-to-photograph items, such as daguerreotypes, ambrotypes, and tintypes, is an ongoing challenge. Industrial contactless sensors offer the potential to improve the quality of scanned images, but their capabilities and limitations have not been fully explored. In our research, a dataset of 48 scans was created using the hi-tech industrial contactless sensor CRUSE. Moreover, 3 rare original photographs were scanned in 16 different modes, the most suitable images were determined by specialists in the restoration, and validated through experiments involving eye-tracking, multiple computer vision, and image processing methods. Our study identified the Cruse scanning modes, which can be utilized to achieve the most accurate digital representation of scanned originals. Secondly, we proposed several methods for highlighting the degradation and minor scratches on photographs that otherwise might not be detected by the restorer's naked eye. Our findings belong to four overlapping areas, i.e., image understanding, digital heritage, information visualization, and industrial sensors research. We publish the entire dataset under the CC BY-NC 4.0 license. The CRUSE sensor shows promise as a tool for improving the quality of scanned images of difficult-to-photograph items. Further research is necessary to fully understand its capabilities and limitations in this context.

RESUMEN

This study focuses on the effect of the substrate temperature (TS) on the quality of VO2 thin films prepared by DC magnetron sputtering. TS was varied from 350 to 600 °C and the effects on the surface morphology, microstructure, optical and electrical properties of the films were investigated. The results show that TS below 500 °C favors the growth of V2O5 phase, whereas higher TS (≥500 °C) facilitates the formation of the VO2 phase. Optical characterization of the as-prepared VO2 films displayed a reduced optical transmittance (T˜) across the near-infrared region (NIR), reduced phase transition temperature (Tt), and broadened hysteresis width (ΔH) through the phase transition region. In addition, a decline of the luminous modulation (ΔT˜lum) and solar modulation (ΔT˜sol) efficiencies of the as-prepared films have been determined. Furthermore, compared with the high-quality films reported previously, the electrical conductivity (σ) as a function of temperature (T) reveals reduced conductivity contrast (Δσ) between the insulating and metallic phases of the VO2 films, which was of the order of 2. These outcomes indicated the presence of defects and unrelaxed lattice strain in the films. Further, the comparison of present results with those in the literature from similar works show that the preparation of high-quality films at TS lower than 650 °C presents significant challenges.

RESUMEN

Objectives: To evaluate whether applying image filters (smooth 3D+ and edge-2) improves image quality in coronary CT angiography (CCTA). Methods: Ninety patients (routine group) with suspected coronary artery diseases based on 16-cm wide coverage detector CT findings were retrospectively enrolled at a chest pain center from December 2019 to September 2021. Two image filters, smooth 3D+ and edge-2 available on the Advantage Workstation (AW) were subsequently applied to the images to generate the research group (SE group). Quantitative parameters, including CT value, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), image sharpness and image quality score, and diagnostic accuracy were compared between the two groups. Results: A total of 900 segments from 270 coronary arteries in 90 patients were analyzed. SNR, CNR, and image sharpness for vessels and image quality scores in the SE group were significantly better than those in the routine group (all p < 0.001). The SE group showed a slightly higher negative predictive value (NPV) on the left anterior descending artery and right coronary artery (RCA) stenosis evaluations, as well as total NPV. The SE group also showed slightly higher sensitivity and accuracy than the routine group on RCA stenosis evaluation. Conclusion: The use of an image filter combining smooth 3D+ and edge-2 on an AW could improve the image quality of CCTA and increase radiologists' diagnostic confidence.