Tiny Regions Research Articles

Rational partitioning of spectral feature space for effective clustering of massive spectral image data

We have successfully proposed and demonstrated a clustering method that overcomes the “needle-in-a-haystack problem” (finding minuscule important regions from massive spectral image data sets). The needle-in-a-haystack problem is of central importance in the characterization of materials since in bulk materials, the properties of a very tiny region often dominate the entire function. To solve this problem, we propose that rational partitioning of the spectral feature space in which spectra are distributed, or defining of the decision boundaries for clustering, can be performed by focusing on the discrimination limit defined by the measurement noise and partitioning the space at intervals of this limit. We verified the proposed method, applied it to actual measurement data, and succeeded in detecting tiny (~ 0.5%) important regions that were difficult for human researchers and other machine learning methods to detect in discovering unknown phases. The ability to detect these crucial regions helps in understanding materials and designing more functional materials.

ECF-Net: Enhanced, Channel-Based, Multi-Scale Feature Fusion Network for COVID-19 Image Segmentation

Accurate segmentation of COVID-19 lesion regions in lung CT images aids physicians in analyzing and diagnosing patients’ conditions. However, the varying morphology and blurred contours of these regions make this task complex and challenging. Existing methods utilizing Transformer architecture lack attention to local features, leading to the loss of detailed information in tiny lesion regions. To address these issues, we propose a multi-scale feature fusion network, ECF-Net, based on channel enhancement. Specifically, we leverage the learning capabilities of both CNN and Transformer architectures to design parallel channel extraction blocks in three different ways, effectively capturing diverse lesion features. Additionally, to minimize irrelevant information in the high-dimensional feature space and focus the network on useful and critical information, we develop adaptive feature generation blocks. Lastly, a bidirectional pyramid-structured feature fusion approach is introduced to integrate features at different levels, enhancing the diversity of feature representations and improving segmentation accuracy for lesions of various scales. The proposed method is tested on four COVID-19 datasets, demonstrating mIoU values of 84.36%, 87.15%, 83.73%, and 75.58%, respectively, outperforming several current state-of-the-art methods and exhibiting excellent segmentation performance. These findings provide robust technical support for medical image segmentation in clinical practice.

An Image Identification System for Agricultural Items Based on Machine Vision

The purpose of this work was to use machine learning classification models and hyperspectral camera technologies to create a model of surface damage to garlic. 140 of the 184 garlic plants of which 44 were used for test validation were pre-treated for surface damage. First, we examined the data in ENVI under various damage scenarios using the normalised vegetation index (NDVI) approach. 579 pixels were then chosen for the training of the logistic regression model. Finally, we used 54 garlic bulbs to practically validate the model. Although tiny regions could not be precisely identified, the mouldy portion of the garlic’s surface could be identified using the NDVI technique. 90% accuracy was attained using the 90% classification model constructed using the logistic regression approach. Garlic’s surface damage, even at first mild ones, was correctly identified. The creation of this model for identifying garlic damage lowers the cost of detecting garlic damage and broadens the use of hyperspectral technologies in garlic detection.

Emerging functions of the mitochondria-ER-lipid droplet three-way junction in coordinating lipid transfer, metabolism, and storage in cells.

Over the past two decades, we have witnessed a growing appreciation for the importance of membrane contact sites (CS) in facilitating direct communication between organelles. CS are tiny regions where the membranes of two organelles meet but do not fuse and allow the transfer of metabolites between organelles, playing crucial roles in the coordination of cellular metabolic activities. The significant advancements in imaging techniques and molecular and cell biology research have revealed that CS are more complex than what originally thought, and as they are extremely dynamic, they can remodel their shape, composition, and functions in accordance with metabolic and environmental changes and can occur between more than two organelles. Here, we describe how recent studies led to the identification of a three-way mitochondria-ER-lipid droplet CS and discuss the emerging functions of these contacts in maintaining lipid storage, homeostasis, and balance. We also summarize the properties and functions of key protein components localized at the mitochondria-ER-lipid droplet interface, with a special focus on lipid transfer proteins. Understanding tripartite CS is essential for unraveling the complexities of inter-organelle communication and cooperation within cells.

Extracting a functional representation from a dictionary for non-rigid shape matching

Shape matching is a fundamental problem in computer graphics with many applications. Functional maps translate the point-wise shape-matching problem into its functional counterpart and have inspired numerous solutions over the last decade. Nearly all the solutions based on functional maps rely on the eigenfunctions of the Laplace–Beltrami Operator (LB) to describe the functional spaces defined on the surfaces and then convert the functional correspondences into point-wise correspondences. However, this final step is often error-prone and inaccurate in tiny regions and protrusions, where the energy of LB does not uniformly cover the surface. We propose a new functional basis Principal Components of a Dictionary (PCD) to address such intrinsic limitation. PCD constructs an orthonormal basis from the Principal Component Analysis (PCA) of a dictionary of functions defined over the shape. These dictionaries can target specific properties of the final basis, such as achieving an even spreading of energy. Our experimental evaluation compares seven different dictionaries on established benchmarks, showing that PCD is suited to target different shape-matching scenarios, resulting in more accurate point-wise maps than the LB basis when used in the same pipeline. This evidence provides a promising alternative for improving correspondence estimation, confirming the power and flexibility of functional maps.

CTCD-Net: A Cross-Layer Transmission Network for Tiny Road Crack Detection

Crack detection is essential for the safety maintenance of road infrastructure. However, there are two major limitations to detecting road cracks accurately: (1) tiny cracks usually possess less distinctive features and are more susceptible to noises, so they are apt to be ignored; (2) most existing methods extract cracks with coarse and thicker boundaries, which needs further improvement. To address the above limitations, we propose CTCD-Net: a Cross-layer Transmission network for tiny road Crack Detection. Firstly, we propose a cross-layer information transmission module based on an attention mechanism to compensate for the disadvantage of unobvious features of tiny cracks. With this module, the feature information from upper layers is transmitted to the next one, layer by layer, to achieve information enhancement and emphasize the feature representation of tiny crack regions. Secondly, we design a boundary refinement block to further improve the accuracy of crack boundary locations, which refines boundaries by learning the residuals between the label images and the interim coarse maps. Extensive experiments conducted on three crack datasets demonstrate the superiority and effectiveness of the proposed CTCD-Net. In particular, our method largely improves the accuracy and completeness of tiny crack detection.

Texture and semantic convolutional auto‐encoder for anomaly detection and segmentation

AbstractAnomaly detection is a challenging task, especially detecting and segmenting tiny defect regions in images without anomaly priors. Although deep encoder‐decoder‐based convolutional neural networks have achieved good anomaly detection results, existing methods operate uniformly on all extracted image features without considering disentangling these features. To fully explore the texture and semantic information of images, A novel unsupervised anomaly detection method is proposed. Specifically, discriminative features are extracted from images by using a deep pre‐trained network, where shallow and deep features are aggregated into texture and semantic modules, respectively. Then, a feature fusion module is developed to interactively enable feature information in two different modules. The texture and semantic segmentation results are obtained by comparing the texture features and semantic features before and after reconstruction, respectively. Finally, an anomaly segmentation module is designed to generate anomaly detection results by integrating the results of the texture and semantic modules by setting a threshold. Experimental results on benchmark datasets for anomaly detection demonstrate that our proposed method can efficiently and effectively detect anomalies, outperforming some state‐of‐the‐art methods by 2.7% and 0.6% in classification and segmentation.

Electrochemo-poromechanics of Ionic Polymer Metal Composites: Towards the Accurate Finite Element Modelling of Actuation and Sensing

Ionic polymer metal composites (IPMCs) consist of an electroactive polymeric membrane, which is plated with metal electrodes and includes a fluid phase of ions in a solvent, whose diffusion allows for actuation and sensing applications. We build on a previous finite-deformation theory of our group that accounts for the cross-diffusion of ions and solvent and couples the mass balances of these species with the stress balance and the Gauss law. Here, we abandon the assumption that the fluid phase is a dilute solution, with benefits on both modelling and computation. A reliable finite element (FE) implementation of electrochemomechanical theories for IPMCs is challenging because the IPMC behaviour is governed by boundary layers (BLs) occurring in tiny membrane regions adjacent to the electrodes, where steep gradients of species concentrations occur. We address this issue by adopting the generalized FE method to discretise the BLs. This allows unprecedented analyses of the IPMC behaviour since it becomes possible to explore it under external actions consistent with applications, beside obtaining accurate predictions with a reasonable computational cost. Hence, we provide novel results concerning the influence of the membrane permittivity on the species profiles at the BLs. Additionally, by leveraging on the mobility matrix, we establish that the initial peak deflection in actuation strongly depends on the constitutive equations for the species transport and discuss the predictions of some experimental results from the literature. Overall, we demonstrate the potential of the proposed model to be an effective tool for the thorough analysis and design of IPMCs.

IHA-Net: An automatic segmentation framework for computer-tomography of tiny intracerebral hemorrhage based on improved attention U-net

Intracerebral hemorrhage (ICH) is a serious category of head injury, which means bleeding occurs inside the skull, and in many cases leads to a high disability or mortality. Therefore, accurate and rapid hematoma region segmentation is of great significance for ICH diagnosis and treatment. Nevertheless, even with a powerful deep neural network, there are still certain limitations in the case of tiny ICH regions. In this paper, a novel model based on convolutional neural network for the segmentation of hematoma regions in brain Computed Tomography images of patients with ICH is developed. Specifically, in order to prevent the degradation of tiny lesion regions due to repeated down-sampling, we also proposed a Residual Hybrid Atrous Convolution strategy. Furthermore, a multi-object function for joint optimization is introduced to ease the severe pixel imbalanced problem, which is of benefit to training procedure. In addition, to avoid gradient disappearance and slow convergence, intermediate supervision with a lightweight head is utilized during the training. The performance of developed segmentation algorithm for Intracranial hemorrhage is analyzed by using Dice coefficient, Jaccard index, Sensitivity and Accuracy. From both quantitative and visual results, we found that the proposed model outperformed the previous methods and achieved a satisfying segmentation performance for Intracranial hemorrhage test in both internal testing data (Dice of 0.725, Jaccard of 0.605) and external clinical data (Dice of 0.869, Jaccard of 0.774), which indicates that proposed model has better segmentation effect and potential clinical prospect.

AMD-Net: Automatic subretinal fluid and hemorrhage segmentation for wet age-related macular degeneration in ocular fundus images

Age-related macular degeneration (AMD) is a chronic, progressively degenerative disorder of the macular. Wet AMD is responsible for 80 to 90 percent of all AMD-related blindness. Nowadays, ophthalmologists can diagnose wet AMD based on subretinal fluid and hemorrhage in ocular fundus images. However, this diagnosis process is time-consuming and influenced by subjective factors. Deep learning methods have achieved significant improvements for segmentation tasks in the retinal image, but only a few studies have focused on the subretinal fluid and hemorrhage lesion segmentation of wet AMD. This paper proposed AMD-Net, a novel U-Net architecture, which can segment the subretinal fluid and hemorrhage lesions of the wet AMD in ocular fundus images. The proposed AMD-Net consists of three components: encoder feature fusion unit (EFFU), skip connection block (SKB), and decoder attention block (DAB). The AMD-Net adopts the EFFU to extract and fuse the multi-scale features. And with the attention module, the EFFU can assign a higher weight for discriminative features. To reduce the semantic gap between the encoder features and decoder features, the AMD-Net designs an SKB. Furthermore, the AMD-Net introduces a DAB, a new module based on the UNet 3+ decoder, designed to leverage the local context information and enhance the contribution of high-level semantic features in tiny regions segmentation. We evaluate the proposed method on a private dataset collected by ourselves. Our model achieves 67.18% subretinal fluid Dice, 66.51% hemorrhage Dice, and 77.07% average Dice on this dataset. The experimental results indicate that our proposed AMD-Net is superior to the state-of-the-art deep learning methods.

Attention‐based convolutional neural network deep learning approach for robust malware classification

AbstractRecently, transforming windows files into images and its analysis using machine learning and deep learning have been considered as a state‐of‐the art works for malware detection and classification. This is mainly due to the fact that image‐based malware detection and classification is platform independent, and the recent surge of success of deep learning model performance in image classification. Literature survey shows that convolutional neural network (CNN) deep learning methods are successfully employed for image‐based windows malware classification. However, the malwares were embedded in a tiny portion in the overall image representation. Identifying and locating these affected tiny portions is important to achieve a good malware classification accuracy. In this work, a multi‐headed attention based approach is integrated to a CNN to locate and identify the tiny infected regions in the overall image. A detailed investigation and analysis of the proposed method was done on a malware image dataset. The performance of the proposed multi‐headed attention‐based CNN approach was compared with various non‐attention‐CNN‐based approaches on various data splits of training and testing malware image benchmark dataset. In all the data‐splits, the attention‐based CNN method outperformed non‐attention‐based CNN methods while ensuring computational efficiency. Most importantly, most of the methods show consistent performance on all the data splits of training and testing and that illuminates multi‐headed attention with CNN model's generalizability to perform on the diverse datasets. With less number of trainable parameters, the proposed method has achieved an accuracy of 99% to classify the 25 malware families and performed better than the existing non‐attention based methods. The proposed method can be applied on any operating system and it has the capability to detect packed malware, metamorphic malware, obfuscated malware, malware family variants, and polymorphic malware. In addition, the proposed method is malware file agnostic and avoids usual methods such as disassembly, de‐compiling, de‐obfuscation, or execution of the malware binary in a virtual environment in detecting malware and classifying malware into their malware family.

QS ride app

The quickly increasing number of vehicles associated commuter’s results in a rise in holdup and also the issues related to it. It needs the study of different measures to scale down the amount of vehicles movement daily, particularly single-occupant vehicles. Many researches conducted discovered that carpooling are often an economical resolution to alleviate the pressure thanks to traffic. The carpooling may be a system by that an individual offers his or her personal vehicle to at least one or additional commuters who have similar destinations or routes. Carpooling is effective solution to cut back the traffic. It consists of accelerating the occupancy of vehicles by occupying empty seats within the vehicle effectively. This system isn't enough as so much as tiny regions are thought-about and it's quantifiability issue. Our system will offer a secure interface for connecting drivers and commuters to share the ride whereby we have a tendency to are reaching to provide this service in small regions also in order that daily commuters like office-goers and students may also be expedited and might contribute in reducing traffic so saving environment.

Electroweak phase transition in the presence of hypermagnetic field and the generation of gravitational waves

We investigate the effects of a large-scale background hypermagnetic field on the electroweak phase transition (EWPT). We propose a model in which an effective weak angle varies during the EWPT and upon its use we show that, although for the majority of the parameter space the phase transition is a crossover, there are tiny regions in which the phase transition occurs in two steps and can be first-order. We obtain all of the important quantities characterizing the details of the first-order phase transition, including the latent heat, transition temperature and duration. We then calculate the gravitational wave energy spectrum generated during the first-order part of the EWPT and find that, for strong enough background hypermagnetic fields, these signals can be detected by the Ultimate-DECIGO interferometer.

Weakly supervised attention-based models using activation maps for citrus mite and insect pest classification

Citrus juices and fruits are commodities with great economic potential in the international market, but productivity losses caused by mites and other pests are still far from being a good mark. Despite the integrated pest mechanical aspect, only a few works on automatic classification have handled images with orange mite characteristics, which means tiny and noisy regions of interest. Attention-based models have gained prominence in deep learning research, and, along with weakly supervised learning algorithms, they have improved tasks performed with some label restrictions. In this context, this work proposes an attention-based activation map approach developed to improve the classification of tiny regions called Two-Weighted Activation Mapping, which also produces locations using feature map scores learned from class labels. We apply our method in a two-stage network process called Attention-based Multiple Instance Learning Guided by Saliency Maps. We analyze the proposed approach in two challenging datasets, the Citrus Pest Benchmark, captured directly in the field using magnifying glasses, and the Insect Pest, a large pest image benchmark. We evaluate and compare our models with weakly supervised methods, such as Attention-based Deep MIL and WILDCAT. The results show that our classifier is superior to literature methods, surpassing them in all scenarios by at least 16 percentage points. Moreover, our approach infers bounding box locations for salient insects, even training without location labels.

Flood risk mapping and analysis: A case study of Andheri Khola catchment, Sindhuli district, Nepal

Nepal is one of the world's most vulnerable countries to a variety of risks, including severe floods. This could result in the loss of lives and property, the relocation of people, the damage of physical infrastructures, homes, and the disruption of people's socioeconomic functions and the country's economy in a variety of ways. River flooding is caused by heavy monsoon rainfall, weak geology, unplanned infrastructure construction along the embankments, and mining in upstream riverbeds. The Andheri Khola (river) is a tributary of the Sunkoshi River that frequently floods, affecting the inhabitants along the way. In Nepal, little effort has been made to comprehend the flood risk in tiny catchment regions such as Andheri Khola, despite the fact that this sort of small catchment is affecting Nepal's numerous new rising towns and urban areas in many ways. To analyse the one-dimensional flood plain, HEC-RAS, Ras Mapper, and ArcGIS were used. The WECS/DHM approach was used to estimate flood frequency in different return periods in order to determine the flood risk in the research area. The study finds that the floods of 2, 50, 100, and 1000 years return periods cover a maximum of 35, 41.9, 42.7, and 49.72 hectare area, respectively. The majority of the flooded sites had water depths of more than 3 meters. More than 70% of the sandy area in the study region is prone to flooding. Furthermore, the cultivated areas are located in a low to moderate risk area.

Functional connectome of arousal and motor brainstem nuclei in living humans by 7 Tesla resting-state fMRI

Brainstem nuclei play a pivotal role in many functions, such as arousal and motor control. Nevertheless, the connectivity of arousal and motor brainstem nuclei is understudied in living humans due to the limited sensitivity and spatial resolution of conventional imaging, and to the lack of atlases of these deep tiny regions of the brain. For a holistic comprehension of sleep, arousal and associated motor processes, we investigated in 20 healthy subjects the resting-state functional connectivity of 18 arousal and motor brainstem nuclei in living humans. To do so, we used high spatial-resolution 7 Tesla resting-state fMRI, as well as a recently developed in-vivo probabilistic atlas of these nuclei in stereotactic space. Further, we verified the translatability of our brainstem connectome approach to conventional (e.g. 3 Tesla) fMRI. Arousal brainstem nuclei displayed high interconnectivity, as well as connectivity to the thalamus, hypothalamus, basal forebrain and frontal cortex, in line with animal studies and as expected for arousal regions. Motor brainstem nuclei showed expected connectivity to the cerebellum, basal ganglia and motor cortex, as well as high interconnectivity. Comparison of 3 Tesla to 7 Tesla connectivity results indicated good translatability of our brainstem connectome approach to conventional fMRI, especially for cortical and subcortical (non-brainstem) targets and to a lesser extent for brainstem targets. The functional connectome of 18 arousal and motor brainstem nuclei with the rest of the brain might provide a better understanding of arousal, sleep and accompanying motor functions in living humans in health and disease.

Attention deep learning‐based large‐scale learning classifier for Cassava leaf disease classification

AbstractCassava is a rich source of carbohydrates, and it is vulnerable to virus diseases. Literature survey shows that the image recognition and integrated deep learning approach is successfully employed for Cassava leaf disease classification. Mostly, transfer learning based on a convolutional neural network (CNN) models were successfully applied for Cassava leaf disease classification. However, existing approaches are not effective in identifying the tiny portion of the disease in the overall leaf area. Identifying and focussing on regions affected by the disease is vital to achieving a good classification accuracy. An attention‐based approach is integrated into pretrained CNN‐based EfficientNet models to locate and identify the tiny infected regions in Cassava leaf. Penultimate layer features of attention‐based EfficientNet models such as A_EfficientNetB4, A_EfficientNetB5, and A_EfficientNetB6 were extracted. Next, the dimensionality of the extracted features was reduced using kernel principal component analysis. The reduced features were fused and passed into a stacked ensemble meta‐classifier for Cassava leaf disease classification. A stacked ensemble meta‐classifier is a two‐stage approach in which the first stage employs random forest and support vector machine (SVM) for prediction followed by logistic regression for classification. Detailed investigation and analysis of the proposed method, attention, and non‐attention‐based approaches with CNN pretrained models were tested using a publicly available benchmark dataset of Cassava leaf disease images. The proposed method achieved better performances in all experiments than several existing methods as well as various attention and non‐attention‐based CNN pretrained models. The proposed approach can be used as a deployable tool for Cassava leaf disease classification in agricultural field.

A Multimodal Affinity Fusion Network for Predicting the Survival of Breast Cancer Patients.

Accurate survival prediction of breast cancer holds significant meaning for improving patient care. Approaches using multiple heterogeneous modalities such as gene expression, copy number alteration, and clinical data have showed significant advantages over those with only one modality for patient survival prediction. However, existing survival prediction methods tend to ignore the structured information between patients and multimodal data. We propose a multimodal data fusion model based on a novel multimodal affinity fusion network (MAFN) for survival prediction of breast cancer by integrating gene expression, copy number alteration, and clinical data. First, a stack-based shallow self-attention network is utilized to guide the amplification of tiny lesion regions on the original data, which locates and enhances the survival-related features. Then, an affinity fusion module is proposed to map the structured information between patients and multimodal data. The module endows the network with a stronger fusion feature representation and discrimination capability. Finally, the fusion feature embedding and a specific feature embedding from a triple modal network are fused to make the classification of long-term survival or short-term survival for each patient. As expected, the evaluation results on comprehensive performance indicate that MAFN achieves better predictive performance than existing methods. Additionally, our method can be extended to the survival prediction of other cancer diseases, providing a new strategy for other diseases prognosis.

Comparison between Optical coherence tomography Angiography and fundus Fluorescein Angiography in cases of Diabetic Maculopathy

Background: Diabetic macular edoema (DME) in industrialised nations has become the most common cause of vision loss. The major cause of serious vision impairment in diabetes individuals is also diabetic macular ischemia (DMI). It is also one of the world's top causes of avoidable blindness, as the WHO states. The objective of this research was to evaluate the state of a capillary bed in diabetic maculopathy in the macular region and the connection between findings with OCT-Angiography and Fundus Fluorescein angiography. Methods: This research consisted of 40 eyes of 25 individuals aged 49 to 70 years with a clinically significant maculo-diabetic edoema as determined by the Early Treatment Diabetic Retinopathy (ETDRS) study without additional eye conditions. Outcomes and conclusion: This research shows the angiographical characteristics of OCT diabetic retinopathy. Some of their benefits over fluorescence angiography include the fast acquirement time, lack of requirement for an intravenous dye, the detection of tiny neovascular tufts and regions of capillary drop-out not masked by leaking. This technique may be helpful for regular diabetic retinopathy monitoring. It has limited field of vision, relatively poor resolution and difficulties identifying micro-aneurysms over a wide region. The lack of the retinal peripheral technology for measuring its vascular health is also a significant barrier to the adoption of OCT-A as the only means of evaluating diabetic retinopathy patients.

Performance Evaluation Towards Automatic Building and Road Detection Technique for High-Resolution Remote Sensing Images

Building and road detection from the high-resolution remote sensing images has many applications in a wide range of areas including urban design, real-estate management, and disaster relief. Extracting buildings and roads from the remote sensing images have been performed by human experts manually, but it takes too much time and the cost of labor to build is also very high. So, an automated system that can emulate a human operator is desired. Our goal is to develop a system for automatically detecting buildings and roads directly from high-resolution satellite images. Therefore, we propose Internal Gray Variance (IGV) to detect the buildings and roads in the urban areas. First, the satellite images are enhanced by using morphological operators, which enhance the edges of the objects. Then multiseed-based clustering technique detects the building and road edges using the variance in the gray levels. To reduce the false alarm, tiny regions that are improbable to be buildings and roads are separated. Finally, by using the adaptive threshold-based segmentation technique, the buildings are segmented and the road network is extracted based on supervised directional homogeneity. Finally, we evaluate our system on a large-scale road and building detection data sets that are publicly available.