Optical Coherence Tomography (OCT) is becoming one of the most important modalities for noninvasive assessment retinal eye diseases. As number acquired OCT volumes increases, automating image analysis increasingly relevant. In this paper, we propose a surrogate-assisted classification method to classify images automatically based on convolutional neural networks (CNNs). Image denoising first performed reduce noise. Thresholding and morphological dilation are applied extract masks. The...

Recently, many methods based on hand-designed convolutional neural networks (CNNs) have achieved promising results in automatic retinal vessel segmentation. However, these CNNs remain constrained capturing vessels complex fundus images. To improve their segmentation performance, tend to parameters, which may lead overfitting and high computational complexity. Moreover, the manual design of competitive is time-consuming requires extensive empirical knowledge. Herein, a novel automated method,...

Automated optic disk (OD) detection plays an important role in developing a computer aided system for eye diseases. In this paper, we propose algorithm the OD based on structured learning. A classifier model is trained Then, use to achieve edge map of OD. Thresholding performed map, thus binary image obtained. Finally, circle Hough transform carried out approximate boundary by circle. The proposed has been evaluated three public datasets and obtained promising results. results (an area...

Convolutional neural networks (CNNs) are effective tools for regression tasks. However, their black-box nature limits applicability in high-impact and high-risk In this paper, a novel method is proposed to identify particular patterns an image that can make the output of CNN model equal specified value, thereby helping users understand behaviours CNNs. Specifically, method, set binary filters first randomly initialized. A genetic algorithm then employed evolve such value when taking filtered...

Advanced prediction of the daily incidence COVID-19 can aid policy making on prevention disease spread, which profoundly affect people's livelihood. In previous studies, predictions were investigated for single or several countries and territories.

Internet search engine data, such as Google Trends, was shown to be correlated with the incidence of COVID-19, but only in several countries. We aim develop a model from small number countries predict epidemic alert level all worldwide.The "interest over time" and by region" Trends data Coronavirus, pneumonia, six COVID symptom-related terms were searched. The daily COVID-19 10 January 23 April 2020 202 retrieved World Health Organization. Three levels defined. Ten weeks' 20 used for...

PurposeThis study aims to estimate the regional choroidal thickness from color fundus images convolutional neural networks in different network structures and task learning models.Method1276 photos their corresponding values healthy subjects were obtained Topcon DRI Triton optical coherence tomography machine. Initially, ten commonly used deployed identify most accurate model, which was subsequently selected for further training. This model then employed combination with single-, multiple-,...

Automated detection of blood vessel structures is becoming a crucial interest for better management vascular disease. In this paper, we propose an algorithm segmentation in digital retinal images based on integral channel features and random forests. the first stage, preprocessing performed to obtain candidate pixels vessels, then host simple are extracted each channels. Furthermore forests used classify as vessels or not. Finally, postprocessing applied fill pixel gaps classified vessels....

This paper proposes a novel and simple unsupervised vessel segmentation algorithm using fundus images. At first, the green channel of image is preprocessed to extract binary after isotropic undecimated wavelet transform, another from morphologically reconstructed image. Secondly, two initial images are extracted according region features for connected regions in Next, common both as major vessels. Then all remaining pixels processed with skeleton extraction linear iterative clustering....

Multi-objective evolutionary algorithms (MOEAs) have been widely used in solving multi-objective optimization problems. A great number of the-state-of-art MOEAs proposed. These can be classified into the following categories: decomposition-based, domination-based, indicator-based, and probability-based methods. Among them, first four categories belong to non-model based methods, while fifth one is considered model-based method, which machine learning techniques are often build models....

Automatic and accurate estimation of choroidal thickness plays a very important role in computer-aided system for eye diseases. One the most common methods automatic is segmentation-based methods, which boundaries choroid are first detected from optical coherence tomography (OCT) images. The then computed based on boundaries. A shortcoming that estimating precision greatly depends segmentation results. To avoid dependence step, this paper, we propose direct method convolutional neural...

Tobacco plants recognizing and counting accurately is very important in the tobacco plant management. In this paper, we propose an algorithm for which consists of four main steps: first, apply unmanned aircraft to acquire images. Second, image converted into Lab color space, then b channel space processed based on morphological reconstruction. Third, candidate regions might contain are extracted channel. Finally, SVM (Support Vector Machine) employed classify as or not. The proposed method...

In our previous work, by combining the Hilbert scan with symbol grouping method, efficient run-length-based entropy coding was developed, and high-efficiency image compression algorithms based on were obtained. However, 2-D curves, which are a critical part of above-mentioned coding, defined squares side length being powers 2, i.e., 2n, while subband is normally rectangle arbitrary sizes. It not straightforward to modify curve from lengths 2n an rectangle. this short article, we provide...

Detecting the optic disk (OD) is very important in fundus image analysis. In this paper, we propose a new OD detection algorithm consisting of four main steps: first, obtaining sub-image which includes from based on saliency map; second, generating super-pixel with simple linear iterative clustering (SLIC) algorithm; third, classifying into or non-OD AdaBoost fourth, fitting detected region circle active geometric shape model. The proposed method has been evaluated Digital Retinal Images for...

A multichannel biomedical signal acquisition system based on core processor STM32F405 of the Cortex-M4 is presented in this paper. The paper describes mainly from perspective hardware and software design. design includes three major parts - power conversion circuit, circuit processing circuit. For module, data collection transmission, digital filters are research focus. It demonstrated experimental tests that can extract collect a variety human physiological signals complex environments. In...

Deep convolutional neural networks (CNNs) have been widely used for fundus image classification and achieved very impressive performance. However, the explainability of CNNs is poor because their black-box nature, which limits application in clinical practice. In this paper, we propose a novel method to search discriminative regions increase confidence features specific category, thereby helping users understand an are important CNN make particular prediction. proposed method, set...

Optic Disk (OD) detection plays an important role for fundus image analysis. In this paper, we propose algorithm detecting OD mainly based on a classifier model trained by structured learning. Then use the to achieve edge map of OD. Thresholding is performed obtain binary image. Finally, circle Hough transform carried out approximate boundary circle. The proposed has been evaluated public database and obtained promising results. results (an area overlap Dices coefficients 0.8636 0.9196,...

Deformable image registration can obtain dynamic information about images, which is of great significance in medical analysis. The unsupervised deep learning method quickly achieve high accuracy without labels. However, these methods generally suffer from uncorrelated features, poor ability to register large deformations and details, unnatural deformation fields. To address the issues above, we propose an multi-scale correlation iterative network (SearchMorph). In proposed network, introduce...

Cone-beam computed tomography (CBCT) often has suboptimal image quality compared with standard fan-beam CT due to increased sensitivity scattering and motion artifacts. In radiation treatment of cancers, the convenience in acquisition as it is fully incorporated into station, CBCT plays a significant role plan re-evaluation adaptation adaptive therapy but limited by quality. this paper, we propose deep convolutional neural network (DCNN) based method for head neck (HN) enhancement, where...

In recent years, convolutional neural networks (CNNs) have achieved excellent performance in pavement crack segmentation tasks. However, CNNs usually need high computational cost and large storage. As a result, CNN slimming is important for embedded systems with limited resources. Inspired by the channel attention techniques CNNs, we propose an effective method to slim based on mechanism. The proposed first removes unimportant filters their connecting feature maps according scales....

<sec> <title>BACKGROUND</title> Advanced prediction of the daily incidence COVID-19 can aid policy making on prevention disease spread, which profoundly affect people's livelihood. In previous studies, predictions were investigated for single or several countries and territories. </sec> <title>OBJECTIVE</title> We aimed to develop models that be applied real-time activity in all individual territories worldwide. <title>METHODS</title> Data infoveillance data (search volume via Google Trends)...