Abstract Cardiovascular disease, primarily caused by atherosclerotic plaque formation, is a significant health concern. The early detection of these plaques crucial for targeted therapies and reducing the risk cardiovascular diseases. This study presents PlaqueNet, solution segmenting coronary artery from computed tomography angiography (CCTA) images. For feature extraction, advanced residual net module was utilized, which integrates deepwise optimization into network branches, enhances...

The aim of this study is to analyze the risk factors associated with development adenomatous and malignant polyps in gallbladder. Adenomatous gallbladder are considered precancerous have a high likelihood progressing into malignancy. Preoperatively, distinguishing between benign polyps, challenging. Therefore, objective develop neural network model that utilizes these accurately predict nature polyps. This predictive can be employed differentiate before surgery, enhancing diagnostic...

Detection of retinal lesions like micro-aneurysms and exudates are important for the clinical diagnosis diabetes retinopathy. The traditional subjective judgments by clinicians dependent on their experience can be subject to lack consistency therefore a quantification method is worthwhile. In this study, 10 moderate non-proliferative retinopathy (NPDR) patients severe NPDR ones were retrospectively selected as cohort. Mathematical morphological methods used automatic segmentation lesions....

Abstract Purpose: We aim to develop a back-propagation artificial neural network (BP-ANN) improved by priori knowledge and compare its efficacy with other methods in early diabetic retinopathy (DR) detection. Methods: A total of 240 fundus images, composed 120 early-stage DR normal were obtained the same 45° field view camera, macula at center, as cohort for further training. All retinal images processed, features such blood vessel width tortuosity semi-automatically extracted. An BP-ANN was...

Topic detection and tracking (TDT) algorithms have long been developed for the discovery of topics. However, most existing TDT suffer from paying less attention to: (1) temporal distance between a pair topics; (2) mutual effect highly correlated topic terms. In this paper, we proposed novel approach by applying hierarchical clustering on constructed concept graph (HCCG), which is able to solve aforementioned shortcomings simultaneously. approach, first defined as well behavior curve. Then,...

In this paper, a method for removing ultrasound medical image noise is proposed. This improves the original median filter by designing suit of directional templates. order to obtain local directions, filter, which combines Gaussion blur with direction parameter, designed. Then, templates based on directions mentioned above are used as reducing spark noise. At last, finial denoising result each pixel obtained from filtered or noise-polluted judging their differences. If difference between two...

Community detection has long been an important yet challenging task to analyze complex networks with a focus on detecting topological structures of graph data. Essentially, real-world data is generally heterogeneous which dynamically varies over time, and this invalidates most existing community approaches. To cope these issues, paper proposes the temporal-heterogeneous convolutional (THGCN) detect communities using learnt feature representations set temporal graphs. Particularly, we first...

The optic disc (OD) is one of the important anatomic structures on retina, changes which shape and area may indicate disease processes, thus needs computerized quantification assistance. In this study, we proposed a self-adaptive distance

Change detection is a critical preprocessing step of visual perception with broad prospects. Its primary challenge to identify all the meaningful changes from target image source image, which observed same scene and has different perspective as well. A robust change method involving graph matching geometric constraints proposed in this paper. Maximum common sub-graph applied for alleviating risk suboptimal results are used remove possible mistaken results. Detection real-world scenes respect...

Community detection has long been an important yet challenging task to analyze complex networks with a focus on detecting topological structures of graph data. Essentially, real-world data contains various features, node and edge types which dynamically vary over time, this invalidates most existing community approaches. To cope these issues, paper proposes the heterogeneous-temporal convolutional (HTGCN) detect communities from hetergeneous temporal graphs. Particularly, we first design...

This paper presents an algorithm to identify the handwritten and printed texts among document images. The characteristic of stroke thickness is used a kind calculating method designed for this feature. proposed method, which clearly defined easily realized, calculates feature by counting edge pixels in neighborhood. Document images are generally divided into text lines or characters. However, line character not conducive judgment between distinction. too rough small. Using characteristics,...

We employ an edge co-occurrence matrix (ECM) to distinguish handwritten and machine printed text without resorting line or word information. The ECM is a modified (CM) on images. First, the whole image divided into overlapping range blocks with fixed size. Then, ECMs are abstracted from these blocks. only counts co-occurring edges connected each other its up direction part most distribution. liner Support Vector Machine (SVM) used classify features. Because of similarities neighboring...

Community detection is an extremely important task for complex network analysis. There still remains a challenge of how to improve the performance community in real-world scenario. Some researchers think that content networks helpful identify communities, and also focus on combining topology with node contents, alongside eradication inauspicious performance. Furthermore, often sparse, which reflected lack capability represent communities. To address above problems, this study identifies...

Visual Simultaneous Localization and Mapping (VSLAM) occupies a pivotal position in the robotics field. The loop closure detection module, which is related to quality of mapping accuracy positioning, an indispensable part SLAM. In recent years, neural networks are often used replace feature extraction detection. These methods can extract more helpful features, but effect not significant. this paper, improved siamese network proposed view as whole improve real-time performance Firstly,...

In recent years, deep learning technology has greatly improved the ability to recognize human activities in daily life.Traditional approaches activity recognition often rely on handcrafted features obtained through a heuristic process from single sensing modality.However, techniques overcome many of these limitations by automatically extracting multi-modal modalities, enabling more powerful recognition.In this paper, we propose learningbased multi-channel architecture that combines two...

Registration is the most fundmental and important task when processing point-clouds. In correspondence-based alignment of point-clouds, using point features to match correspondences can result in a very high outlier rate. Current methods still have low efficiency, recognition rates accuracy. this paper, we propose point-cloud descriptor based on SIFT feature points method for eliminating outliers similar triangles. Firstly, algorithm used extract point-cloud, then FPFH calculated. After...