This paper presents a new supervised method for vessel segmentation in retinal images. remolds the task of as problem cross-modality data transformation from image to map. A wide and deep neural network with strong induction ability is proposed model transformation, an efficient training strategy presented. Instead single label center pixel, can output map all pixels given patch. Our approach outperforms reported state-of-the-art methods terms sensitivity, specificity accuracy. The result...

Purpose . In this study, we proposed an automated deep learning (DL) method for head and neck cancer (HNC) gross tumor volume (GTV) contouring on positron emission tomography-computed tomography (PET-CT) images. Materials Methods PET-CT images were collected from 22 newly diagnosed HNC patients, of whom 17 (Database 1) 5 2) two centers, respectively. An oncologist a radiologist decided the gold standard GTV manually by consensus. We developed convolutional neural network (DCNN) trained based...

Colorectal cancer is the third common in United States and most colorectal associated with polyps. In hospital, colonoscopy a way to detect polyps segmentation plays an important role diagnosis prevention of digestive system related diseases. Therefore, there pressure-need for polyp computer-aided help doctors diagnosis. this paper, we propose new, end-to-end fully convolutional neural network structure segmenting This method can directly output prediction map same size as original image...

Purpose: To automate the segmentation of retinal layers, we propose DeepRetina, a method based on deep neural networks. Methods: DeepRetina uses improved Xception65 to extract and learn characteristics layers. The Xception65-extracted feature maps are inputted an atrous spatial pyramid pooling module obtain multiscale information. This information is then recovered capture clearer layer boundaries in encoder-decoder module, thus completing auto-segmentation optical coherence tomography (OCT)...

To evaluate the application of a deep learning architecture, based on convolutional neural network (CNN) technique, to perform automatic tumor segmentation magnetic resonance imaging (MRI) for nasopharyngeal carcinoma (NPC).In this prospective study, 87 MRI containing regions were acquired from newly diagnosed NPC patients. These augmented >60,000 images. The proposed CNN is composed two phases: feature representation and scores map reconstruction. We designed stepwise scheme train our...

Flexible and low-voltage photosensors with high near-infrared (NIR) sensitivity are critical for realization of interacting humans robots environments by thermal imaging or night vision techniques. In this work, we the first time develop an easy cost-effective process to fabricate flexible ultrathin electrolyte-gated organic phototransistors (EGOPTs) transparent nanocomposite membranes high-conductivity silver nanowire (AgNW) networks large-capacitance iontronic films. A responsivity 1.5 ×...

The scale space-based method has been recently studied for multispectral alignment; however, due to the significant intensity difference between image pairs, there are usually not enough keypoint correspondences found, and robustness of alignment tends be compromised. In this letter, we attempt improve performance from following two aspects: 1) avoid boundary blurring Gaussian space, adopt nonlinear space explore more keypoints with potential being correctly matched, 2) a robust feature...

Objectives The purpose of this study was to develop an automatic tracking method for the muscle cross‐sectional area (CSA) on ultrasound (US) images using a convolutional neural network (CNN). performance proposed evaluated and compared with that state‐of‐the art segmentation method. Methods A real‐time US image sequence obtained from rectus femoris during voluntary contraction. CNN built segment calculate CSA in each frame. This consisted 2 stages: feature extraction score map...

Purpose To automate the detection and identification of visible components in feces for early diagnosis gastrointestinal diseases, we propose FecalNet, a method using multiple deep neural networks. Methods FecalNet uses ResNet152 residual network to extract learn characteristics fecal microscopic images, acquire feature maps combination with pyramid network, apply full convolutional classify locate components, implement improved focal loss function reoptimize classification results. This...

The analysis of semen plays an important role in male fertility evaluations. Computer-aided Sperm Analysis systems have been working on providing more accurate information about the sperm motility and quantity. However, existing detection algorithms which segment sperms according to grey levels are not able preclude bright non-sperm objects, like round cells. contribution this paper is a solution problem. use Gaussian-modeling method makes our algorithm filter non-target objects. We also...

ABSTRACT Protein S100B is a clinically useful non‐invasive biomarker for brain cell damage. A rapid chemiluminescence immunoassay (CLIA) in human serum has been developed. Fluorescein isothiocyanate (FITC) and N ‐(aminobutyl)‐ ‐(ethylisoluminol) (ABEI) are used to label two different monoclonal antibodies of anti‐S100B. combines with labeled can form sandwiched immunoreaction. simplified separation procedure based on the use magnetic particles (MPs) that were coated anti‐FITC antibody...

Computer-aided sperm analysis (CASA) can detect male infertility by classifying and counting motility. However, a commonly used threshold segmentation method is prone to loss of target when impurities collide, resulting in detection errors. In this regard, we propose based on the Gaussian mixture model for cell tracking recognition classification. Through background modeling, motile be sensitively tracked, which effectively avoid caused collision impurities. Experimental results show that...

Purpose Urinary particles are particularly important parameters in clinical urinalysis, especially for the diagnosis of nephropathy. Therefore, it is highly to precisely detect urinary setting. However, artificial microscopy subjective and time consuming, various previous detection algorithms lack adequate accuracy. In this study, a method proposed analysis based on deep learning. Methods We used seven cellular components (i.e., erythrocytes, leukocytes, epithelial, low‐transitional...

Retinal vessel delineation is a hot research topic owing to its importance in lot of clinic application. Several methods have been proposed the past decades. Here we will present new supervised method for retinal segmentation. The designed explore complex relationship between images and their corresponding label maps. Specifically, order build model describing direct transformation from image map, introduce deep convolutional neural network (abbreviation as CNN), which has strong enough...

Normalized cross correlation (NCC) has been widely used to match control points (CP) in image alignment. This method will produce a lot of incorrect matches owing the significant difference intensity between multispectral pairs, and furthermore, it is very computationally expensive handle rotational displacement. letter presents using rotation-invariant distance CPs; local descriptor matrix built describe each CP, fast Fourier transform introduced compute matrices. The computational load...

The purpose of retinal image registration is to establish the coherent correspondences between multi-model for applying into ophthalmological surgery. Vessel landmarks detection in vital step registration. In this paper, a novel approach proposed, firstly, deep learning technology used vessel segmentation generate probability map image, which more reliable optimizing feature image. Secondly, we detect using multi-scale Hessian response on Compared traditional methods, results show that our...