We present a new method for image reconstruction which replaces the projector in projected gradient descent (PGD) with convolutional neural network (CNN). CNNs trained as high-dimensional (image-to-image) regressors have recently been used to efficiently solve inverse problems imaging. However, these approaches lack feedback mechanism enforce that reconstructed is consistent measurements. This crucial problems, and more so biomedical imaging, where reconstructions are diagnosis. In our...

Abstract Mammography, or breast X-ray imaging, is the most widely used imaging modality to detect cancer and other diseases. Recent studies have shown that deep learning-based computer-assisted detection diagnosis (CADe/x) tools been developed support physicians improve accuracy of interpreting mammography. A number large-scale mammography datasets from different populations with various associated annotations clinical data introduced study potential methods in field radiology. With aim...

Downsampling of signals living on a general weighted graph is not as trivial regular where we can simply keep every other samples. In this paper propose simple, yet effective downsampling scheme in which the underlying approximated by maximum spanning tree (MST) that naturally defines multiresolution. This MST-based method significantly outperforms two previous schemes, coloring-based and SVD-based, both random specific graphs terms computations partition efficiency quantified cuts. The...

Recent years have experienced phenomenal growth in computer-aided diagnosis systems based on machine learning algorithms for anomaly detection tasks the medical image domain. However, performance of these greatly depends quality labels since subjectivity a single annotator might decline certainty datasets. In order to alleviate this problem, aggregating from multiple radiologists with different levels expertise has been established. particular, under reliance their own biases and proficiency...

Advanced deep learning (DL) algorithms may predict the patient's risk of developing breast cancer based on Breast Imaging Reporting and Data System (BI-RADS) density standards. Recent studies have suggested that combination multi-view analysis improved overall exam classification. In this paper, we propose a novel DL approach for BI-RADS assessment mammograms. The proposed first deploys convolutional networks feature extraction each view separately. extracted features are then stacked fed...

ABSTRACT Mammography, or breast X-ray, is the most widely used imaging modality to detect cancer and other diseases. Recent studies have shown that deep learning-based computer-assisted detection diagnosis (CADe/x) tools been developed support physicians improve accuracy of interpreting mammography. However, published datasets mammography are either limited on sample size digitalized from screen-film (SFM), hindering development CADe/x which based full-field digital (FFDM). To overcome this...

The next step in immersive communication beyond video from a single camera is object-based free viewpoint video, which the capture and compression of dynamic object such that it can be reconstructed viewed an arbitrary viewpoint. moving human body particularly useful subclass for relevant to both telepresence entertainment. In this paper, we compress sequences by applying recently developed Graph Wavelet Filter Banks time-varying geometry color signals living on mesh representation body....

Laser Optoacoustic Ultrasonic Imaging System Assembly (LOUISA-3D) was developed in response to demand of diagnostic radiologists for an advanced screening system the breast improve on low sensitivity x-ray based modalities mammography and tomosynthesis dense heterogeneous specificity magnetic resonance imaging. It is our working hypothesis that co-registration quantitatively accurate functional images vasculature microvasculature, anatomical morphological structures will provide a clinically...

We study a dynamic spectrum access situation where, in each time slot, single cognitive agent decides to either stay idle or one of the N frequency channels based on its sensing whole spectrum. The are occupied vacant according independent nonidentical 2-state Markov chains. prove that optimal policy can easily be found if state transition probabilities all known. When has no knowledge about channel model, we propose use deep Q-learning method learn state-action value function determines an...

Interpretation of chest radiographs (CXR) is a difficult but essential task for detecting thoracic abnormalities. Recent artificial intelligence (AI) algorithms have achieved radiologist-level performance on various medical classification tasks. However, only few studies addressed the localization abnormal findings from CXR scans, which in explaining image-level to radiologists. Additionally, actual impact AI diagnostic radiologists clinical practice remains relatively unclear. To bridge...

A fully automated system for interpreting abdominal computed tomography (CT) scans with multiple phases of contrast enhancement requires an accurate classification the phases. Current approaches to classify CT are commonly based on three-dimensional (3D) convolutional neural network (CNN) high computational complexity and latency. This work aims at developing validating a precise, fast multiphase classifier recognize three main types in scans.We propose this study novel method that uses...

The National Academy of Engineering recently identified 14 grand challenges for engineering in the 21st century (www.engineeringchallenges. org). We believe that continuing advances ubiquitous sensing, processing, and computing provide potential to tackle two these challenges: specifically, enhancing virtual reality advancing personalized learning.

We propose a data-driven algorithm for the Bayesian estimation of stochastic processes from noisy observations. The primary statistical properties sought signal are specified by penalty function (i.e., negative logarithm prior probability density function). Our alternating direction method multipliers (ADMM) based approach translates task into successive applications proximal mapping function. Capitalizing on this direct link, we define operator as parametric spline curve and optimize...

Image augmentation techniques have been widely investigated to improve the performance of deep learning (DL) algorithms on mammography classification tasks. Recent methods proved efficiency image data deficiency or imbalance issues. In this paper, we propose a novel transparency strategy boost Breast Imaging Reporting and Data System (BI-RADS) scores mammogram classifiers. The proposed approach utilizes Region Interest (ROI) information generate more high-risk training examples for breast...

We propose a vector space approach for relighting Lambertian convex object with distant light source, whose crucial task is the decomposition of reflectance function into albedos (or reflection coefficients) and lightings based on set images same its 3-D model. Making use fact that functions are well approximated by low-dimensional linear subspace spanned first few spherical harmonics, this inverse problem can be formulated as matrix factorization, in which basis encoded harmonic <i>S</i>. A...

Chest radiograph (CXR) interpretation is critical for the diagnosis of various thoracic diseases in pediatric patients. This task, however, error-prone and requires a high level understanding radiologic expertise. Recently, deep convolutional neural networks (D-CNNs) have shown remarkable performance interpreting CXR adults. However, there lack evidence indicating that D-CNNs can recognize accurately multiple lung pathologies from scans. In particular, development diagnostic models detection...

Frame permutation quantization (FPQ) is a new vector technique using finite frames. In FPQ, encoded source code to quantize its frame expansion. This means that the encoding partial ordering of expansion coefficients. Compared ordinary coding, FPQ produces greater number possible rates and higher maximum rate. Various representations for partitions induced by are presented reconstruction algorithms based on linear programming quadratic derived. Reconstruction canonical dual also studied,...

Tyrosine is mainly degraded in the liver by a series of enzymatic reactions. Abnormal expression tyrosine catabolic enzyme aminotransferase (TAT) has been reported patients with hepatocellular carcinoma (HCC). Despite this, aberration metabolism not investigated cancer development. In this work, we conduct comprehensive cross-platform study to obtain foundation for discoveries potential therapeutics and preventative biomarkers HCC. We explore data from The Cancer Genome Atlas (TCGA), Gene...

The Poisson summation formula (PSF), which relates the sampling of an analog signal with periodization its Fourier transform, plays a key role in classical theory. In current forms, is only applicable to limited class signals L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> . However, this assumption on too strict for many applications processing that require non-decaying signals. paper we generalize PSF functions living weighted...