Medical Imaging Detection Research Articles

Bilayer lead oxide X-ray photoconductor for lag-free operation

Polycrystalline Lead Oxide (poly-PbO) was considered one of the most promising photoconductors for the direct conversion X-ray medical imaging detectors due to its previous success in optical imaging, i.e., as an optical target in so-called Plumbicon video pick-up tubes. However, a signal lag which accompanies X-ray excitation, makes poly-PbO inapplicable as an X-ray-to-charge transducer in real-time X-ray imaging. In contrast, the recently synthesized Amorphous Lead Oxide (a-PbO) photoconductor is essentially lag-free. Here, we report on our approach to a PbO detector where a thin layer of a-PbO is combined with a thick layer of poly-PbO for lag-free operation. In the presented a-PbO/poly-PbO bilayer structure, the poly-PbO layer serves as an X-ray-to-charge transducer while the a-PbO acts as a lag prevention layer. The hole mobility in the a-PbO/poly-PbO bilayer structure was measured by photo-Charge Extraction by Linearly Increasing Voltage technique at different temperatures and electric fields to investigate charge transport properties. It was found that the hole mobility is similar to that in a-Se—currently the only commercially viable photoconductor for the direct conversion X-ray detectors. Evaluation of the X-ray temporal performance demonstrated complete suppression of signal lag, allowing operation of the a-PbO/poly-PbO detector in real-time imaging.

New lasers that fire terahertz beams could propel medical imaging and contraband detection

New lasers that fire terahertz beams could propel medical imaging and contraband detection

Musculoskeletal Abnormality Detection in Medical Imaging Using GnCNNr (Group Normalized Convolutional Neural Networks with Regularization)

Musculoskeletal abnormality detection serves as an advantage to the professionals in the medical domain and also serves as an assistance in the diagnosis as well as the treatment of the abnormalities. This paper mainly focuses on accurately detecting musculoskeletal abnormalities using various deep learning models and techniques. MURA dataset has been used for experimentation. MURA dataset has 14,863 images of finger, wrist, elbow, shoulder, forearm and hand which has been analyzed using deep learning models. In this research paper, authors have proposed GnCNNr model which utilizes group normalization, weight standardization and cyclic learning rate scheduler to enhance the accuracy, precision and other model interpretation metrics. The musculoskeletal abnormality has been detected by using various deep learning models. Accuracy and Cohen Kappa have been taken as the evaluation criteria. The highest accuracy of 85% and Cohen Kappa statistic of 0.698 was achieved by the GnCNNr model in comparison with the conventional deep learning methods like DenseNet, Inception, Inception v2 model.

Weakly and semi supervised detection in medical imaging via deep dual branch net

This study presents a novel deep learning architecture for multi-class classification and localization of abnormalities in medical imaging illustrated through experiments on mammograms. The proposed network combines two learning branches. One branch is for region classification with a newly added normal-region class. Second branch is region detection branch for ranking regions relative to one another. Our method enables detection of abnormalities at full mammogram resolution for both weakly and semi-supervised settings. A novel objective function allows for the incorporation of local annotations into the model. We present the impact of our schemes on several performance measures for classification and localization, to evaluate the cost effectiveness of the lesion annotation effort. Our evaluation was primarily conducted over a large multi-center mammography dataset of ~3,000 mammograms with various findings. The results for weakly supervised learning showed significant improvement compared to previous approaches. We show that the time consuming local annotations involved in supervised learning can be addressed by a weakly supervised method that can leverage a subset of locally annotated data. Weakly and semi-supervised methods coupled with detection can produce a cost effective and explainable model to be adopted by radiologists in the field.

Review Research of Medical Image Analysis Using Deep Learning

In modern globe, medical image analysis significantly participates in diagnosis process. In general, it involves five processes, such as medical image classification, medical image detection, medical image segmentation, medical image registration, and medical image localization. Medical imaging uses in diagnosis process for most of the human body organs, such as brain tumor, chest, breast, colonoscopy, retinal, and many other cases relate to medical image analysis using various modalities. Multi-modality images include magnetic resonance imaging, single photon emission computed tomography (CT), positron emission tomography, optical coherence tomography, confocal laser endoscopy, magnetic resonance spectroscopy, CT, X-ray, wireless capsule endoscopy, breast cancer, papanicolaou smear, hyper spectral image, and ultrasound use to diagnose different body organs and cases. Medical image analysis is appropriate environment to interact with automate intelligent system technologies. Among the intelligent systems deep learning (DL) is the modern one to manipulate medical image analysis processes and processing an image into fundamental components to extract meaningful information. The best model to establish its systems is deep convolutional neural network. This study relied on reviewing of some of these studies because of these reasons; improvements of medical imaging increase demand on automate systems of medical image analysis using DL, in most tested cases, accuracy of intelligent methods especially DL methods higher than accuracy of hand-crafted works. Furthermore, manually works need a lot of time compare to systematic diagnosis.

Detection of pulmonary nodules based on a multiscale feature 3D U-Net convolutional neural network of transfer learning.

A new computer-aided detection scheme is proposed, the 3D U-Net convolutional neural network, based on multiscale features of transfer learning to automatically detect pulmonary nodules from the thoracic region containing background and noise. The test results can be used as reference information for doctors to assist in the detection of early lung cancer. The proposed scheme is composed of three major steps: First, the pulmonary parenchyma area is segmented by various methods. Then, the 3D U-Net convolutional neural network model with a multiscale feature structure is built. The network model structure is subsequently fine-tuned by the transfer learning method based on weight, and the optimal parameters are selected in the network model. Finally, datasets are extracted to train the fine-tuned 3D U-Net network model to detect pulmonary nodules. The five-fold cross-validation method is used to obtain the experimental results for the LUNA16 and TIANCHI17 datasets. The experimental results show that the scheme not only has obvious advantages in the detection of medium and large-sized nodules but also has an accuracy rate of more than 70% for the detection of small-sized nodules. The scheme provides automatic and accurate detection of pulmonary nodules that reduces the overfitting rate and training time and improves the efficiency of the algorithm. It can assist doctors in the diagnosis of lung cancer and can be extended to other medical image detection and recognition fields.

GANs for medical image analysis.

Generative adversarial networks (GANs) and their extensions have carved open many exciting ways to tackle well known and challenging medical image analysis problems such as medical image de-noising, reconstruction, segmentation, data simulation, detection or classification. Furthermore, their ability to synthesize images at unprecedented levels of realism also gives hope that the chronic scarcity of labeled data in the medical field can be resolved with the help of these generative models. In this review paper, a broad overview of recent literature on GANs for medical applications is given, the shortcomings and opportunities of the proposed methods are thoroughly discussed, and potential future work is elaborated. We review the most relevant papers published until the submission date. For quick access, essential details such as the underlying method, datasets, and performance are tabulated. An interactive visualization that categorizes all papers to keep the review alive is available at http://livingreview.in.tum.de/GANs_for_Medical_Applications/.

Ultrawideband Antenna Combined with a Reconfigurable Stop-Band Filter for Medical Imaging Detection Applications

In this paper, a ultrawideband antenna combined with a reconfigurable stop-band filter for medical imaging detection applications is presented. The proposed antenna provides dual-band operation at 1.8–3.2 GHz and 4–6.2 GHz, which satisfies widely the bandwidth for ISM application systems.These bands are achieved by the introduction of an open square ring on the top side and a pair of square of CSRRs on the ground, respectively. Equivalent circuit of the filter has been presented using ADS simulator. Its performance results have been compared to those obtained with CST simulator. The results of both methods show reasonably good agreement in terms of the frequency of operation and the impedance bandwidth. The study of the antenna properties is also taken into account. The resulting design has a compact configuration and enables easy control of the two desired operating frequencies. This proposed antenna is matched with $$50\Omega $$ to allow an easier connection. To validate the proposed concept, a prototype was fabricated and measured. The experimental results are presented and discussed. The obtained results show a good agreement between experimental and simulation results.

Highly Sensitive Narrowband Si Photodetector With Peak Response at Around 1060 nm

Photodetection at a wavelength of about 1060 nm is very important for applications including medical imaging, optical communication, and light detection and ranging. In this article, a self-powered near-infrared light detector with a narrowband at around 1060 nm is realized based on a simple Si Schottky structure, in which the Ohmic and Schottky electrodes are configured on the front and rear surfaces of the Si substrate, respectively. The as-assembled device exhibits a tunable peak response near 1060 nm with a full width at half maximum of 107 nm, which could be due to the combined effect of the narrow photo-current generation and the self-filtering effect of the silicon substrate. At zero bias, a specific detectivity of ~1 x 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">11</sup> Jones and linear dynamic range about 101 dB are achieved, in spite of the weak absorption of Si at this wavelength. The external quantum efficiency can be improved to 135% under a low bias of -1 V, indicating the existence of gain mechanism during photodetection. Finally, it is also found that the asassembled near-infrared device shows excellent antiinterference capability during the photodetection process.These results corroborate that the present Si photodetector may find promising application in future near-infrared optoelectronic devices and systems.

Fully automated intracranial aneurysm detection and segmentation from digital subtraction angiography series using an end-to-end spatiotemporal deep neural network

BackgroundIntracranial aneurysms (IAs) are common in the population and may cause death.ObjectiveTo develop a new fully automated detection and segmentation deep neural network based framework to assist neurologists in evaluating...

Luminescence Efficiency of Cadmium Tungstate (CdWO4) Single Crystal for Medical Imaging Applications

Background: In this study, the light output of a cadmium tungstate (CdWO4) single crystal was measured under various X-ray radiographic energies. Methods: A CdWO4 single crystal (10 × 10 × 10 mm3) was exposed to X-rays in the 50–130 kVp range. Measurements were evaluated against published data for single crystals of equal dimensions (CaF2:Eu and Lu3Al5O12:Ce). Since the crystal was examined for application in medical imaging detectors, the emitted optical spectrum was classified with respect to the spectral compatibility of numerous commercial optical sensors. Results: The luminescence efficiency (LE) was found to constantly increase with X-ray energy and was higher than that of CaF2:Eu for energies above 90 kVp. However, the efficiency of the previously published Lu3Al5O12:Ce was found to be constantly higher than that of CdWO4. The light emitted from CdWO4 can be optimally detected by certain charge-coupled devices (CCDs), amorphous silicon photodiodes, and photocathodes. Conclusions: The high density (7.9 g/cm3) of CdWO4 and the luminescence signal of this material make it suitable for medical imaging (such as dual energy), high-energy physics or for applications of scintillators in harsh environments.

Research on Medical Image Storage and Retrieval System Based on Hadoop

With the Continuous Improvement of Medical Information, the Continuous Development of Image Processing Technology and the Continuous Upgrading of Hospital Image Acquisition Equipment, Medical Image Detection Has Become More and More Reliable and Effective, Becoming an Important Basis for Doctors’ Diagnosis. Now Large Hospitals Use the Combination of Radiology Information Management System and Image Archiving and Communication System to Store and Access Patients’ Images. with the Establishment of Regional Image Sharing and Cooperation Platform, the Online - Near Line - Offline Storage Strategy Adopted by the Traditional Image Archiving and Communication System Can No Longer Meet the Needs of Mass Image Data Storage. in the Retrieval of Mass Image Data, the Relational Database Used by the Image Archiving and Communication System is Inefficient in Reading the Image on the Scale of Mass Data, and the Existing Image the Archive and Communication System Lacks Complete Data Backup and Disaster Recovery Scheme. This Paper Analyses the Feasibility, Advantages and Problems to Be Solved of Using Hdfs to Store Massive Medical Image Files, and Studies How to Realize the Function of Precise Access and Range Access Provided by Pacs Efficiently in Hadoop.

ElixirNet: Relation-Aware Network Architecture Adaptation for Medical Lesion Detection

Most advances in medical lesion detection network are limited to subtle modification on the conventional detection network designed for natural images. However, there exists a vast domain gap between medical images and natural images where the medical image detection often suffers from several domain-specific challenges, such as high lesion/background similarity, dominant tiny lesions, and severe class imbalance. Is a hand-crafted detection network tailored for natural image undoubtedly good enough over a discrepant medical lesion domain? Is there more powerful operations, filters, and sub-networks that better fit the medical lesion detection problem to be discovered? In this paper, we introduce a novel ElixirNet that includes three components: 1) TruncatedRPN balances positive and negative data for false positive reduction; 2) Auto-lesion Block is automatically customized for medical images to incorporates relation-aware operations among region proposals, and leads to more suitable and efficient classification and localization. 3) Relation transfer module incorporates the semantic relationship and transfers the relevant contextual information with an interpretable graph, thus alleviates the problem of lack of annotations for all types of lesions. Experiments on DeepLesion and Kits19 prove the effectiveness of ElixirNet, achieving improvement of both sensitivity and precision over FPN with fewer parameters.

Siamese neural networks for continuous disease severity evaluation and change detection in medical imaging

Using medical images to evaluate disease severity and change over time is a routine and important task in clinical decision making. Grading systems are often used, but are unreliable as domain experts disagree on disease severity category thresholds. These discrete categories also do not reflect the underlying continuous spectrum of disease severity. To address these issues, we developed a convolutional Siamese neural network approach to evaluate disease severity at single time points and change between longitudinal patient visits on a continuous spectrum. We demonstrate this in two medical imaging domains: retinopathy of prematurity (ROP) in retinal photographs and osteoarthritis in knee radiographs. Our patient cohorts consist of 4861 images from 870 patients in the Imaging and Informatics in Retinopathy of Prematurity (i-ROP) cohort study and 10,012 images from 3021 patients in the Multicenter Osteoarthritis Study (MOST), both of which feature longitudinal imaging data. Multiple expert clinician raters ranked 100 retinal images and 100 knee radiographs from excluded test sets for severity of ROP and osteoarthritis, respectively. The Siamese neural network output for each image in comparison to a pool of normal reference images correlates with disease severity rank (ρ = 0.87 for ROP and ρ = 0.89 for osteoarthritis), both within and between the clinical grading categories. Thus, this output can represent the continuous spectrum of disease severity at any single time point. The difference in these outputs can be used to show change over time. Alternatively, paired images from the same patient at two time points can be directly compared using the Siamese neural network, resulting in an additional continuous measure of change between images. Importantly, our approach does not require manual localization of the pathology of interest and requires only a binary label for training (same versus different). The location of disease and site of change detected by the algorithm can be visualized using an occlusion sensitivity map-based approach. For a longitudinal binary change detection task, our Siamese neural networks achieve test set receiving operator characteristic area under the curves (AUCs) of up to 0.90 in evaluating ROP or knee osteoarthritis change, depending on the change detection strategy. The overall performance on this binary task is similar compared to a conventional convolutional deep-neural network trained for multi-class classification. Our results demonstrate that convolutional Siamese neural networks can be a powerful tool for evaluating the continuous spectrum of disease severity and change in medical imaging.

Temperature Dependence of the Luminescence output of CdWO4 Crystal. Comparison with CaF2:Eu

Scintillators are radiation converters applied in medical imaging detectors, in applications at harsh environments, including in geophysical detectors for deep geology boreholes, non-destructive testing (NDT) in gas and oil facilities, space, marine exploration, etc. In this study the luminescence efficiency dependence of single-crystal scintillators was examined with increasing temperature. Cadmium tungstate (CdWO4) was examined against calcium fluoride doped with europium (CaF2:Eu). The dimensions of the single crystals’ samples were 10x10x10 mm3 and were irradiated using X-ray radiographic exposures (90 kVp, 63mAs) to measure the light output with temperature (22 to 128 °C). The luminescence efficiency was found in both cases maximum at the lowest examined temperature (23.06 efficiency units-E.U for CdWO4 and 22.01 E.U. for CaF2:Eu, at 22 °C-environmental). With increasing temperature, the luminescence efficiency constantly decreased for both crystals due to thermal quenching (5.32 efficiency units for CdWO4 and 4.43 for CaF2:Eu, at 128 °C). In the mid-range (50-80 °C) CdWO4 shows increased differences compared to CaF2:Eu. CdWO4 has a higher density (7.9 g/cm 3) and luminescence signal than CaF2:Eu (3.18 g/cm 3), thus it is suitable, besides medical imaging, also for operation in harsh environments.

Diabetic Retinopathy Detection Using Prognosis of Microaneurysm and Early Diagnosis System for Non-Proliferative Diabetic Retinopathy Based on Deep Learning Algorithms

Predicting the presence of Microaneurysms in the fundus images and the identification of diabetic retinopathy in early-stage has always been a major challenge for decades. Diabetic Retinopathy (DR) is affected by prolonged high blood glucose level which leads to microvascular complications and irreversible vision loss. Microaneurysms formation and macular edema in the retinal is the initial sign of DR and diagnosis at the right time can reduce the risk of non proliferated diabetic retinopathy. The rapid improvement of deep learning makes it gradually become an efficient technique to provide an interesting solution for medical image analysis problems. The proposed system analysis the presence of microaneurysm in fundus image using convolutional neural network algorithms that embeds deep learning as a core component accelerated with GPU(Graphics Processing Unit) which will perform medical image detection and segmentation with high-performance and low-latency inference. The semantic segmentation algorithm is utilized to classify the fundus picture as normal or infected. Semantic segmentation divides the image pixels based on their common semantic to identify the feature of microaneurysm. This provides an automated system that will assist ophthalmologists to grade the fundus images as early NPDR, moderate NPDR, and severe NPDR. The Prognosis of Microaneurysm and early diagnosis system for non - proliferative diabetic retinopathy system has been proposed that is capable to train effectively a deep convolution neural network for semantic segmentation of fundus images which can increase the efficiency and accuracy of NPDR (non proliferated diabetic retinopathy) prediction.

Research on Recognition of Medical Image Detection Based on Neural Network

Bowel cancer, which is easily affected by diet and drugs, has some restrictive factors such as the fecal occult blood test (FOBT) in the routine detection and the high cost and inconvenience of microscopy. In order to break through these restrictive factors, a possible alternative method of FOBT is sought. In this paper, error back propagation neural network (BPNN) algorithm is used, and expression spectrum is used as an auxiliary method to detect medical images, and a colorectal cancer (CRC) diagnosis model based on neural network is constructed. The results show that the accuracy of the model on the training set and the test set are 0.943 and 0.935, respectively, the AUC reaches more than 0.95. Therefore, the CRC diagnosis model based on neural network provides a possible alternative method of FOBT. Experimental results show that the proposed algorithm have high robustness and accuracy, which meets the current clinical needs.

Adversarial explanations for understanding image classification decisions and improved neural network robustness

For sensitive problems, such as medical imaging or fraud detection, Neural Network (NN) adoption has been slow due to concerns about their reliability, leading to a number of algorithms for explaining their decisions. NNs have also been found vulnerable to a class of imperceptible attacks, called adversarial examples, which arbitrarily alter the output of the network. Here we demonstrate both that these attacks can invalidate prior attempts to explain the decisions of NNs, and that with very robust networks, the attacks themselves may be leveraged as explanations with greater fidelity to the model. We show that the introduction of a novel regularization technique inspired by the Lipschitz constraint, alongside other proposed improvements, greatly improves an NN's resistance to adversarial examples. On the ImageNet classification task, we demonstrate a network with an Accuracy-Robustness Area (ARA) of 0.0053, an ARA 2.4x greater than the previous state of the art. Improving the mechanisms by which NN decisions are understood is an important direction for both establishing trust in sensitive domains and learning more about the stimuli to which NNs respond.

Design of broadband impedance-matching Bessel lens with acoustic metamaterials

In this paper, a gradient-index lens is designed to implement an acoustic Bessel beam. Here, the wave from the point source is reshaped to the Bessel beam with the energy concentrated near the axial direction and almost no divergence. The two-dimensional distribution of the gradient refractive index is obtained based on the analysis of the impedance matching theory. The interface reflection is reduced significantly due to the impedance matching with air. The effect of the acoustic Bessel lens is found to work in a broadband with the use of the subwavelength unit cell and the finite-element simulations. Our results may provide the potential applications for medical ultrasound imaging and signal detection.

Development of a technology for the fabrication of Low-Gain Avalanche Diodes at BNL

Low-Gain Avalanche Detectors are gathering interest in the High-Energy Physics community thanks to their fast-timing and radiation-hardness properties, which are planned to be exploited, for example, in timing detectors for the upgrades of the ATLAS and CMS detectors at the High Luminosity LHC. This new technology has also raised interest for its possible application for photon detection in medical physics, imaging and photon science. The main characteristic of this type of device is a thin and highly-doped layer that provides internal and moderate gain, in the order of 10-20, that enhances the signal amplitude. Furthermore, the thin substrate of a few tens of microns allows fast carrier collection. This paper details on the fabrication technology, specifically developed at Brookhaven National Laboratory for the detection of minimum ionizing particles. The static electrical characterization and the gain measurements on prototypes will be reported, too.