Applications In Medical Diagnosis Research Articles

Automatic Quality Control in Lung X-Ray Imaging with Deep Learning

The development of deep learning and its growing application in medical diagnosis have focused the attention on automatic control of image quality for neural-network medical image analysis algorithms. This article presents a method for automatic determination of the hardness (penetration) of lung X-ray images using standard criteria from chest X-ray diagnosis. The proposed method can be applied to automatically filter images by hardness (penetration) level and to detect low-quality images, thus facilitating the creation of high-quality data sets and increasing the efficiency of neural-network approaches to the analysis of lung X-ray images.

Using single-vesicle technologies to unravel the heterogeneity of extracellular vesicles.

Extracellular vesicles (EVs) are heterogeneous lipid containers with a complex molecular cargo comprising several populations with unique roles in biological processes. These vesicles are closely associated with specific physiological features, which makes them invaluable in the detection and monitoring of various diseases. EVs play a key role in pathophysiological processes by actively triggering genetic or metabolic responses. However, the heterogeneity of their structure and composition hinders their application in medical diagnosis and therapies. This diversity makes it difficult to establish their exact physiological roles, and the functions and composition of different EV (sub)populations. Ensemble averaging approaches currently employed for EV characterization, such as western blotting or 'omics' technologies, tend to obscure rather than reveal these heterogeneities. Recent developments in single-vesicle analysis have made it possible to overcome these limitations and have facilitated the development of practical clinical applications. In this review, we discuss the benefits and challenges inherent to the current methods for the analysis of single vesicles and review the contribution of these approaches to the understanding of EV biology. We describe the contributions of these recent technological advances to the characterization and phenotyping of EVs, examination of the role of EVs in cell-to-cell communication pathways and the identification and validation of EVs as disease biomarkers. Finally, we discuss the potential of innovative single-vesicle imaging and analysis methodologies using microfluidic devices, which promise to deliver rapid and effective basic and practical applications for minimally invasive prognosis systems.

Vacuum ultraviolet silicon photomultipliers applied to BaF2 cross-luminescence detection for high-rate ultrafast timing applications

Inorganic scintillators are widely used for fast timing applications in high-energy physics (HEP) experiments, time-of-flight positron emission tomography and time tagging of soft and hard x-ray photons at advanced light sources. As the best coincidence time resolution (CTR) achievable is proportional to the square root of the scintillation decay time it is worth studying fast cross-luminescence, for example in BaF2 which has an intrinsic yield of about 1400 photons/MeV. However, emission bands in BaF2 are located in the deep-UV at 195 nm and 220 nm, which sets severe constraints on photodetector selection. Recent developments in dark matter and neutrinoless double beta decay searches have led to silicon photomultipliers (SiPMs) with photon detection efficiencies of 20%–25% at wavelengths of 200 nm. We tested state-of-the-art devices from Fondazione Bruno Kessler and measured a best CTR of 51 ± 5 ps full width at half maximum when coupling 2 mm × 2 mm × 3 mm BaF2 crystals excited by 511 keV electron–positron annihilation gammas. Using these vacuum ultraviolet SiPMs we recorded the scintillation kinetics of samples from Epic Crystal under 511 keV excitation, confirming a fast decay time of 855 ps with 12.2% relative light yield and 805 ns with 84.0% abundance, together with a smaller rise time of 4 ps beyond the resolution of our setup. The total intrinsic light yield was determined to be 8500 photons/MeV. We also revealed a faster component with 136 ps decay time and 3.7% light yield contribution, which is extremely interesting for the fastest timing applications. Timing characteristics and CTR results on BaF2 samples from different producers and with different dopants (yttrium, cadmium and lanthanum) are given, and clearly show that the the slow 800 ns emission can be effectively suppressed. Such results ultimately pave the way for high-rate ultrafast timing applications in medical diagnosis, range monitoring in proton or heavy ion therapy and HEP.

A Study of Fuzzy Relation and Its Application in Medical Diagnosis

Objective: Medical diagnosis process extends within the degree to which they plan to affect different complicating aspects of diagnosis. In this research work, the concept of fuzzy relation with medical diagnosis is studied and the application of fuzzy relations to such problems by extending the Sanchez’s approach is introduced.
 Method: An application of fuzzy relation with Sanchez's approach for medical diagnosis is presented. Based on the composition of the fuzzy relations, an algorithm for medical diagnosis as follows- first input the number of objects and attributes to obtain patient symptom matrix, symptom-disease matrix and the composition of fuzzy relations to get the patient-diagnosis matrix. Then find the maximum value to evaluate which patient is suffering from what disease.
 Result: Using the algorithm for medical diagnosis, the disease for which the membership value is maximum gives the final decision. If almost equal values for different diagnosis in composition are obtained, the case for which non-membership is minimum and hesitation is least is considered. The output matched well with the doctor’s diagnosis.
 Conclusion: In the process of medical diagnosis, state of patient are given by the patient through linguistic terminology like as temperature, cough, stomach pain etc., consideration of fuzzy sets as grades for association instead of membership grades in [0,1] is more advantageous to model the state of the patient. Similarly fuzzy relation has been introduced representing the association between symptoms and diseases. Sanchez’s approach has been extended for medical diagnosis in this reference. The approach used to form fuzzy matrix showing the association of symptoms and diseases is based on the sanchez’s approach.

An advanced uncertainty measure using fuzzy soft sets: Application to decision-making problems

In this paper, uncertainty has been measured in the form of fuzziness which arises due to imprecise boundaries of fuzzy sets. Uncertainty caused due to human's cognition can be decreased by the use of fuzzy soft sets. There are different approaches to deal with the measurement of uncertainty. The method we proposed uses fuzzified evidence theory to calculate total degree of fuzziness of the parameters. It consists of mainly four parts. The first part is to measure uncertainties of parameters using fuzzy soft sets and then to modulate the uncertainties calculated. Afterward, the appropriate basic probability assignments with respect to each parameter are produced. In the last, we use Dempster's rule of combination to fuse independent parameters into integrated one. To validate the proposed method, we perform an experiment and compare our outputs with grey relational analysis method. Also, a medical diagnosis application in reference to COVID-19 has been given to show the effectiveness of advanced method by comparing with other method.

AEC Classifier: A Tree-Based Classifier with Error Control for Medical Disease Diagnosis and Other Applications

In medical disease diagnosis, the cost of a false negative could greatly outweigh the cost of a false positive. This is because the former could cost a life, whereas the latter may only cause medical costs and stress to the patient. The unique nature of this problem highlights the need of asymmetric error control for binary classification applications. In this domain, traditional machine learning classifiers may not be ideal as they do not provide a way to control the number of false negatives below a certain threshold. This paper proposes a novel tree-based binary classification algorithm that can control the number of false negatives with a mathematical guarantee, based on Neyman–Pearson (NP) Lemma. This classifier is evaluated on the data obtained from different heart studies and it predicts the risk of cardiac disease, not only with comparable accuracy and AUC-ROC score but also with full control over the number of false negatives. The methodology used to construct this classifier can be expanded to many more use cases, not only in medical disease diagnosis but also beyond as shown from analysis on different diverse datasets.

Classification Under Human Assistance

Most supervised learning models are trained for full automation. However, their predictions are sometimes worse than those by human experts on some specific instances. Motivated by this empirical observation, our goal is to design classifiers that are optimized to operate under different automation levels. More specifically, we focus on convex margin-based classifiers and first show that the problem is NP-hard. Then, we further show that, for support vector machines, the corresponding objective function can be expressed as the difference of two functions f = g - c, where g is monotone, non-negative and gamma-weakly submodular, and c is non-negative and modular. This representation allows a recently introduced deterministic greedy algorithm, as well as a more efficient randomized variant of the algorithm, to enjoy approximation guarantees at solving the problem. Experiments on synthetic and real-world data from several applications in medical diagnosis illustrate our theoretical findings and demonstrate that, under human assistance, supervised learning models trained to operate under different automation levels can outperform those trained for full automation as well as humans operating alone.

A Review of Artificial Intelligence's Neural Networks (Deep Learning) Applications in Medical Diagnosis and Prediction

This paper reviews deep learning applications in medical diagnosis and prediction, such as Convolutional Neural Networks, Fully Convolutional Networks, and Generative Adversarial Networks in medical image analysis. It further summarizes the strength and weaknesses of deep learning in medical imaging and suggests that deep learning's great potential in medical fields.

Aptamer-functionalized Au nanoparticles array as the effective SERS biosensor for label-free detection of interleukin-6 in serum

• Homogeneous Au nanoparticles array was fabricated which showed high SERS activity. • The SERS substrate was modified with IL-6 aptamer which could recognized IL-6 specifically. • The aptamer-SERS assay could detect IL-6 in blood in a wide range with lower limit of detection of 1 pM. • Bacterial infection and radiation injure effect on mice could be evaluated based on IL-6 SERS detection. A label-free, highly sensitive, and selective biosensor for detection of interleukin-6 (IL-6), which is an important inflammatory and cancer progression cytokine, has been achieved by combining gold nanoparticles (NPs) array surface-enhanced Raman spectroscopy (SERS) substrate and DNA aptamer. The SERS substrate consisted of Au NPs array is functionalized with the aptamer which is composed of the IL-6 recognition sequence and the output signal reporter. Upon recognition of target IL-6 in serum, the aptamer changes its conformation, resulting in the corresponding change of the output Raman intensity ratio (I 660 /I 736 ), with which the quantitative evaluation of IL-6 in 10 -12 -10 -7 M range can be realized. This SERS approach was employed to detect IL-6 in mice serum for inspecting the effect of drug and bacterial infection, and particularly, the radiation injury in the treated animals. This work therefore exemplifies that the aptamer-based SERS biosensor can serve as a promising tool for quick and convenient medical diagnosis applications.

A Novel Approach of Complex Dual Hesitant Fuzzy Sets and Their Applications in Pattern Recognition and Medical Diagnosis

Complex dual hesitant fuzzy set (CDHFS) is an assortment of complex fuzzy set (CFS) and dual hesitant fuzzy set (DHFS). In this manuscript, the notion of the CDHFS is explored and its operational laws are discussed. The new methodology of the complex interval-valued dual hesitant fuzzy set (CIvDHFS) and its necessary laws are introduced and are also defensible with the help of examples. Further, the antilogarithmic and with-out exponential-based similarity measures, generalized similarity measures, and their important characteristics are also developed. These similarity measures are applied in the environment of pattern recognition and medical diagnosis to evaluate the proficiency and feasibility of the established measures. We also solved some numerical examples using the established measures to examine the reliability and validity of the proposed measures by comparing these with existing measures. To strengthen the proposed study, the comparative analysis is made and it is conferred that the proposed study is much more superior to the existing studies.

Hydrated Ionic Liquids Boost the Trace Detection Capacity of Proteins on TiO2 Support.

Trace detection based on surface-enhanced Raman scattering (SERS) has attracted considerable attention, and exploiting efficient strategies to stretch the limit of detection and understanding the mechanisms on molecular level are of utmost importance. In this work, we use ionic liquids (ILs) as trace additives in a protein-TiO2 system, allowing us to obtain an exceptionally low limit of detection down to 10–9 M. The enhancement factors (EFs) were determined to 2.30 × 104, 6.17 × 104, and 1.19 × 105, for the three systems: one without ILs, one with ILs in solutions, and one with ILs immobilized on the TiO2 substrate, respectively, corresponding to the molecular forces of 1.65, 1.32, and 1.16 nN quantified by the atomic force microscopy. The dissociation and following hydration of ILs, occurring in the SERS system, weakened the molecular forces but instead improved the electron transfer ability of ILs, which is the major contribution for the observed excellent detection. The weaker diffusion of the hydrated IL ions immobilized on the TiO2 substrate did provide a considerably greater EF value, compared to the ILs in the solution. This work clearly demonstrates the importance of the hydration of ions, causing an improved electron transfer ability of ILs and leading to an exceptional SERS performance in the field of trace detection. Our results should stimulate further development to use ILs in SERS and related applications in bioanalysis, medical diagnosis, and environmental science.

Active training of physics-informed neural networks to aggregate and interpolate parametric solutions to the Navier-Stokes equations

The goal of this work is to train a neural network which approximates solutions to the Navier-Stokes equations across a region of parameter space, in which the parameters define physical properties such as domain shape and boundary conditions. The contributions of this work are threefold:1.To demonstrate that neural networks can be efficient aggregators of whole families of parametric solutions to physical problems, trained using data created with traditional, trusted numerical methods such as finite elements. Advantages include extremely fast evaluation of pressure and velocity at any point in physical and parameter space (asymptotically, ∼3 μs/query), and data compression (the network requires 99% less storage space compared to its own training data).2.To demonstrate that the neural networks can accurately interpolate between finite element solutions in parameter space, allowing them to be instantly queried for pressure and velocity field solutions to problems for which traditional simulations have never been performed.3.To introduce an active learning algorithm, so that during training, a finite element solver can automatically be queried to obtain additional training data in locations where the neural network's predictions are in most need of improvement, thus autonomously acquiring and efficiently distributing training data throughout parameter space. In addition to the obvious utility of Item 2, above, we demonstrate an application of the network in rapid parameter sweeping, very precisely predicting the degree of narrowing in a tube which would result in a 50% increase in end-to-end pressure difference at a given flow rate. This capability could have applications in both medical diagnosis of arterial disease, and in computer-aided design.

Monochromatic image reconstruction via machine learning

X-ray computed tomography (CT) is a nondestructive imaging technique to reconstruct cross-sectional images of an object using x-ray measurements taken from different view angles for medical diagnosis, therapeutic planning, security screening, and other applications. In clinical practice, the x-ray tube emits polychromatic x-rays, and the x-ray detector array operates in the energy-integrating mode to acquire energy intensity. This physical process of x-ray imaging is accurately described by an energy-dependent non-linear integral equation on the basis of the Beer–Lambert law. However, the non-linear model is not invertible using a computationally efficient solution and is often approximated as a linear integral model in the form of the Radon transform, which basically loses energy-dependent information. This approximate model produces an inaccurate quantification of attenuation images, suffering from beam-hardening effects. In this paper, a machine learning-based approach is proposed to correct the model mismatch to achieve quantitative CT imaging. Specifically, a one-dimensional network model is proposed to learn a non-linear transform from a training dataset to map a polychromatic CT image to its monochromatic sinogram at a pre-specified energy level, realizing virtual monochromatic (VM) imaging effectively and efficiently. Our results show that the proposed method recovers high-quality monochromatic projections with an average relative error of less than 2%. The resultant x-ray VM imaging can be applied for beam-hardening correction, material differentiation and tissue characterization, and proton therapy treatment planning.

Magnetoinductive waves in attenuating media

The capability of magnetic induction to transmit signals in attenuating environments has recently gained significant research interest. The wave aspect—magnetoinductive (MI) waves—has been proposed for numerous applications in RF-challenging environments, such as underground/underwater wireless networks, body area networks, and in-vivo medical diagnosis and treatment applications, to name but a few, where conventional electromagnetic waves have a number of limitations, most notably losses. To date, the effects of eddy currents inside the dissipative medium have not been characterised analytically. Here we propose a comprehensive circuit model of coupled resonators in a homogeneous dissipative medium, that takes into account all the electromagnetic effects of eddy currents, and, thereby, derive a general dispersion equation for the MI waves. We also report laboratory experiments to confirm our findings. Our work will serve as a fundamental model for design and analysis of every system employing MI waves or more generally, magnetically-coupled circuits in attenuating media.

A brief history of intelligent medicine

Intelligent medicine is a discipline system developed and established in the background of a rapid development of science and technology. Information technology represented by artificial intelligence, VR(virtual reality) technology and the 1G (first-generation) communication laid a foundation for intelligent medicine. Internet technology, such as 3D printing, robot system and wearable medical devices have helped to improve the development of intelligent medicine. The application of intelligent medicine infiltrated medical imaging, auxiliary diagnosis, surgical operation, hospital management, pharmaceutical development, healthcare system and other aspects of the medical system. Currently, the cutting-edge technologies such as 5G medical, medical cloud platform/big data, advanced material synthesis, 4D printing, AR(Augmented Reality)/MR(Mixed Reality), super intelligence and other technologies have comprehensively driven human society into the era of intelligent medicine.

Compositions on lattice ordered multi-fuzzy soft matrix and its simulated application in medical diagnosis

The essential goal of the manuscript is to initiate lattice ordered multi-fuzzy soft matrix (ℒℳℱSℳ) using lattice ordered multi-fuzzy soft set (ℒℳℱSS) and discuss the operations and depict rudimentary properties on ℒℳℱSℳ. Moreover, we originate the various compositions of ℒℳℱSℳ and propose algorithm for medical diagnosis by applying score matrix which have been implemented with the numerical example. Finally, the comparitive analysis of three composition methods is also included for improved reasonability and reliability.

Single-resonator, stable dual-longitudinal-mode optofluidic microcavity laser based on a hollow-core microstructured optical fiber.

A single-resonator, stable dual-longitudinal-mode optofluidic microcavity laser based on a hollow-core microstructured optical fiber is proposed and experimentally demonstrated. The resonator and microfluidic channel are integrated in the hollow-core region of the fiber, inside which a hexagonal silica ring is used as the only resonator of the laser. Experimental results show that with mixing a small amount of Rhodamine B into a 1 mM Rhodamine 6G solution to form a dual-dye solution as a gain medium, the laser obtained by the method of lateral pumping can operate at dual longitudinal modes, with a threshold of 90 nJ/mm2. By adjusting the concentration of Rhodamine B, the lasing wavelength of the laser and the power ratio of the two wavelengths can be controlled. And because the laser emission is co-excited by different kinds of dye molecules, the mode competition is diminished, enabling the simultaneously efficient optical gain and therefore lasing at dual longitudinal modes stably with a maximum lasing intensity fluctuation of 3.2% within 30 minutes even if the dual longitudinal modes have the same linear polarization states. This work can open up promising opportunities for diverse applications in biosensing and medical diagnosis with high sensitivity and integrated photonics with compact structure.

An intelligent hybrid classification algorithm integrating fuzzy rule-based extraction and harmony search optimization: Medical diagnosis applications

Uncertainty is a critical factor in medical datasets needed to be overcome for increasing diagnosis efficiency. This paper proposes an intelligent classification algorithm comprising a fuzzy rule-based approach, a harmony search (HS) algorithm, and a heuristic algorithm to classify medical datasets intelligently. Two fuzzy approaches, as well as orthogonal and triangular fuzzy sets, are first utilized to define the attributes of data. Then, an HS algorithm is integrated with a heuristic to generate fuzzy rules to select the best rules in the fuzzy rule-based systems. Moreover, to improve the performance of the proposed classification approach, a three-phase parameter tuning approach is applied. First, the Taguchi method (phase I) is employed to tune the parameters of the HS algorithm using a fixed number of training data and find the central points of the parameters’ values. Then, a nested cross-validation (CV) approach consisting of an outer CV (phase II) and an inner CV (phase III) is utilized. Using the Taguchi approach gives the advantage of not considering a wide range of parameters by the nested CV which produces better results on the medical dataset. Nine well-known medical datasets are used to evaluate the efficiency of the proposed hybrid algorithm. To this aim, the results obtained by the algorithm are compared with the ones of several related works in the literature where several statistical tests and graphical approaches are used for comparisons. The results show that the proposed methods are robust in data analysis and classification of clinical datasets.

Frequency division denoising algorithm based on VIF adaptive 2D-VMD ultrasound image.

Ultrasound imaging has developed into an indispensable imaging technology in medical diagnosis and treatment applications due to its unique advantages, such as safety, affordability, and convenience. With the development of data information acquisition technology, ultrasound imaging is increasingly susceptible to speckle noise, which leads to defects, such as low resolution, poor contrast, spots, and shadows, which affect the accuracy of physician analysis and diagnosis. To solve this problem, we proposed a frequency division denoising algorithm combining transform domain and spatial domain. First, the ultrasound image was decomposed into a series of sub-modal images using 2D variational mode decomposition (2D-VMD), and adaptively determined 2D-VMD parameter K value based on visual information fidelity (VIF) criterion. Then, an anisotropic diffusion filter was used to denoise low-frequency sub-modal images, and a 3D block matching algorithm (BM3D) was used to reduce noise for high-frequency images with high noise. Finally, each sub-modal image was reconstructed after processing to obtain the denoised ultrasound image. In the comparative experiments of synthetic, simulation, and real images, the performance of this method was quantitatively evaluated. Various results show that the ability of this algorithm in denoising and maintaining structural details is significantly better than that of other algorithms.

ARTIFICIAL NEURAL NETWORKS IN MEDICAL DIAGNOSIS

Artificial neural networks are finding many uses in the medical diagnosis application. The article examines cases of renopathy in type 2 diabetes. Data are symptoms of disease. The multilayer perceptron networks (MLP) is used as a classifier to distinguish between a sick and a healthy person. The results of applying artificial neural networks for diagnose renopathy based on selected symptoms show the network's ability to recognize to recognize diseases corresponding to human symptoms. Various parameters, structures and learning algorithms of neural networks were tested in the modeling process.