Abstract

This work proposes a tractable estimation of the maximum a posteriori (MAP) threshold of various families of sparse-graph code ensembles, by using an approximation for the extended belief propagation generalized extrinsic information transfer (EBP-GEXIT) function, first proposed by M ́easson et al. We consider the transmission over non-binary complex-input additive white Gaussian noise channel and extend the existing results to obtain an expression for the GEXIT function. We estimate the MAP threshold by applying the Maxwell construction to the obtained approximate EBP-GEXIT charts for various families of low-density parity-check (LDPC), generalized LDPC, doubly generalized LDPC, and serially concatenated turbo codes (SC-TC). When codewords of SC-TC are modulated using Gray mapping, we also explore where the spatially-coupled belief propagation (BP) threshold is located with respect to the previously computed MAP threshold. Numerical results indicate that the BP threshold of the spatially-coupled SC-TC does saturate to the MAP threshold obtained via EBP-GEXIT chart.

Abstract

The quantum speed limit indicates the maximal evolution speed of the quantum system. In this work, we determine speed limits on the informational measures, namely the von Neumann entropy, maximal information, and coherence of quantum systems evolving under dynamical processes. These speed limits ascertain the fundamental limitations on the evolution time required by the quantum systems for the changes in their informational measures. Erasing of quantum information to reset the memory for future use is crucial for quantum computing devices. We use the speed limit on the maximal information to obtain the minimum time required to erase the information of quantum systems via some quantum processes of interest.

Abstract

The normal foveal anatomy consists of the extrusion of the plexiform layers, a formed foveal pit, cone photoreceptor outer segment (OS) lengthening, and outer nuclear layer (ONL) widening relative to the parafoveal OS and ONL.1 Foveal development begins with the centrifugal displacement of the inner retinal layers, followed by the centripetal migration of the cone photoreceptors toward the incipient fovea and finally cone photoreceptor specialization (PRS) (Fig 1A–E ).2 3 , 4 Foveal pit formation begins midgestation and continues postnatally characterized by pit deepening and widening. Simultaneously, outer retinal specialization occurs at a rapid rate in the first 2 years of life, but ONL widening continues up to the age of 13 years.4 Lengthening of the OS layer represents cone PRS and reflects peak cone photoreceptor density.5 The maturation of the human fovea is a complex process and is critical for visual function.4 Many genes are involved in the development of the retina. Pathogenic variants in these genes often cause disruption to the foveal developmental process, resulting in foveal hypoplasia (FH), which describes the underdevelopment of the fovea.Foveal hypoplasia is characterized by the continuation of inner retinal layers posterior to the foveola, and progressive loss of these foveal elements is represented by increasing grades of FH. Three key foveal development stages form the basis of the Leicester Grading System for FH, developed by Thomas et al1 (Fig 1F–H). The Leicester Grading System is divided into 4 grades of typical FH (grades 1–4) and 1 grade of atypical FH, which is associated with photoreceptor degeneration. The Leicester Grading System can be applied to a diverse range of genetic disorders, including albinism, achromatopsia, and those caused by pathogenic variants in PAX6, SLC38A8, FRMD7, and AHR. In addition to providing insight into the degree of foveal development, identifying the grade of FH has diagnostic and prognostic implications.1 ,7 Rufai et al7 reported that identifying the grade of FH can predict future visual acuity (VA) in preverbal children with nystagmus. The introduction of OCT has revolutionized ophthalmic diagnosis with the ability to promptly visualize retinal morphology in a high-resolution, noninvasive manner.8 ,9 Furthermore, the advent of handheld OCT has facilitated the foveal examination of pediatric patients.10 OCT is now available in most ophthalmology clinics and subsequently has rapidly become part of routine ophthalmic assessment.8 Thus, the necessary technology to identify foveal developmental abnormalities is now accessible in most centers. The phenotypic spectrum of albinism has been reported, including the variance in degree of arrested retinal development.11 ,12 Furthermore, high grades of FH (grades 3 and 4) have recently been consistently associated with SLC38A8 variants13 ; however, to date, the full spectrum of FH linked with variants of known genes involved in retinal development has not been investigated. Thus, it is unclear whether variants of certain genes known to be related to FH and nystagmus are associated with a more underdeveloped foveal morphology. We performed a comparative study to investigate and characterize the genotypic and phenotypic spectrum of FH in albinism, achromatopsia, PAX6, SLC38A8, FRMD7, and AHR variants.

Abstract

Complementary Split Ring Resonators (CSRR) and Inter-Digitated Capacitors (IDC) based RF sensors have been ubiquitously optimized assuming empty cavities. How- ever, in bio-sensing applications, most of the samples are dissolved in respective solvents. As a result, designing sen-sors based on specific solvent-filled cavities is the optimal approach for future RF bio-sensors. This paper presents the application of Binary Particle Swarm Optimization (BPSO) technique for maximizing the sensitivity of CSRR and IDC sensors. BPSO optimizes the designs by cell patterning the sensing region in the presence of phosphate buffer as a solvent. An enhancement of 47.38× for the CSRR and 2.03× for the IDC in the detection of L-Lysine at frequencies 2.18 GHz and 0.72 GHz respectively has been achieved, when compared to sensitivities of their traditional counterparts. Promising results have also been observed in the detection of sucrose and glucose, making it a suitable contender for optimization of next-generation RF bio-sensors

Abstract

Greenhouse gases (GHGs) play an important role in controlling local air pollution as well as climate change. In this study, we retrieved column-averaged dry-air (X) mole fractions of carbon dioxide (CO2), methane (CH4), and carbon monoxide (CO) using a ground-based EM27/SUN Fourier transform infrared spectrometer (FTIR). The EM27/SUN spectrometers are widely in use in the COllaborative Carbon Column Observing Network (COCCON). The PROFFAST software provided by COCCON has been used to analyze the measured atmospheric solar absorption spectra. In this letter, the diurnal variation and the time series of daily averaged XCO2, XCH4, and XCO covering the period from December 2020 to May 2021 are analyzed. The maximum values of XCO2, XCH4, and XCO are observed to be 420.57 ppm, 1.93 ppm, and 170.40 ppb, respectively. Less diurnal but clear seasonal changes are observed during the study period. XCH4 and XCO from the Sentinel-5Precursor (S5P)/TROPOspheric Monitoring Instrument (TROPOMI) are compared against the EM27/SUN retrievals. The correlation coefficient for the EM27/SUN retrieved XCH4 and XCO, with the S5P/TROPOMI products, are 0.75 and 0.94, respectively

Abstract

Locomotion in Virtual Reality (VR) acts as a motion tracking unit for simulating user movements based on the Degree-of-Freedom (DOF) of the application. For effective locomotion, VR practitioners may have to transform their hardware from 3-DOF to 6-DOF. In this context, we conducted a literature review on different motion tracking methods employed in the Head-Mounted-Devices (HMD) to understand such hardware transformation to conduct locomotion in VR. Our observations led us to formulate a taxonomy of the tracking methods for locomotion in VR based on system design. Our study also captures different metrics that VR practitioners use to evaluate the hardware based on the context, performance, and significance for conducting locomotion.

Abstract

Deep learning based methods have shown great promise in achieving accurate automatic detection of Coronavirus Disease (COVID) - 19 from Chest X-Ray ( CXR) images.However, incorporating explainability in these so- lutions remains relatively less explored. We present a hi- erarchical classification approach for separating normal, non- COVID pneumonia (NCP ) and COVID cases using CXR im- ages. We demonstrate that the proposed method achieves clinically consistent explainations. We achieve this using a novel multi-scale attention architecture called Multi-scale Attention Residual Learning (MARL ) and a new loss function based on conicity for training the proposed architecture. The proposed classification strategy has two stages. The first stage uses a model derived from DenseNet to sepa- rate pneumonia cases from normal cases while the second stage uses the MARL architecture to discriminate between COVID and NCP cases. With a five-fold cross validation the proposed method achieves 93%, 96.28%, and 84.51% accu- racy respectively over three large, public datasets for nor- mal vs. NCP vs. COVID classification. This is competitive to the state-of-the-art methods. We also provide explanations in the form of GradCAM attributions, which are well aligned with expert annotations. The attributions are also seen to clearly indicate that MARL deems the peripheral regions of the lungs to be more important in the case of COVID cases while central regions are seen as more important in NCP cases. This observation matches the criteria described by radiologists in clinical literature, thereby attesting to the utility of the derived explanations.

Abstract

Segmentation and analysis of sub-cortical structures is of interest in diagnosing some neurological diseases. Segmentation is a challenging task because of brain tissue ambiguity and data scarcity. Deep learning (DL) solutions are widely used for this purpose by considering the problem as a semantic segmentation of the brain. In general, DL approaches exhibit a bias towards larger structures when training is done on the whole brain. We propose a method to address this problem wherein a pre-training step is used to learn tissue characteristics and a rough ROI extraction step aids focusing on local context. We use a Residual U-net for demonstrating the proposed method. Experiments on the IBSR and MICCAI datasets show that our proposed solution leads to an improvement in segmentation performance in general with medium and small size structures benefiting significantly. The performance with the proposed method is also marginally better than a more complex, state of art sub-cortical structure segmentation method. A strength of the proposed method is that it can also be applied as a modification to any existing segmentation solution.

Abstract

Deep neural networks suffer from Catastrophic Forgetting (CF) on old tasks when they are trained to learn new tasks sequentially, since the parameters of the model will change to optimize on the new class. The problem of alleviating CF is of interest to Computer aided diagnostic (CAD) sys- tems community to facilitate class incremental learning (IL): learn new classes as and when new data/annotations are made available and old data is no longer accessible. However, IL has not been explored much in CAD development. We pro- pose a novel approach that ensures that a model remembers the causal factor behind the decisions on the old classes, while incrementally learning new classes. We introduce a common auxiliary task during the course of incremental training, whose hidden representations are shared across all the classification heads. Since the hidden representation is no longer task-specific, it leads to a significant reduction in CF. We demonstrate our approach by incrementally learning 5 different tasks on Chest-Xrays and compare the results with the state-of-the-art regularization methods. Our approach performs consistently well in reducing CF in all the tasks with almost zero CF in most of the cases unlike standard regularisation-based approaches.

Abstract

Due to the increasing number of Internet of Things (IoT) devices, a large amount of data is being generated. However, factors such as hardware malfunctions, network failures, or cyber-attacks affect data quality and result in inaccurate data generation. Therefore, to facilitate the data usage, we propose a novel 5D-IoT framework for heterogeneous IoT systems that provides uniform data quality assessment with meaningful data descriptions. Based on the quality assessment result, a data consumer can directly access data from any IoT source, which ultimately speeds up the analysis process and helps gain important insights in less time. The framework relies on semantic descriptions of sensor observations and SHACL shapes assessing the quality of such data. Evaluations carried out on real-time data show the added value of such a framework.