Search

During the diagnostic process, doctors incorporate multimodal information including imaging and the medical history - and similarly medical AI development has increasingly become multimodal. In this paper we tackle a more subtle challenge: doctors take a targeted medical history to obtain only the most pertinent pieces of information; how do we enable AI to do the same? We develop a wrapper method named MINT (Make your model INTeractive) that automatically determines what pieces of information are most valuable at each step, and ask for only the most useful information. We demonstrate the efficacy of MINT wrapping a skin disease prediction model, where multiple images and a set of optional answers to $25$ standard metadata questions (i.e., structured medical history) are used by a multi-modal deep network to provide a differential diagnosis. We show that MINT can identify whether metadata inputs are needed and if so, which question to ask next. We also demonstrate that when collecting multiple images, MINT can identify if an additional image would be beneficial, and if so, which type of image to capture. We showed that MINT reduces the number of metadata and image inputs needed by 82% and 36.2% respectively, while maintaining predictive performance. Using real-world AI dermatology system data, we show that needing fewer inputs can retain users that may otherwise fail to complete the system submission and drop off without a diagnosis. Qualitative examples show MINT can closely mimic the step-by-step decision making process of a clinical workflow and how this is different for straight forward cases versus more difficult, ambiguous cases. Finally we demonstrate how MINT is robust to different underlying multi-model classifiers and can be easily adapted to user requirements without significant model re-training., Comment: 15 pages, 7 figures

Neutron deficient nuclei from $^{74}$Se$-^{196}$Hg are thought to be produced by $\gamma$-induced reactions ($\gamma$,n), ($\gamma$,p) and ($\gamma,\alpha$) processes. The relatively high abundance of $^{113}$In odd A $p$-nuclei has inspired to study its production processes. As reaction with $\gamma$-beam is difficult to perform in the laboratory, $\gamma$-induced reaction rate is calculated from the inverse reaction data employing reciprocity theorem. Stacked foil activation method was used to measure the $^{113}$In($\alpha,\gamma$) and $^{113}$In($\alpha$, n) reactions cross-section near the astrophysical energies. Theoretical statistical model calculations were performed with different nuclear input parameters and compared with the experimental results. An appropriate $\alpha$-optical potential has been identified from the ($\alpha,\gamma$) and ($\alpha$, n) fitting, which provides the major source of uncertainty in the statistical model calculations. The other nuclear input parameters like level density, and $\gamma$-ray strength function were also constrained for theoretical calculations. $^{113}$In($\alpha,\gamma$)$^{117}$Sb and $^{117}$Sb($\alpha,\gamma$)$^{113}$In reaction rates were calculated using best-fitted input parameters.

The recent IceCube detection of significant neutrino flux from the inner Galactic plane has provided us valuable insights on the spectrum of cosmic rays in our Galaxy. This flux can be produced either by a population of Galactic point sources or by diffused emission from cosmic ray interactions with the interstellar medium or by a mixture of both. In this work, we compute diffused gamma-ray and neutrino fluxes produced by a population of giant molecular clouds (GMCs) in our Galaxy, assuming different parametrizations of the Galactic diffused cosmic ray distribution. In particular, we take into account two main cases: (I) constant cosmic ray luminosity in our Galaxy, and (II) space-dependent cosmic ray luminosity, based on the supernovae distribution in our Galaxy. For Case-I, we found that the neutrino flux from GMCs is a factor of $\sim 10$ below compared to $\pi^0$ and KRA$_\gamma$ best-fitted models of IceCube observations at $10^5$ GeV. Instead, for Case-II the model can explain up to $\sim 90 \%$ of the neutrino flux at that energy. Moreover, for this last scenario IceCube detector could be able to detect neutrino events from the Galactic centre regions. We then calculated gamma-ray and neutrino fluxes from individual GMCs and noticed that several current and future Cherenkov telescopes and neutrino observatories have the right sensitivities to study these objects. In particular, very neutrino-bright region such as Aquila Rift is favourable for detection by the IceCube-Gen2 observatory., Comment: 30 pages, 17 Figures

The sky is the limit with regards to the societal impact nanomaterials can have on our lives. However, in this study we show that their potential is out of this world. The planet Mars has an abundant source of calcium sulfate minerals and in our work, we show that these deposits can be the basis of transformative nanomaterials to potentially support future space endeavors. Through a scalable eco-friendly liquid processing technique performed on two common terrestrial gypsum, our simple method presented a cost-efficient procedure to yield the commercially valuable intermediate phase of gypsum, known as bassanite. Through the liquid exfoliation of bassanite powders, suspensions of large aspect ratio anhydrite nanobelts with long-term stability were characterized through scanning electron microscopy and Raman spectroscopy. Transmission electron microscopy showed nanobelts to have a mesocrystal structure, with distinct nanoparticle constituents making up the lattice. Unexpectedly, anhydrite nanobelts had remarkable electronic properties, namely a bandgap that was easily tuned between semiconducting (~2.2 eV) and insulating (~4 eV) behaviors through dimensional control measured via atomic force microscopy. To demonstrate the application potential of our nanobelts; optoelectronic, electrochemical and nanocomposite measurements were made. For the hydrogen evolution reaction and mechanical reinforcement, selenite-based anhydrite nanobelts displayed superlative performances.

Conformal Prediction (CP) allows to perform rigorous uncertainty quantification by constructing a prediction set $C(X)$ satisfying $\mathbb{P}(Y \in C(X))\geq 1-\alpha$ for a user-chosen $\alpha \in [0,1]$ by relying on calibration data $(X_1,Y_1),...,(X_n,Y_n)$ from $\mathbb{P}=\mathbb{P}^{X} \otimes \mathbb{P}^{Y|X}$. It is typically implicitly assumed that $\mathbb{P}^{Y|X}$ is the "true" posterior label distribution. However, in many real-world scenarios, the labels $Y_1,...,Y_n$ are obtained by aggregating expert opinions using a voting procedure, resulting in a one-hot distribution $\mathbb{P}_{vote}^{Y|X}$. For such ``voted'' labels, CP guarantees are thus w.r.t. $\mathbb{P}_{vote}=\mathbb{P}^X \otimes \mathbb{P}_{vote}^{Y|X}$ rather than the true distribution $\mathbb{P}$. In cases with unambiguous ground truth labels, the distinction between $\mathbb{P}_{vote}$ and $\mathbb{P}$ is irrelevant. However, when experts do not agree because of ambiguous labels, approximating $\mathbb{P}^{Y|X}$ with a one-hot distribution $\mathbb{P}_{vote}^{Y|X}$ ignores this uncertainty. In this paper, we propose to leverage expert opinions to approximate $\mathbb{P}^{Y|X}$ using a non-degenerate distribution $\mathbb{P}_{agg}^{Y|X}$. We develop Monte Carlo CP procedures which provide guarantees w.r.t. $\mathbb{P}_{agg}=\mathbb{P}^X \otimes \mathbb{P}_{agg}^{Y|X}$ by sampling multiple synthetic pseudo-labels from $\mathbb{P}_{agg}^{Y|X}$ for each calibration example $X_1,...,X_n$. In a case study of skin condition classification with significant disagreement among expert annotators, we show that applying CP w.r.t. $\mathbb{P}_{vote}$ under-covers expert annotations: calibrated for $72\%$ coverage, it falls short by on average $10\%$; our Monte Carlo CP closes this gap both empirically and theoretically.

For safety, AI systems in health undergo thorough evaluations before deployment, validating their predictions against a ground truth that is assumed certain. However, this is actually not the case and the ground truth may be uncertain. Unfortunately, this is largely ignored in standard evaluation of AI models but can have severe consequences such as overestimating the future performance. To avoid this, we measure the effects of ground truth uncertainty, which we assume decomposes into two main components: annotation uncertainty which stems from the lack of reliable annotations, and inherent uncertainty due to limited observational information. This ground truth uncertainty is ignored when estimating the ground truth by deterministically aggregating annotations, e.g., by majority voting or averaging. In contrast, we propose a framework where aggregation is done using a statistical model. Specifically, we frame aggregation of annotations as posterior inference of so-called plausibilities, representing distributions over classes in a classification setting, subject to a hyper-parameter encoding annotator reliability. Based on this model, we propose a metric for measuring annotation uncertainty and provide uncertainty-adjusted metrics for performance evaluation. We present a case study applying our framework to skin condition classification from images where annotations are provided in the form of differential diagnoses. The deterministic adjudication process called inverse rank normalization (IRN) from previous work ignores ground truth uncertainty in evaluation. Instead, we present two alternative statistical models: a probabilistic version of IRN and a Plackett-Luce-based model. We find that a large portion of the dataset exhibits significant ground truth uncertainty and standard IRN-based evaluation severely over-estimates performance without providing uncertainty estimates.

Recently, the Fermi-LAT gamma-ray satellite has detected six Giant Molecular Clouds (GMCs) located in the Gould Belt and the Aquila Rift regions. In half of these objects (Taurus, Orion A, Orion B), the observed gamma-ray spectrum can be explained using the Galactic diffused Cosmic Ray (CR) interactions with the gas environments. In the remaining three GMCs (Rho Oph, Aquila Rift, Cepheus), the origin of the gamma-ray spectrum is still not well established. We use the GEometry ANd Tracking (GEANT4) simulation framework in order to simulate gamma-ray emission due to CR/GMC interaction in these three objects, taking into account the gas density distribution inside the GMCs. We find that propagation of diffused Galactic CRs inside these GMCs can explain the Fermi-LAT detected gamma-ray spectra. Further, our estimated TeV-PeV fluxes are consistent with the HAWC upper limits, available for the Aquila Rift GMC. As last step, we compute the total neutrino flux estimated for these GMCs and compare it with the IceCube detection sensitivity., Comment: 17 pages, 10 figures

A ubiquitous challenge in machine learning is the problem of domain generalisation. This can exacerbate bias against groups or labels that are underrepresented in the datasets used for model development. Model bias can lead to unintended harms, especially in safety-critical applications like healthcare. Furthermore, the challenge is compounded by the difficulty of obtaining labelled data due to high cost or lack of readily available domain expertise. In our work, we show that learning realistic augmentations automatically from data is possible in a label-efficient manner using generative models. In particular, we leverage the higher abundance of unlabelled data to capture the underlying data distribution of different conditions and subgroups for an imaging modality. By conditioning generative models on appropriate labels, we can steer the distribution of synthetic examples according to specific requirements. We demonstrate that these learned augmentations can surpass heuristic ones by making models more robust and statistically fair in- and out-of-distribution. To evaluate the generality of our approach, we study 3 distinct medical imaging contexts of varying difficulty: (i) histopathology images from a publicly available generalisation benchmark, (ii) chest X-rays from publicly available clinical datasets, and (iii) dermatology images characterised by complex shifts and imaging conditions. Complementing real training samples with synthetic ones improves the robustness of models in all three medical tasks and increases fairness by improving the accuracy of diagnosis within underrepresented groups. This approach leads to stark improvements OOD across modalities: 7.7% prediction accuracy improvement in histopathology, 5.2% in chest radiology with 44.6% lower fairness gap and a striking 63.5% improvement in high-risk sensitivity for dermatology with a 7.5x reduction in fairness gap.

Recent progress in Medical Artificial Intelligence (AI) has delivered systems that can reach clinical expert level performance. However, such systems tend to demonstrate sub-optimal "out-of-distribution" performance when evaluated in clinical settings different from the training environment. A common mitigation strategy is to develop separate systems for each clinical setting using site-specific data [1]. However, this quickly becomes impractical as medical data is time-consuming to acquire and expensive to annotate [2]. Thus, the problem of "data-efficient generalization" presents an ongoing difficulty for Medical AI development. Although progress in representation learning shows promise, their benefits have not been rigorously studied, specifically for out-of-distribution settings. To meet these challenges, we present REMEDIS, a unified representation learning strategy to improve robustness and data-efficiency of medical imaging AI. REMEDIS uses a generic combination of large-scale supervised transfer learning with self-supervised learning and requires little task-specific customization. We study a diverse range of medical imaging tasks and simulate three realistic application scenarios using retrospective data. REMEDIS exhibits significantly improved in-distribution performance with up to 11.5% relative improvement in diagnostic accuracy over a strong supervised baseline. More importantly, our strategy leads to strong data-efficient generalization of medical imaging AI, matching strong supervised baselines using between 1% to 33% of retraining data across tasks. These results suggest that REMEDIS can significantly accelerate the life-cycle of medical imaging AI development thereby presenting an important step forward for medical imaging AI to deliver broad impact.

Efficient air quality sensing serves as one of the essential services provided in any recent smart city. Mostly facilitated by sparsely deployed Air Quality Monitoring Stations (AQMSs) that are difficult to install and maintain, the overall spatial variation heavily impacts air quality monitoring for locations far enough from these pre-deployed public infrastructures. To mitigate this, we in this paper propose a framework named AQuaMoHo that can annotate data obtained from a low-cost thermo-hygrometer (as the sole physical sensing device) with the AQI labels, with the help of additional publicly crawled Spatio-temporal information of that locality. At its core, AQuaMoHo exploits the temporal patterns from a set of readily available spatial features using an LSTM-based model and further enhances the overall quality of the annotation using temporal attention. From a thorough study of two different cities, we observe that AQuaMoHo can significantly help annotate the air quality data on a personal scale., Comment: 26 Pages, 17 Figures, Journal

The Cosmic Ray Albedo Neutron Decay (CRAND) is believed to be the principal mechanism for the formation of inner proton radiation belt -- at least for relatively higher energy particles. We implement this mechanism in a Monte Carlo simulation procedure to calculate the trapped proton radiation at the low Earth orbits, through event-by-event interaction of the cosmic ray particles in the Earth's atmosphere and their transportation in the magnetosphere. We consider the generation of protons from subsequent decay of the secondary neutrons from the cosmic ray interaction in the atmosphere and their transport (and/or trapping) in the geomagnetic field. We address the computational challenges for this type of calculations and develop an optimized algorithm to minimize the computation time. We consider a full 3D description of the Earth's atmospheric and magnetic-field configurations using the latest available models. We present the spatial and phase-space distribution of the trapped protons considering the adiabatic invariants and other parameters at the low Earth orbits. We compare the simulation results with the trapped proton flux measurements made by PAMELA experiment at low Earth orbit and explain certain features observed by the measurement., Comment: 28 pages, 11 figures, Accepted for publication in Advances in Space Research

The study of secondary particles produced by the cosmic-ray interaction in the Earth's atmosphere is very crucial as these particles mainly constitute the background counts produced in the high-energy detectors at balloon and satellite altitudes. In the present work, we calculate the abundance of cosmic-ray generated secondary particles at various heights of the atmosphere by means of a Monte Carlo simulation and use this result to understand the background counts in our X-ray observations using balloon-borne instruments operating near the tropical latitude (geomagnetic latitude: $\sim 14.50^{\circ}$ N). For this purpose, we consider a 3D description of the atmospheric and geomagnetic field configurations surrounding the Earth, as well as the electromagnetic and nuclear interaction processes using Geant4 simulation toolkit. Subsequently, we use a realistic mass model description of the detector under consideration, to simulate the counts produced in the detector due to secondary cosmic-ray particles., Comment: 24 pages, 6 figures

We performed a detailed modelling of the background counts observed in a phoswich scintillator X-ray detector at balloon altitude, used for astronomical observations, on board small scientific balloon. We used Monte Carlo simulation technique in Geant4 simulation environment, to estimate the detector background from various plausible sources. High energy particles and radiation generated from the interaction of Galactic Cosmic Rays with the atmospheric nuclei is a major source of background counts (under normal solar condition) for such detectors. However, cosmogenic or induced radioactivity in the detector materials due to the interaction of high energy particles and natural radioactive contamination present in the detector can also contribute substantially to the detector background. We considered detailed 3D modelling of the earth's atmosphere and magnetosphere to calculate the radiation environment at the balloon altitude and deployed a proper mass model of the detector to calculate the background counts in it. The calculation satisfactorily explains the observed background in the detector at 30 km altitude (atmospheric depth: 11.5 $g/cm^{2}$) during the balloon flight experiment from a location near 14.5$^{\circ}$N geomagnetic latitude., Comment: 28 pages, 8 figures

The diverse near-Earth radiation environment due to cosmic rays and solar radiation has direct impact on human civilization. In the present and upcoming era of increasing air transfer, it is important to have precise idea of radiation dose effects on human body during air travel. Here, we calculate the radiation dose on the human body at the aviation altitude, also considering the shielding effect of the aircraft structure, using Monte Carlo simulation technique based on Geant4 toolkit. We consider proper 3D mathematical model of the atmosphere and geomagnetic field, updated profile of the incoming particle flux due to cosmic rays and appropriate physics processes. We use quasi-realistic computational phantoms to replicate the human body (male/female) for the effective dose calculation and develop a simplified mathematical model of the aircraft (taking Boeing 777--200LR as reference) for the shielding study. We simulate the radiation environment at the flying altitude (considering geomagnetic latitude in the range of 45-50$^{\circ}$), as well as at various locations inside the fuselage of the aircraft. Then, we calculate the dose rates in the different organs for both male and female phantoms, based on latest recommendations of International Commission on Radio logical Protection. This calculation shows that the sex-averaged effective dose rate in human phantom is 5.46 $\mu$Sv/h, whereas, if we calculate weighted sum of equivalent dose contributions separately in female and male body: total weighted sum of equivalent dose rate received by the female phantom is 5.72 $\mu$Sv/h and that by the male phantom is 5.20 $\mu$Sv/h. From the simulation, we also calculate the numerous cosmogenic radionuclides produced inside the phantoms through activation or spallation processes which may induce long-term biological effects., Comment: 37 pages, 15 figures

A generalized theory of spatial coherence for superposition of two speckle patterns with polarization diversity is presented. The presented theory deals with superposition in different scenarios i.e. superposition of two fully correlated, partially correlated and completely uncorrelated speckle patterns, and describes the effect on the spatial coherence-polarization (CP) property of the superposed random field from the study of the spatial degree of coherence and degree of polarization. The change in the spatial CP property with the polarization diversity is studied for different correlation factors of the constituent random fields, and a modulation of the spatial CP property is observed, when the constituent speckle patterns are not fully correlated. The effect of the average intensity ratio of the two random fields on the spatial CP property is also studied. Thus, this theoretical study completes the case of superposition of two speckle patterns with polarization diversity and its effect on the spatial CP property.

Mahalanobis distance (MD) is a simple and popular post-processing method for detecting out-of-distribution (OOD) inputs in neural networks. We analyze its failure modes for near-OOD detection and propose a simple fix called relative Mahalanobis distance (RMD) which improves performance and is more robust to hyperparameter choice. On a wide selection of challenging vision, language, and biology OOD benchmarks (CIFAR-100 vs CIFAR-10, CLINC OOD intent detection, Genomics OOD), we show that RMD meaningfully improves upon MD performance (by up to 15% AUROC on genomics OOD).

We develop and rigorously evaluate a deep learning based system that can accurately classify skin conditions while detecting rare conditions for which there is not enough data available for training a confident classifier. We frame this task as an out-of-distribution (OOD) detection problem. Our novel approach, hierarchical outlier detection (HOD) assigns multiple abstention classes for each training outlier class and jointly performs a coarse classification of inliers vs. outliers, along with fine-grained classification of the individual classes. We demonstrate the effectiveness of the HOD loss in conjunction with modern representation learning approaches (BiT, SimCLR, MICLe) and explore different ensembling strategies for further improving the results. We perform an extensive subgroup analysis over conditions of varying risk levels and different skin types to investigate how the OOD detection performance changes over each subgroup and demonstrate the gains of our framework in comparison to baselines. Finally, we introduce a cost metric to approximate downstream clinical impact. We use this cost metric to compare the proposed method against a baseline system, thereby making a stronger case for the overall system effectiveness in a real-world deployment scenario., Comment: Under Review, 19 Pages

A detailed theoretical and experimental study on the effect of the superposition of uncorrelated speckle patterns with polarization diversity on the spatial statistics of the superposed speckle pattern is presented. It is shown that depending on the mutual orientation of the polarization vectors of the constituent speckle patterns, the maximum degree of coherence (DoC) and degree of polarization (DoP) of the superposed speckle pattern changes between a maximum and minimum value in a sinusoidal fashion. Moreover, the average intensity ratio of the constituent speckle patterns is also found to be affecting these variations. A study of the change in the visibility of the two-point intensity correlation function also reveals a sinusoidal nature of the variation and its dependence on the ratio of the average intensity, which are found to be similar to the variations of the maximum DoC and DoP. A detailed study on the changes in the normalized probability density function is also performed for better understanding of the effect on the spatial statistics.

Non-invasive medical diagnostics demonstrate a propensity for droplet generation and should be studied to devise risk mitigation strategies against the spread of the SARS-CoV-2 virus. We investigate the air-puff tonometry, which uses a short-timed air-puff to applanate the human eye in a bid to detect the early onset of glaucoma by measuring the intraocular pressure. The air-puff consists of a vortex trailed by a high-speed jet. High-speed imaging of the eye during a typical tonometry measurement reveals a sequence of events starting with the interaction between the tear layer and the air puff leading to an initial sheet ejection. It is immediately followed by the trailing jet applanating the central corneal section, causing capillary waves to form and interact with the highly 3D transient expanding sheet. Such interaction with the capillary waves and the surrounding airfield due to the trailing jet causes the expanding sheet to undergo bag breakup, finger formation by Rayleigh Taylor instability and further break up into subsequent droplets by Rayleigh Plateau instability, which eventually splashes onto nearby objects, potentially forming fomites or aerosols which can lead to infections. The complex spatiotemporal phenomenon is carefully documented by rigorous experiments and corroborated using comprehensive theoretical analyses.

The random intensity distribution observed due to the propagation of a coherent beam of light through a scattering medium is known as a speckle pattern. The interaction of the speckles with a birefringent crystal, here, YVO4 results in the e-ray and o-ray speckles. It is observed that the e-ray and o-ray speckles experience an angular shift with the rotation of the crystal and the shift depends on the rotation angle of the crystal. It is also found that the rate of the shift is different for e-ray and o-ray speckles, and is independent of the topological charge of the beam incident on the scattering medium. The observed rate of shift of the speckle pattern is found to be same as that of the e-ray and o-ray beams. It is experimentally demonstrated that the rotation of a birefringent crystal can be tracked very accurately from the speckle shift measurement.

Reliable detection of out-of-distribution (OOD) inputs is increasingly understood to be a precondition for deployment of machine learning systems. This paper proposes and investigates the use of contrastive training to boost OOD detection performance. Unlike leading methods for OOD detection, our approach does not require access to examples labeled explicitly as OOD, which can be difficult to collect in practice. We show in extensive experiments that contrastive training significantly helps OOD detection performance on a number of common benchmarks. By introducing and employing the Confusion Log Probability (CLP) score, which quantifies the difficulty of the OOD detection task by capturing the similarity of inlier and outlier datasets, we show that our method especially improves performance in the `near OOD' classes -- a particularly challenging setting for previous methods.

The rising demand for electricity and its essential nature in today's world calls for intelligent home energy management (HEM) systems that can reduce energy usage. This involves scheduling of loads from peak hours of the day when energy consumption is at its highest to leaner off-peak periods of the day when energy consumption is relatively lower thereby reducing the system's peak load demand, which would consequently result in lesser energy bills, and improved load demand profile. This work introduces a novel way to develop a learning system that can learn from experience to shift loads from one time instance to another and achieve the goal of minimizing the aggregate peak load. This paper proposes a Deep Reinforcement Learning (DRL) model for demand response where the virtual agent learns the task like humans do. The agent gets feedback for every action it takes in the environment; these feedbacks will drive the agent to learn about the environment and take much smarter steps later in its learning stages. Our method outperformed the state of the art mixed integer linear programming (MILP) for load peak reduction. The authors have also designed an agent to learn to minimize both consumers' electricity bills and utilities' system peak load demand simultaneously. The proposed model was analyzed with loads from five different residential consumers; the proposed method increases the monthly savings of each consumer by reducing their electricity bill drastically along with minimizing the peak load on the system when time shiftable loads are handled by the proposed method.

The ability of neural networks to continuously learn and adapt to new tasks while retaining prior knowledge is crucial for many applications. However, current neural networks tend to forget previously learned tasks when trained on new ones, i.e., they suffer from Catastrophic Forgetting (CF). The objective of Continual Learning (CL) is to alleviate this problem, which is particularly relevant for medical applications, where it may not be feasible to store and access previously used sensitive patient data. In this work, we propose a Continual Learning approach for brain segmentation, where a single network is consecutively trained on samples from different domains. We build upon an importance driven approach and adapt it for medical image segmentation. Particularly, we introduce learning rate regularization to prevent the loss of the network's knowledge. Our results demonstrate that directly restricting the adaptation of important network parameters clearly reduces Catastrophic Forgetting for segmentation across domains.

Fully Convolutional Neural Networks (F-CNNs) achieve state-of-the-art performance for segmentation tasks in computer vision and medical imaging. Recently, computational blocks termed squeeze and excitation (SE) have been introduced to recalibrate F-CNN feature maps both channel- and spatial-wise, boosting segmentation performance while only minimally increasing the model complexity. So far, the development of SE blocks has focused on 2D architectures. For volumetric medical images, however, 3D F-CNNs are a natural choice. In this article, we extend existing 2D recalibration methods to 3D and propose a generic compress-process-recalibrate pipeline for easy comparison of such blocks. We further introduce Project & Excite (PE) modules, customized for 3D networks. In contrast to existing modules, Project \& Excite does not perform global average pooling but compresses feature maps along different spatial dimensions of the tensor separately to retain more spatial information that is subsequently used in the excitation step. We evaluate the modules on two challenging tasks, whole-brain segmentation of MRI scans and whole-body segmentation of CT scans. We demonstrate that PE modules can be easily integrated into 3D F-CNNs, boosting performance up to 0.3 in Dice Score and outperforming 3D extensions of other recalibration blocks, while only marginally increasing the model complexity. Our code is publicly available on https://github.com/ai-med/squeeze_and_excitation ., Comment: Accepted for publication at IEEE Transactions on Medical Imaging

A new, different kind of intensity correlation, denoted as polarization based intensity correlation, is proposed and investigated to study the correlation between different polarization components of polarization speckle, which has non-uniform spatial polarization distribution. It is shown both theoretically and experimentally that the range of the polarization based intensity correlation for a particular polarization component of the polarization speckle depends on the spatial average intensity of the speckles corresponding to that particular polarization component. The experimentally determined nature of the change of range of the intensity correlation for different polarization components, due to variation in the average intensity, is found to be matching well with the theoretical prediction. The existence of non-zero correlation between two orthogonally polarized speckle patterns, filtered from a partially depolarized speckle pattern, is also observed. This study may be useful in exploiting the polarization based intensity correlation for different applications such as speckle cryptography etc.

We use existing light weight balloon facility of Indian Centre for Space Physics to detect the X-ray radiation from Crab pulsar with a phoswich detector. We present the design considerations and characterization of the detector used for this purpose. We model the background radiation in the detector environment at various altitudes and use this in spectral analysis. The background radiation level and limitations on the detector allowed us to calculate minimum detection limit for extrasolar radiation sources with our set up., Comment: 15 pages, 9 figures

Fully Convolutional Neural Networks (F-CNNs) achieve state-of-the-art performance for image segmentation in medical imaging. Recently, squeeze and excitation (SE) modules and variations thereof have been introduced to recalibrate feature maps channel- and spatial-wise, which can boost performance while only minimally increasing model complexity. So far, the development of SE has focused on 2D images. In this paper, we propose `Project & Excite' (PE) modules that base upon the ideas of SE and extend them to operating on 3D volumetric images. `Project & Excite' does not perform global average pooling, but squeezes feature maps along different slices of a tensor separately to retain more spatial information that is subsequently used in the excitation step. We demonstrate that PE modules can be easily integrated in 3D U-Net, boosting performance by 5% Dice points, while only increasing the model complexity by 2%. We evaluate the PE module on two challenging tasks, whole-brain segmentation of MRI scans and whole-body segmentation of CT scans. Code: https://github.com/ai-med/squeeze_and_excitation, Comment: Accepted for International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) 2019

Access to sufficient annotated data is a common challenge in training deep neural networks on medical images. As annotating data is expensive and time-consuming, it is difficult for an individual medical center to reach large enough sample sizes to build their own, personalized models. As an alternative, data from all centers could be pooled to train a centralized model that everyone can use. However, such a strategy is often infeasible due to the privacy-sensitive nature of medical data. Recently, federated learning (FL) has been introduced to collaboratively learn a shared prediction model across centers without the need for sharing data. In FL, clients are locally training models on site-specific datasets for a few epochs and then sharing their model weights with a central server, which orchestrates the overall training process. Importantly, the sharing of models does not compromise patient privacy. A disadvantage of FL is the dependence on a central server, which requires all clients to agree on one trusted central body, and whose failure would disrupt the training process of all clients. In this paper, we introduce BrainTorrent, a new FL framework without a central server, particularly targeted towards medical applications. BrainTorrent presents a highly dynamic peer-to-peer environment, where all centers directly interact with each other without depending on a central body. We demonstrate the overall effectiveness of FL for the challenging task of whole brain segmentation and observe that the proposed server-less BrainTorrent approach does not only outperform the traditional server-based one but reaches a similar performance to a model trained on pooled data.

Model architectures have been dramatically increasing in size, improving performance at the cost of resource requirements. In this paper we propose 3DQ, a ternary quantization method, applied for the first time to 3D Fully Convolutional Neural Networks (F-CNNs), enabling 16x model compression while maintaining performance on par with full precision models. We extensively evaluate 3DQ on two datasets for the challenging task of whole brain segmentation. Additionally, we showcase our method's ability to generalize on two common 3D architectures, namely 3D U-Net and V-Net. Outperforming a variety of baselines, the proposed method is capable of compressing large 3D models to a few MBytes, alleviating the storage needs in space critical applications., Comment: Accepted to MICCAI 2019