RESUMEN
BACKGROUND: Research suggests that treatment of multiple brain metastases (BMs) with stereotactic radiosurgery shows improvement when metastases are detected early, providing a case for BM detection capabilities on small lesions. PURPOSE: To demonstrate automatic detection of BM on three MRI datasets using a deep learning-based approach. To improve the performance of the network is iteratively co-trained with datasets from different domains. A systematic approach is proposed to prevent catastrophic forgetting during co-training. STUDY TYPE: Retrospective. POPULATION: A total of 156 patients (105 ground truth and 51 pseudo labels) with 1502 BM (BrainMetShare); 121 patients with 722 BM (local); 400 patients with 447 primary gliomas (BrATS). Training/pseudo labels/validation data were distributed 84/51/21 (BrainMetShare). Training/validation data were split: 121/23 (local) and 375/25 (BrATS). FIELD STRENGTH/SEQUENCE: A 5 T and 3 T/T1 spin-echo postcontrast (T1-gradient echo) (BrainMetShare), 3 T/T1 magnetization prepared rapid acquisition gradient echo postcontrast (T1-MPRAGE) (local), 0.5 T, 1 T, and 1.16 T/T1-weighted-fluid-attenuated inversion recovery (T1-FLAIR) (BrATS). ASSESSMENT: The ground truth was manually segmented by two (BrainMetShare) and four (BrATS) radiologists and manually annotated by one (local) radiologist. Confidence and volume based domain adaptation (CAVEAT) method of co-training the three datasets on a 3D nonlocal convolutional neural network (CNN) architecture was implemented to detect BM. STATISTICAL TESTS: The performance was evaluated using sensitivity and false positive rates per patient (FP/patient) and free receiver operating characteristic (FROC) analysis at seven predefined (1/8, 1/4, 1/2, 1, 2, 4, and 8) FPs per scan. RESULTS: The sensitivity and FP/patient from a held-out set registered 0.811 at 2.952 FP/patient (BrainMetShare), 0.74 at 3.130 (local), and 0.723 at 2.240 (BrATS) using the CAVEAT approach with lesions as small as 1 mm being detected. DATA CONCLUSION: Improved sensitivities at lower FP can be achieved by co-training datasets via the CAVEAT paradigm to address the problem of data sparsity. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 2.
Asunto(s)
Neoplasias Encefálicas , Aprendizaje Profundo , Humanos , Estudios Retrospectivos , Imagen por Resonancia Magnética/métodos , Neoplasias Encefálicas/patología , Redes Neurales de la ComputaciónRESUMEN
Coronavirus disease (COVID-19) is a global health crisis caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Real-time reverse transcriptase-polymerase chain reaction (RT-PCR) is the gold standard test for diagnosing COVID-19. Although it is highly accurate, this lab test requires highly-trained personnel and the turn-around time is long. Rapid and inexpensive immuno-diagnostic tests (antigen or antibody test) are available, but these point of care (POC) tests are not as accurate as the RT-PCR test. Biosensors are promising alternatives to these rapid POC tests. Here we review three types of recently developed biosensors for SARS-CoV-2 detection: surface plasmon resonance (SPR)-based, electrochemical and field-effect transistor (FET)-based biosensors. We explain the sensing principles and discuss the advantages and limitations of these sensors. The accuracies of these sensors need to be improved before they could be translated into POC devices for commercial use. We suggest potential biorecognition elements with highly selective target-analyte binding that could be explored to increase the true negative detection rate. To increase the true positive detection rate, we suggest two-dimensional materials and nanomaterials that could be used to modify the sensor surface to increase the sensitivity of the sensor.
Asunto(s)
Técnicas Biosensibles , COVID-19 , Nanoestructuras , SARS-CoV-2/aislamiento & purificación , COVID-19/diagnóstico , Técnicas Electroquímicas , Humanos , Resonancia por Plasmón de SuperficieRESUMEN
Convolutional neural networks (CNNs) have been used quite successfully for semantic segmentation of brain tumors. However, current CNNs and attention mechanisms are stochastic in nature and neglect the morphological indicators used by radiologists to manually annotate regions of interest. In this paper, we introduce a channel and spatial wise asymmetric attention (CASPIAN) by leveraging the inherent structure of tumors to detect regions of saliency. To demonstrate the efficacy of our proposed layer, we integrate this into a well-established convolutional neural network (CNN) architecture to achieve higher Dice scores, with less GPU resources. Also, we investigate the inclusion of auxiliary multiscale and multiplanar attention branches to increase the spatial context crucial in semantic segmentation tasks. The resulting architecture is the new CASPIANET++, which achieves Dice Scores of 91.19%, 87.6% and 81.03% for whole tumor, tumor core and enhancing tumor respectively. Furthermore, driven by the scarcity of brain tumor data, we investigate the Noisy Student method for segmentation tasks. Our new Noisy Student Curriculum Learning paradigm, which infuses noise incrementally to increase the complexity of the training images exposed to the network, further boosts the enhancing tumor region to 81.53%. Additional validation performed on the BraTS2020 data shows that the Noisy Student Curriculum Learning method works well without any additional training or finetuning.
Asunto(s)
Neoplasias Encefálicas , Procesamiento de Imagen Asistido por Computador , Neoplasias Encefálicas/diagnóstico por imagen , Curriculum , Humanos , Redes Neurales de la Computación , EstudiantesRESUMEN
Although the application of sub-sensory mechanical noise to the soles of the feet has been shown to enhance balance, there has been no study on how the bandwidth of the noise affects balance. Here, we report a single-blind randomized controlled study on the effects of a narrow and wide bandwidth mechanical noise on healthy young subjects' sway during quiet standing on firm and compliant surfaces. For the firm surface, there was no improvement in balance for both bandwidths-this may be because the young subjects could already balance near-optimally or optimally on the surface by themselves. For the compliant surface, balance improved with the introduction of wide but not narrow bandwidth noise, and balance is improved for wide compared to narrow bandwidth noise. This could be explained using a simple model, which suggests that adding noise to a sub-threshold pressure stimulus results in markedly different frequency of nerve impulse transmitted to the brain for the narrow and wide bandwidth noise-the frequency is negligible for the former but significantly higher for the latter. Our results suggest that if a person's standing balance is not optimal (for example, due to aging), it could be improved by applying a wide bandwidth noise to the feet.
RESUMEN
It has been recently demonstrated that one-port surface acoustic wave (SAW) resonators known for their high Q value and relatively small device footprint could be utilized for in-liquid mass loading sensing applications where only the reflectors of the device are coated with the sensing film, while the interdigital transducer (IDT) is isolated from the sensing environment. The sensor relies on changes induced in reflectivity and phase velocity of SAW in the region of the reflectors upon detection of the measurand and is particularly advantageous for SAW resonator-type sensors as any contact of the sensing film with the IDT could change its static capacitance during sensing and thereby introduce serious instability in the sensor response. Accordingly, in the present work, the existing scattering matrix approach to the design of one-port SAW resonator filters, which does not cater to the integration of sensing film on the resonator surface, is adapted to develop a method to design one-port SAW resonator sensors utilizing reflectors as sensing element. The reflector block of the one-port SAW resonator is readily split into sensing-active and sensing-inactive parts using the SAW grating transmission matrix in order to study the changes introduced in input admittance of the device for varying level of coverage of the sensing film. The theoretical design approach presented in this work could be used to fabricate high-performance one-port SAW resonator sensors operating at its point of highest sensitivity while utilizing one of the device reflectors as sensing element, without the use of additional impedance matching circuit elements.
RESUMEN
The biophotovoltaic cell (BPV) is deemed to be a potent green energy device as it demonstrates the generation of renewable energy from microalgae; however, inadequate electron generation from microalgae is a significant impediment for functional employment of these cells. The photosynthetic process is not only affected by the temperature, CO2 concentration and light intensity but also the spectrum of light. Thus, a detailed understanding of the influences of light spectrum is essential. Accordingly, we developed spectrally optimized light using programmable LED arrays (PLA)s to study the effect on algae growth and bioelectricity generation. Chlorella is a green microalga and contains chlorophyll-a (chl-a), which is the major light harvesting pigment that absorbs light in the blue and red spectrum. In this study, Chlorella is grown under a PLA which can optimally simulate the absorption spectrum of the pigments in Chlorella. This experiment investigated the growth, photosynthetic performance and bioelectricity generation of Chlorella when exposed to an optimally-tuned light spectrum. The algal BPV performed better under PLA with a peak power output of 0.581 mW m-2 for immobilized BPV device on day 8, which is an increase of 188% compared to operation under a conventional white LED light source. The photosynthetic performance, as measured using pulse amplitude modulation (PAM) fluorometry, showed that the optimized spectrum from the PLA gave an increase of 72% in the rETRmax value (190.5 µmol electrons m-2 s-1), compared with the conventional white light source. Highest algal biomass (1100 mg L-1) was achieved in the immobilized system on day eight, which translates to a carbon fixation of 550 mg carbon L-1. When artificial light is used for the BPV system, it should be optimized with the light spectrum and intensity best suited to the absorption capability of the pigments in the cells. Optimum artificial light source with algal BPV device can be integrated into a power management system for low power application (eg. environment sensor for indoor agriculture system).
RESUMEN
This study investigates the effects of aircraft cabin pressure on intracranial pressure (ICP) elevation of a pneumocephalus patient. We propose an experimental setup that simulates the intracranial hydrodynamics of a pneumocephalus patient during flight. It consists of an acrylic box (skull), air-filled balloon [intracranial air (ICA)], water-filled balloon (cerebrospinal fluid and blood) and agarose gel (brain). The cabin was replicated using a custom-made pressure chamber. The setup can measure the rise in ICP during depressurization to levels similar to that inside the cabin at cruising altitude. ΔICP, i.e. the difference between mean cruising ICP and initial ICP, was found to increase with ICA volume and ROC. However, ΔICP was independent of the initial ICP. The largest ΔICP was 5 mmHg; obtained when ICA volume and ROC were 20 ml and 1,600 ft/min, respectively. The postulated ICA expansion and the subsequent increase in ICP in pneumocephalus patients during flight were successfully quantified in a laboratory setting. Based on the quantitative and qualitative analyses of the results, an ICA volume of 20 ml and initial ICP of 15 mmHg were recommended as conservative thresholds that are required for safe air travel among pneumocephalus patients. This study provides laboratory data that may be used by doctors to advise post-neurosurgical patients if they can safely fly.
Asunto(s)
Medicina Aeroespacial , Aeronaves , Ambiente Controlado , Hipertensión Intracraneal/fisiopatología , Presión Intracraneal , Modelos Teóricos , Neumocéfalo/fisiopatología , Presión del Aire , Viaje en Avión , Altitud , Encéfalo/fisiopatología , Humanos , Masculino , Persona de Mediana Edad , Cráneo/fisiopatologíaRESUMEN
Complex dynamical systems can shift abruptly from a stable state to an alternative stable state at a tipping point. Before the critical transition, the system either slows down in its recovery rate or flickers between the basins of attraction of the alternative stable states. Whether the heart critically slows down or flickers before it transitions into and out of paroxysmal atrial fibrillation (PAF) is still an open question. To address this issue, we propose a novel definition of cardiac states based on beat-to-beat (RR) interval fluctuations derived from electrocardiogram data. Our results show the cardiac state flickers before PAF onset and termination. Prior to onset, flickering is due to a "tug-of-war" between the sinus node (the natural pacemaker) and atrial ectopic focus/foci (abnormal pacemakers), or the pacing by the latter interspersed among the pacing by the former. It may also be due to an abnormal autonomic modulation of the sinus node. This abnormal modulation may be the sole cause of flickering prior to termination since atrial ectopic beats are absent. Flickering of the cardiac state could potentially be used as part of an early warning or screening system for PAF and guide the development of new methods to prevent or terminate PAF. The method we have developed to define system states and use them to detect flickering can be adapted to study critical transition in other complex systems.
Asunto(s)
Fibrilación Atrial/fisiopatología , Bases de Datos Factuales , Electrocardiografía , Frecuencia Cardíaca/fisiología , Humanos , Nodo Sinoatrial/fisiopatologíaRESUMEN
Giancardo et al. recently introduced the neuroQWERTY index (nQi), which is a novel motor index derived from computer-key-hold-time data using an ensemble regression algorithm, to detect early-stage Parkinson's disease. Here, we derive a much simpler motor index from their hold-time data, which is the standard deviation (SD) of the hold-time fluctuations, where fluctuation is defined as the difference between successive natural-log of hold time. Our results show the performance of the SD and nQi tests in discriminating early-stage subjects from controls do not differ, although the SD index is much simpler. There is also no difference in performance between the SD and alternating-finger-tapping tests.
Asunto(s)
Destreza Motora/fisiología , Enfermedad de Parkinson/diagnóstico , Enfermedad de Parkinson/fisiopatología , Algoritmos , Estudios de Casos y Controles , Computadores , Dedos/fisiopatología , Humanos , Movimiento/fisiología , Análisis y Desempeño de Tareas , Factores de TiempoRESUMEN
PURPOSE: Lung nodules have very diverse shapes and sizes, which makes classifying them as benign/malignant a challenging problem. In this paper, we propose a novel method to predict the malignancy of nodules that have the capability to analyze the shape and size of a nodule using a global feature extractor, as well as the density and structure of the nodule using a local feature extractor. METHODS: We propose to use Residual Blocks with a 3 × 3 kernel size for local feature extraction and Non-Local Blocks to extract the global features. The Non-Local Block has the ability to extract global features without using a huge number of parameters. The key idea behind the Non-Local Block is to apply matrix multiplications between features on the same feature maps. RESULTS: We trained and validated the proposed method on the LIDC-IDRI dataset which contains 1018 computed tomography scans. We followed a rigorous procedure for experimental setup, namely tenfold cross-validation, and ignored the nodules that had been annotated by < 3 radiologists. The proposed method achieved state-of-the-art results with AUC = 95.62%, while significantly outperforming other baseline methods. CONCLUSIONS: Our proposed deep Local-Global network has the capability to accurately extract both local and global features. Our new method outperforms state-of-the-art architecture including Densenet and Resnet with transfer learning.
Asunto(s)
Procesamiento de Imagen Asistido por Computador/métodos , Neoplasias Pulmonares/diagnóstico por imagen , Redes Neurales de la Computación , Nódulo Pulmonar Solitario/diagnóstico por imagen , Tomografía Computarizada por Rayos X/métodos , HumanosRESUMEN
Although heavy-tailed fluctuations are ubiquitous in complex systems, a good understanding of the mechanisms that generate them is still lacking. Optical complex systems are ideal candidates for investigating heavy-tailed fluctuations, as they allow recording large datasets under controllable experimental conditions. A dynamical regime that has attracted a lot of attention over the years is the so-called low-frequency fluctuations (LFFs) of semiconductor lasers with optical feedback. In this regime, the laser output intensity is characterized by abrupt and apparently random dropouts. The statistical analysis of the inter-dropout-intervals (IDIs) has provided many useful insights into the underlying dynamics. However, the presence of large temporal fluctuations in the IDI sequence has not yet been investigated. Here, by applying fluctuation analysis we show that the experimental distribution of IDI fluctuations is heavy-tailed, and specifically, is well-modeled by a non-Gaussian stable distribution. We find a good qualitative agreement with simulations of the Lang-Kobayashi model. Moreover, we uncover a transition from a less-heavy-tailed state at low pump current to a more-heavy-tailed state at higher pump current. Our results indicate that fluctuation analysis can be a useful tool for investigating the output signals of complex optical systems; it can be used for detecting underlying regime shifts, for model validation and parameter estimation.
Asunto(s)
Láseres de Semiconductores , Dinámicas no LinealesRESUMEN
A frequent observation in plant-animal mutualistic networks is that abundant species tend to be more generalised, interacting with a broader range of interaction partners than rare species. Uncovering the causal relationship between abundance and generalisation has been hindered by a chicken-and-egg dilemma: is generalisation a by-product of being abundant, or does high abundance result from generalisation? Here, we analyse a database of plant-pollinator and plant-seed disperser networks, and provide strong evidence that the causal link between abundance and generalisation is uni-directional. Specifically, species appear to be generalists because they are more abundant, but the converse, that is that species become more abundant because they are generalists, is not supported by our analysis. Furthermore, null model analyses suggest that abundant species interact with many other species simply because they are more likely to encounter potential interaction partners.
Asunto(s)
Cadena Alimentaria , Modelos Biológicos , Plantas , Polinización , Dispersión de Semillas , Simbiosis , Animales , Densidad de PoblaciónRESUMEN
The statistical predictions of Newtonian and special-relativistic mechanics, which are calculated from an initially Gaussian ensemble of trajectories, are compared for a low-speed scattering system. The comparisons are focused on the mean dwell time, transmission and reflection coefficients, and the position and momentum means and standard deviations. We find that the statistical predictions of the two theories do not always agree as conventionally expected. The predictions are close if the scattering is non-chaotic but they are radically different if the scattering is chaotic and the initial ensemble is well localized in phase space. Our result indicates that for low-speed chaotic scattering, special-relativistic mechanics must be used, instead of the standard practice of using Newtonian mechanics, to obtain empirically-correct statistical predictions from an initially well-localized Gaussian ensemble.
Asunto(s)
Modelos Estadísticos , Movimiento (Física) , Dinámicas no Lineales , AlgoritmosRESUMEN
The newtonian and special-relativistic statistical predictions for the mean, standard deviation and probability density function of the position and momentum are compared for the periodically-delta-kicked particle at low speed. Contrary to expectation, we find that the statistical predictions, which are calculated from the same parameters and initial gaussian ensemble of trajectories, do not always agree if the initial ensemble is sufficiently well-localized in phase space. Moreover, the breakdown of agreement is very fast if the trajectories in the ensemble are chaotic, but very slow if the trajectories in the ensemble are non-chaotic. The breakdown of agreement implies that special-relativistic mechanics must be used, instead of the standard practice of using newtonian mechanics, to correctly calculate the statistical predictions for the dynamics of a low-speed system.
Asunto(s)
Modelos Teóricos , Fenómenos Físicos , Modelos Estadísticos , Movimiento (Física)RESUMEN
We show, contrary to expectation, that the trajectory predicted by general-relativistic mechanics for a low-speed weak-gravity system is not always well-approximated by the trajectories predicted by special-relativistic and newtonian mechanics for the same parameters and initial conditions. If the system is dissipative, the breakdown of agreement occurs for chaotic trajectories only. If the system is non-dissipative, the breakdown of agreement occurs for chaotic trajectories and non-chaotic trajectories. The agreement breaks down slowly for non-chaotic trajectories but rapidly for chaotic trajectories. When the predictions are different, general-relativistic mechanics must therefore be used, instead of special-relativistic mechanics (newtonian mechanics), to correctly study the dynamics of a weak-gravity system (a low-speed weak-gravity system).
Asunto(s)
Hipogravedad , Dinámicas no Lineales , Algoritmos , Simulación por ComputadorRESUMEN
Newtonian and special-relativistic predictions, based on the same model parameters and initial conditions for the trajectory of a low-speed scattering system are compared. When the scattering is chaotic, the two predictions for the trajectory can rapidly diverge completely, not only quantitatively but also qualitatively, due to an exponentially growing separation taking place in the scattering region. In contrast, when the scattering is nonchaotic, the breakdown of agreement between predictions takes a very long time, since the difference between the predictions grows only linearly. More importantly, in the case of low-speed chaotic scattering, the rapid loss of agreement between the Newtonian and special-relativistic trajectory predictions implies that special-relativistic mechanics must be used, instead of the standard practice of using Newtonian mechanics, to correctly describe the scattering dynamics.
Asunto(s)
Dinámicas no Lineales , Procesos Estocásticos , Factores de TiempoRESUMEN
The dynamics of a periodically delta-kicked Hamiltonian system moving at low speed (i.e., at speed much less than the speed of light) is studied numerically. In particular, the trajectory of the system predicted by Newtonian mechanics is compared with the trajectory predicted by special relativistic mechanics for the same parameters and initial conditions. We find that the Newtonian trajectory, although close to the relativistic trajectory for some time, eventually disagrees completely with the relativistic trajectory, regardless of the nature (chaotic, nonchaotic) of each trajectory. However, the agreement breaks down very fast if either the Newtonian or relativistic trajectory is chaotic, but very much slower if both the Newtonian and relativistic trajectories are nonchaotic. In the former chaotic case, the difference between the Newtonian and relativistic values for both position and momentum grows, on average, exponentially. In the latter nonchaotic case, the difference grows much slower, for example, linearly on average.
Asunto(s)
Física/métodos , Modelos Estadísticos , Modelos Teóricos , Movimiento (Física) , Dinámicas no Lineales , Tamaño de la Partícula , Proyectos de Investigación , Factores de TiempoRESUMEN
We use model five-level systems to study resonance leaking of pi-pulse-induced multiphoton (MP) transitions along a strongly coupled anharmonic ladder. We demonstrate that the presence of a weakly bound background state attached to the ladder either in linear or Lambda configuration can have very pronounced effects on resonant MP ladder transitions, including essentially complete quenching of the primary transition. We also develop control strategies for the elimination of background state population based on phase-adjusted Gaussian pulse pairs and discuss the underlying control mechanisms. Finally we show that these strategies are effective in realistic molecular many-level systems. In particular, we demonstrate efficient pulse-pair control of resonance leaking in a 165-level system modeling vibrational excitation in HCN.