Efficient Deep Network Architectures for Fast Chest X-Ray Tuberculosis Screening and Visualization.

Automated diagnosis of tuberculosis (TB) from chest X-Rays (CXR) has been tackled with either hand-crafted algorithms or machine learning approaches such as support vector machines (SVMs) and convolutional neural networks (CNNs). Most deep neural network applied to the task of tuberculosis diagnosis have been adapted from natural image classification. These models have a large number of parameters as well as high hardware requirements, which makes them prone to overfitting and harder to deploy in mobile settings. We propose a simple convolutional neural network optimized for the problem which is faster and more efficient than previous models but preserves their accuracy. Moreover, the visualization capabilities of CNNs have not been fully investigated. We test saliency maps and grad-CAMs as tuberculosis visualization methods, and discuss them from a radiological perspective.

The use of laser-based diagnostics for the rapid identification of infectious agents in human blood.

AIMS: To investigate the use of a laser-based method of detection as a potential diagnostic test for the rapid identification of infectious agents in human blood. METHODS AND RESULTS: In this study, the successful differentiation of blood spiked with viruses, bacteria or protozoan parasites to clinically relevant levels is demonstrated using six blood types (O+, O-, AB+, A+, A-, B+) using blood from different individuals with blood samples prepared in two different laboratories. Experiments were performed using various compositions of filters, experimental set-ups and experimental parameters for spectral capture. CONCLUSIONS: The potential for developing a laser-based diagnostic instrument to detect the presence of parasites, bacteria and viruses in human blood capable of providing analysis results within minutes was demonstrated. SIGNIFICANCE AND IMPACT OF THE STUDY: There is an ongoing need for clinical diagnostics to adapt to newly emerging agents and to screen simultaneously for multiple infectious agents. A laser-based approach can achieve sensitive, multiplex detection with minimal sample preparation and provide rapid results (within minutes). These properties along with the flexibility to add new agent detection by simply adjusting the detection programming make it a promising tool for clinical diagnosis.

Video Object Segmentation without Temporal Information.

Video Object Segmentation, and video processing in general, has been historically dominated by methods that rely on the temporal consistency and redundancy in consecutive video frames. When the temporal smoothness is suddenly broken, such as when an object is occluded, or some frames are missing in a sequence, the result of these methods can deteriorate significantly. This paper explores the orthogonal approach of processing each frame independently, i.e., disregarding the temporal information. In particular, it tackles the task of semi-supervised video object segmentation: the separation of an object from the background in a video, given its mask in the first frame. We present Semantic One-Shot Video Object Segmentation (OSVOS$^\mathrm {S}$S), based on a fully-convolutional neural network architecture that is able to successively transfer generic semantic information, learned on ImageNet, to the task of foreground segmentation, and finally to learning the appearance of a single annotated object of the test sequence (hence one shot). We show that instance-level semantic information, when combined effectively, can dramatically improve the results of our previous method, OSVOS. We perform experiments on two recent single-object video segmentation databases, which show that OSVOS$^\mathrm {S}$S is both the fastest and most accurate method in the state of the art. Experiments on multi-object video segmentation show that OSVOS$^\mathrm {S}$S obtains competitive results.

Stress sensitivity as a putative mechanism linking childhood trauma and psychopathology in youth's daily life.

OBJECTIVE: Childhood trauma (CT) is associated with a range of psychopathologies, including psychosis. However, evidence on underlying mechanisms remains limited. The study aimed to investigate whether CT impacts on youth mental health by modifying sensitivity to stress in daily life. METHOD: The experience sampling method (ESM) was used to measure momentary stress, negative affect and psychotic experiences in 99 adolescents and young adults (43 help-seeking service users, 16 siblings and 40 controls). Before ESM assessments, CT and depressive, anxiety and psychotic symptoms were assessed. RESULTS: Stress sensitivity, that is, the association between momentary stress and (i) negative affect and (ii) psychotic experiences, was modified by physical and emotional abuse and, partially, emotional and physical neglect, but not sexual abuse in service users and controls. While there was strong evidence for increased stress sensitivity in service users when high vs. low levels of CT were compared, a pattern of resilience was evident in controls, with attenuated, or no differences in, stress sensitivity in those with high vs. low CT levels. Less consistent findings were observed in siblings. CONCLUSIONS: Stress sensitivity may be an important risk and resilience mechanism through which CT impacts on mental health in youth.

Measuring total soil carbon with laser-induced breakdown spectroscopy (LIBS).

Improving estimates of carbon inventories in soils is currently hindered by lack of a rapid analysis method for total soil carbon. A rapid, accurate, and precise method that could be used in the field would be a significant benefit to researchers investigating carbon cycling in soils and dynamics of soil carbon in global change processes. We tested a new analysis method for predicting total soil carbon using laser-induced breakdown spectroscopy (LIBS). We determined appropriate spectral signatures and calibrated the method using measurements from dry combustion of a Mollisol from a cultivated plot. From this calibration curve we predicted carbon concentrations in additional samples from the same soil and from an Alfisol collected in a semiarid woodland and compared these predictions with additional dry combustion measurements. Our initial tests suggest that the LIBS method rapidly and efficiently measures soil carbon with excellent detection limits (approximately 300 mg/kg), precision (4-5%), and accuracy (3-14%). Initial testing shows that LIBS measurements and dry combustion analyses are highly correlated (adjusted r2 = 0.96) for soils of distinct morphology, and that a sample can be analyzed by LIBS in less than one minute. The LIBS method is readily adaptable to a field-portable instrument, and this attribute--in combination with rapid and accurate sample analysis--suggests that this new method offers promise for improving measurement of total soil carbon. Additional testing of LIBS is required to understand the effects of soil properties such as texture, moisture content, and mineralogical composition (i.e., silicon content) on LIBS measurements.

Thermooptic-based differential measurements of weak solute absorptions with an interferometer.

An interferometric method of measuring small differences between weak optical absorptions of solutions has been developed using the thermooptic effect. To record the small changes in optical path length ~lambda/200 due to heating, it was necessary to stabilize the fringe pattern with respect to slow thermal drift using a galvanometer-driven compensator plate controlled by a closed feedback loop. Fringe shifts from background absorptions were nulled out to better than 1 part in 400, permitting the measurement of differences in absorptions between two solutions that were l/100th of background. Using laser powers of 100 mW, absorptions approximately 5 x 10(-6) cm(-1) (base e) could be measured with CC1(4) solutions.

Thermally induced laser pulsing to measure weak optical absorptions.

Intracavity thermal lensing has been used to convert a cw dye laser to pulsed operation. The pulse width is a measure of the absorptivity of the solution in the cell. Experimental data are presented to show how this technique can be used to measure the absorptivity of a pure solvent and the concentration of a solute. The absorptivities (base e) of water, CH(3)OH, CH(3)COCH(3), and CCl(4) were determined to be, respectively, (1.24 +/- 0.17) x 10(-3), (1.01 +/- 0.07) x 10(-3), (1.74 +/- 0.09) x 10(-4), and (3.95 +/- x(-6) cm(-1). For CCl(4) in a 10-cm sample cell, the minimum detectable absorptivity is ~4 x 10(-6) cm(-1).

Reliable, inexpensive microcomputer interface for the optical multichannel analyzer (OMA).

A microprocessor system and interface for the optical multichannel analyzer (OMA) is described. The interface hardware and software are very simple and easy to implement. The microcomputer is used in conjunction with the OMA in the 2-d (two-dimensional) mode as the data acquisition, analysis and storage system for a 8-ps-resolution transient absorption spectrometer. However, the microcomputer and interface hardware and software are of general use in any application where rapid transfer, processing and storage of spectroscopic information from the OMA are required.