To help prevent motor vehicle accidents, there has been significant interest in finding an automated method to recognize signs of driver distraction, such as talking to passengers, fixing hair and makeup, eating and drinking, and using a mobile phone. In this paper, we present an automated supervised learning method called Drive-Net for driver distraction detection. Drive-Net uses a combination of a convolutional neural network (CNN) and a random decision forest for classifying images of a driver. We compare the performance of our proposed Drive-Net to two other popular machine-learning approaches: a recurrent neural network (RNN), and a multi-layer perceptron (MLP). We test the methods on a publicly available database of images acquired under a controlled environment containing about 22425 images manually annotated by an expert. Results show that Drive-Net achieves a detection accuracy of 95%, which is 2% more than the best results obtained on the same database using other methods.

Abstract. This study provides a detailed characterization of stratocumulus clearings off the US West Coast using remote sensing, reanalysis, and airborne in situ data. Ten years (2009–2018) of Geostationary Operational Environmental Satellite (GOES) imagery data are used to quantify the monthly frequency, growth rate of total area (GRArea), and dimensional characteristics of 306 total clearings. While there is interannual variability, the summer (winter) months experienced the most (least) clearing events, with the lowest cloud fractions being in close proximity to coastal topographical features along the central to northern coast of California, including especially just south of Cape Mendocino and Cape Blanco. From 09:00 to 18:00 (PST), the median length, width, and area of clearings increased from 680 to 1231, 193 to 443, and ∼67 000 to ∼250 000 km2, respectively. Machine learning was applied to identify the most influential factors governing the GRArea of clearings between 09:00 and 12:00 PST, which is the time frame of most rapid clearing expansion. The results from gradient-boosted regression tree (GBRT) modeling revealed that air temperature at 850 hPa (T850), specific humidity at 950 hPa (q950), sea surface temperature (SST), and anomaly in mean sea level pressure (MSLPanom) were probably most impactful in enhancing GRArea using two scoring schemes. Clearings have distinguishing features such as an enhanced Pacific high shifted more towards northern California, offshore air that is warm and dry, stronger coastal surface winds, enhanced lower-tropospheric static stability, and increased subsidence. Although clearings are associated obviously with reduced cloud fraction where they reside, the domain-averaged cloud albedo was actually slightly higher on clearing days as compared to non-clearing days. To validate speculated processes linking environmental parameters to clearing growth rates based on satellite and reanalysis data, airborne data from three case flights were examined. Measurements were compared on both sides of the clear–cloudy border of clearings at multiple altitudes in the boundary layer and free troposphere, with results helping to support links suggested by this study's model simulations. More specifically, airborne data revealed the influence of the coastal low-level jet and extensive horizontal shear at cloud-relevant altitudes that promoted mixing between clear and cloudy air. Vertical profile data provide support for warm and dry air in the free troposphere, additionally promoting expansion of clearings. Airborne data revealed greater evidence of sea salt in clouds on clearing days, pointing to a possible role for, or simply the presence of, this aerosol type in clearing areas coincident with stronger coastal winds.

We propose a deep learning based method for classification of commonly occurring pathologies in chest X-ray images. The vast number of publicly available chest X-ray images provides the data necessary for successfully employing deep learning methodologies to reduce the misdiagnosis of thoracic diseases. We applied our method to the classification of two example pathologies, pulmonary nodules and cardiomegaly, and we compared the performance of our method to three existing methods. The results show an improvement in AUC for detection of nodules and cardiomegaly compared to the existing methods.

Purpose-To develop a fast and accurate convolutional neural network based method for segmentation of thalamic nuclei.Methods-A cascaded multi-planar scheme with a modified residual U-Net architecture was used to segment thalamic nuclei on conventional and white-matter-nulled (WMn) magnetization prepared rapid gradient echo (MPRAGE) data. A single network was optimized to work with images from healthy controls and patients with multiple sclerosis (MS) and essential tremor (ET), acquired at both 3T and 7T field strengths. WMn-MPRAGE images were manually delineated by a trained neuroradiologist using the Morel histological atlas as a guide to generate reference ground truth labels. Dice similarity coefficient and volume similarity index (VSI) were used to evaluate performance. Clinical utility was demonstrated by applying this method to study the effect of MS on thalamic nuclei atrophy.Results-Segmentation of each thalamus into twelve nuclei was achieved in under a minute. For 7T WMn-MPRAGE, the proposed method outperforms current state-of-the-art on patients with ET with statistically significant improvements in Dice for five nuclei (increase in the range of 0.05-0.18) and VSI for four nuclei (increase in the range of 0.05-0.19), while performing comparably

No abstract