USA As Represented by Secretary of the Navy

Carnegie Mellon University

University of Virginia Patent Foundation

Smoke Detective, LLC

Gustavo Kunde Rohde

Frederick Lanni

Wei Wang

Jonathan M Nichols

Andrew M Southerland

Russell P Mills

Stephen C Davis, Jr

Omar Uribe

Mark McDonald

Yan Zhuang

Frank Bucholtz

Geoffrey A Cranch

Meredith N Hutchinson

Chad M Aldridge

Dejan Slepcev

Mohamed Abul Hassan

Cheng Chen

Nicole Menkart

John A Ozolek

An automated and quantitative facial weakness screening framework that utilizes a Bi-LSTM network to model the temporal dynamics among the shape and appearance features. The technique is beneficial to assist the paramedics or other users to identify the facial weakness in the field or, more importantly, whenever expertise in neurology is not available either for emergency patient triage (e.g., pre-hospital stroke care) or chronic disease management (e.g., Bell's palsy rehabilitation screen), leading to increased coverage and earlier treatment. The technique provides visualizable and interpretable results to increase its transparency and interpretability. The technique provides for inexpensive solutions that can be used in areas underserved by non-neurologists to more readily identify neurological deficits such as facial weakness in the field or other environment.

System, method and computer readable medium for video-based facial weakness analysis for detecting neurological deficits

The disclosed embodiments provide systems and methods for predicting presence of one or more neurological deficits. The system may include a microphone, a camera, one or more memory devices storing instructions, and one or more processors configured to execute the instructions to extract audio information including a period density entropy coefficient and a mel frequency cepstral coefficient from an audio feed received from the microphone. Additionally, the instructions may cause the processor to determine position and depth information of eye movement from a video feed received from the camera and detect features of interest including facial landmarks, spatial orientation of limbs, and positional information of limb movements from the video feed. The one or more processors may further extract the features of interest from the video feed and process the extracted features of interest by aligning the extracted features of interest to a common reference.

System and method for automated detection of neurological deficits

Methods, apparatuses, and systems for calculating time delays by a Wasserstein approach are provided. A plurality of signals are recorded by a plurality of sensors (three or more), respectively, and received at a controller. The plurality of signals recorded by the plurality of sensors are generated in response to a signal emitted by a source. The plurality of signals are converted into a plurality of probability density functions. A cumulative distribution transform for each of the plurality of probability density functions is calculated. A time delay for each unique pair of the plurality of sensors is calculated by minimizing a Wasserstein distance between two cumulative distribution transforms corresponding to the unique pair of the plurality of sensors.

Methods, apparatuses and systems for time delay estimation

A fire detection device is provided that has a camera that captures a reference image and a measured image. A processor compares intensity of the measured image to intensity of the reference image and uses this comparison to determine if an alarm is generated to indicate the presence of fire. The intensity of the measured image may be the total number of photons of the measured image, and the intensity of the reference image may be the total number of photons of the reference image. In other arrangements, the intensity may be measured between individual corresponding pixels of the reference and measured images.

Fire detection apparatus utilizing a camera

Image-based fire detection methods are provided which include analyzing a field of view based on a plurality of images, determining if changes occur, and analyzing whether any such changes are due to smoke-like material. Changes due to large objects are ruled out, and drift in the scene or shaking of the image collection device is accounted for by the method. The methods employ representing images of the field of view with quantifiable values and comparing to thresholds established for each event, and activating an alarm if changes due to smoke-like material are detected. The fire detection methods may also include monitoring a field of view for changes, and only performing the analysis on the images if changes are detected. The monitoring and validating stages of the methods include collecting images at different sampling rates for different amounts of information.

Smoke detection system and method using a camera

A method performed by one or more processors includes retrieving a set of images; selecting, based on the images retrieved, a reference template; calculating optimal transportation plans between the reference template and each of the images in the set; and calculating, using the optimal transportation plans, a linear optimal transportation metric between at least two images in the set.

Quantitative comparison of image data using a linear optimal transportation

A method performed by one or more processors, includes: receiving an image to be segmented into one or more representations of one or more biological structures; accessing data representing a set of biological structures that are derived from other biological structures delineated in a training image, wherein the training image is associated with a level of modality that corresponds to a level of modality associated with the image to be segmented; computing a normalized cross correlation of the received image against one or more of the biological structures in the set of biological structures; generating, based on computing, seed data representing an estimate of a spatial organization of the one or more biological structures in the received image; and segmenting, based on a normalized cross correction of the received image to the seed data, the received image into the one or more representations of the one or more biological structures.

Segmenting biological structures from microscopy images

A system and method for encoding zero and one bits for transmission, including generating a first signal from a non-linear chaotic system to represent the one bit, with the signal's embedded vectors being within the non-linear system's attractor set, and generating a second signal from the non-linear system to represent the zero bit, with the signal's embedded vectors being outside the non-linear system's attractor set. The second signal encoding the zero bit can be generated by adding together two chaotic signals arising from the non-linear system initialized with different initial conditions, and weighting the second signal to have approximately the same energy as the first signal. One suitable chaotic systems is a Lorenz system. Systems and methods for decoding a transmitted stream of signals compare a detection statistic of the received stream of signals to a threshold value that depends on the chaotic system.

Growing community of inventors

Pittsburgh, PA, United States of America

Gustavo Kunde Rohde