Abstract: An electronic device for gesture recognition on resource-constrained devices is provided. The electronic device controls storage of a plurality of first consecutive image frames in a first buffer of a first length. The plurality of first consecutive image frames corresponds to the first length. The electronic device recognizes a first hand sign of a plurality of hand signs in a first subset of image frames of the plurality of first consecutive image frames. The electronic device controls storage of the recognized first hand sign in a second buffer of a second length based on the determination that a ratio of a number of the first subset of image frames and the first length is one of equal to or greater than the threshold. The electronic device determines a gesture corresponding to one or more hand signs of the plurality of hand signs stored in the second buffer.

Abstract: A system and method for customization of software applications with neural network-based features is disclosed. The system acquires information related to one or more functional components of an electronic device and usage data associated with the electronic device. The system selects a computer vision task, based on the acquired information and the usage data, and determines constraints associated with an implementation of the selected computer vision task on the electronic device. The system selects a first neural network as a seed model for the selected computer vision task and execute operations, including a neural architecture search with the seed model and the constraints as input, to obtain a second neural network which is trained on the selected computer vision task. The system updates a software application on the electronic device to include an end-user feature that implements the second neural network for the selected computer vision task.

Abstract: A system collects first data associated with user. The first data includes historical health data and set of sensor data corresponding to a set of health-monitoring parameters. The system applies first Artificial Intelligence (AI) model on the first data to compute indicators which reflect a deviation in a health condition of the user with respect to reference values. Based on the indicators, the system generates first inference data comprising labels or tags associated with a cause of the deviation. Based on the first inference data, the system determines a first requirement for which the user is required to visit a first healthcare center. The system further determines a first set of user-related data associated with the first requirement, based on the first data and the first inference data. Thereafter, the system transfers the first set of user-related data to an electronic healthcare system associated with the first healthcare center.

Abstract: A cluster-based approach and template-based approach are combined to segment brain matter from a three-dimensional MRI image of voxels. The morphological information captured by the template-based approach may be used to refine the segmentation produced by the cluster-based approach. Conversely, the “similarity” information captured by the cluster-based approach may be used to refine the segmentation produced by the template-based approach.

Abstract: In one aspect, a computer-implemented method corrects a multi-sphere head model used in dipole localization for a set of magnetic field sensors (MEG sensors) by replacing ghost spheres with replacement spheres that are not ghost spheres. One type of ghost sphere completely encloses the brain volume but is so large that a center of the sphere is outside the brain volume. Another type of ghost sphere lies entirely outside the brain volume. Various approaches for correcting ghost spheres are disclosed.

Abstract: In one aspect, a computer-implemented method corrects a multi-sphere head model used in dipole localization for a set of magnetic field sensors (MEG sensors) by replacing ghost spheres with replacement spheres that are not ghost spheres. One type of ghost sphere completely encloses the brain volume but is so large that a center of the sphere is outside the brain volume. Another type of ghost sphere lies entirely outside the brain volume. Various approaches for correcting ghost spheres are disclosed.

Abstract: In one aspect, a computer-implemented method corrects a multi-sphere head model used in dipole localization for a set of magnetic field sensors (MEG sensors) by replacing ghost spheres with replacement spheres that are not ghost spheres. One type of ghost sphere completely encloses the brain volume but is so large that a center of the sphere is outside the brain volume. Another type of ghost sphere lies entirely outside the brain volume. Various approaches for correcting ghost spheres are disclosed.

Abstract: In one aspect, multiscale filtering is used to normalize the intensities of voxels in an MRI image. A multiscale filter is applied to the raw MRI image. This image is compared to the original image. Luma aberrations (i.e., intensity variations) are corrected based on this comparison. In one approach, the intensity of the image is increased for voxels that are dimmer than in the multiscale filtered version, and decreased for voxels that are brighter than the multiscale filtered version. In another aspect, additional features are created based on multiscale gradients. These may be used in combination with other approaches to segment the MRI image. Voxels with positive gradients may represent brain gray matter bordered by brain white matter. Voxels with negative gradients may represent brain white matter bordered by brain grain matter.

Abstract: A cluster-based approach and template-based approach are combined to segment brain matter from a three-dimensional MRI image of voxels. The morphological information captured by the template-based approach may be used to refine the segmentation produced by the cluster-based approach. Conversely, the “similarity” information captured by the cluster-based approach may be used to refine the segmentation produced by the template-based approach.

Abstract: A system assists users in time and frequency analysis of magnetoencephalography (MEG) signals. In one aspect, a system includes an analysis module, a configuration module and a user interface. The analysis module performs a time and frequency analysis of the MEG signal, for example a short time Fourier transform (STFT) or a continuous wavelet transform (CWT) analysis. The analysis is parameterized by a parameter set that affects the time and frequency resolution of the analysis, for example window size and overlap size for STFT or center frequency and decay parameter for CWT. The configuration module automatically determines or assists the user to determine correct values for the parameter set.

Abstract: A surface model is generated from a three-dimensional volume model of a person's head. The person's head is modelled as a three-dimensional volume model of loss values (i.e., absorption values). Wave vectors are launched towards the volume model. Each wave vector is characterized by a wavelength and a capture direction (direction of propagation). The launched wave vectors are absorbed by the volume and the point at which they are absorbed (referred to as the intersection point) is determined. The surface model of the person's head is generated from the intersection points of the wave vectors.

Abstract: A surface model is generated from a three-dimensional volume model of a person's head. The person's head is modelled as a three-dimensional volume model of loss values (i.e., absorption values). Wave vectors are launched towards the volume model. Each wave vector is characterized by a wavelength and a capture direction (direction of propagation). The launched wave vectors are absorbed by the volume and the point at which they are absorbed (referred to as the intersection point) is determined. The surface model of the person's head is generated from the intersection points of the wave vectors.

Abstract: In one aspect, a computer-implemented method corrects a multi-sphere head model used in dipole localization for a set of magnetic field sensors (MEG sensors) by replacing ghost spheres with replacement spheres that are not ghost spheres. One type of ghost sphere completely encloses the brain volume but is so large that a center of the sphere is outside the brain volume. Another type of ghost sphere lies entirely outside the brain volume. Various approaches for correcting ghost spheres are disclosed.

Abstract: In one aspect, a computer-implemented method corrects a multi-sphere head model used in dipole localization for a set of magnetic field sensors (MEG sensors) by replacing ghost spheres with replacement spheres that are not ghost spheres. One type of ghost sphere completely encloses the brain volume but is so large that a center of the sphere is outside the brain volume. Another type of ghost sphere lies entirely outside the brain volume. Various approaches for correcting ghost spheres are disclosed.

Abstract: A surface model is generated from a three-dimensional volume model of a person's head. The person's head is modelled as a three-dimensional volume model of loss values (i.e., absorption values). Wave vectors are launched towards the volume model. Each wave vector is characterized by a wavelength and a capture direction (direction of propagation). The launched wave vectors are absorbed by the volume and the point at which they are absorbed (referred to as the intersection point) is determined. The surface model of the person's head is generated from the intersection points of the wave vectors.

Abstract: A surface model is generated from a three-dimensional volume model of a person's head. The person's head is modelled as a three-dimensional volume model of loss values (i.e., absorption values). Wave vectors are launched towards the volume model. Each wave vector is characterized by a wavelength and a capture direction (direction of propagation). The launched wave vectors are absorbed by the volume and the point at which they are absorbed (referred to as the intersection point) is determined. The surface model of the person's head is generated from the intersection points of the wave vectors.

Abstract: An application delivery controller for improving remote healthcare services is disclosed. The application delivery controller may be configured to provide improved service routing and security. In some implementations, the application delivery controller may receive a service request from a client device. The application delivery controller may determine routing parameters based on a requested service of the service request, authenticate access from the client device based on the requested service, a security zone of the requested service, and a token, and route the requested service to the client device based on the routing parameters and the authenticated access.

Abstract: A system assists users in time and frequency analysis of magnetoencephalography (MEG) signals. In one aspect, a system includes an analysis module, a configuration module and a user interface. The analysis module performs a time and frequency analysis of the MEG signal, for example a short time Fourier transform (STFT) or a continuous wavelet transform (CWT) analysis. The analysis is parameterized by a parameter set that affects the time and frequency resolution of the analysis, for example window size and overlap size for STFT or center frequency and decay parameter for CWT. The configuration module automatically determines or assists the user to determine correct values for the parameter set.

Abstract: The location of the optical axis of a plenoptic imaging system is determined. In one approach, the plenoptic image is noise filtered, down-sampled, low-pass filtered, and again noise filtered. Slices through this image that have the highest power are used to determine the location of the optical axis. Once the location of the optical axis is determined, corrections for aberrations such as distortion can be applied.

Abstract: The location of the optical axis of a plenoptic imaging system is determined. In one approach, the plenoptic image is noise filtered, down-sampled, low-pass filtered, and again noise filtered. Slices through this image that have the highest power are used to determine the location of the optical axis. Once the location of the optical axis is determined, corrections for aberrations such as distortion can be applied.