Abstract: A method can be used to reduce the memory storage and energy used by deep neural networks. The method can include determining the weights associated with the deep neural network. An input feature map can be received and used with the weights to generate approximated weights. Using the approximated weights and the input feature map a convolution inference can be performed.

Abstract: Devices, systems, and methods of using one or more memristors as a radiation sensor are enabled. A memristor can be attractive as a sensor due to its passive low power characteristics. Medical and environment monitoring are contemplated use cases. Sensing radiation as part of a security system (at an airport for example) and screening food for radiation exposure are also possible uses. The memristor as a radiation sensor may possibly provide an inexpensive and easy alternative to personal thermoluminescent dosimeters (TLD). Memristor devices with high current and low power operation may be attached with wearable plastic substrates. An example device includes two metal strips with a 50 ?m thick layer of TiO2 memristor material. The device may be made large relative to traditional memristors which are nanometers in scale but its increased thickness can significantly increase the probability of radiation interaction with the memristor material.

Abstract: Devices, systems, and methods of using one or more memristors as a radiation sensor are enabled. A memristor can be attractive as a sensor due to its passive low power characteristics. Medical and environment monitoring are contemplated use cases. Sensing radiation as part of a security system (at an airport for example) and screening food for radiation exposure are also possible uses. The memristor as a radiation sensor may possibly provide an inexpensive and easy alternative to personal thermoluminescent dosimeters (TLD). Memristor devices with high current and low power operation may be attached with wearable plastic substrates. An example device includes two metal strips with a 50 ?m thick layer of TiO2 memristor material. The device may be made large relative to traditional memristors which are nanometers in scale but its increased thickness can significantly increase the probability of radiation interaction with the memristor material.

Abstract: Devices, systems, and methods of using one or more memristors as a radiation sensor are enabled. A memristor can be attractive as a sensor due to its passive low power characteristics. Medical and environment monitoring are contemplated use cases. Sensing radiation as part of a security system (at an airport for example) and screening food for radiation exposure are also possible uses. The memristor as a radiation sensor may possibly provide an inexpensive and easy alternative to personal thermoluminescent dosimeters (TLD). Memristor devices with high current and low power operation may be attached with wearable plastic substrates. An example device includes two metal strips with a 50 ?m thick layer of TiO2 memristor material. The device may be made large relative to traditional memristors which are nanometers in scale but its increased thickness can significantly increase the probability of radiation interaction with the memristor material.

Abstract: An electrocardiogram (ECG) processor is disclosed. The ECG processor includes ECG sampling circuitry configured in a first mode to acquire a continuous ECG sample set from an ECG signal by digitally sampling the ECG signal at a Nyquist rate for a first predetermined number of heartbeats and in a second mode to acquire a non-continuous ECG sample set from the ECG signal for a second predetermined number of heartbeats by digitally sampling active regions of the ECG signal that contain a PQRST complex and not from silent regions between adjacent PQRST complexes. The ECG processor also includes processing circuitry configured to determine from the continuous ECG sample set relative locations of the active regions and provide the relative locations of the active regions to the ECG sampling circuitry for sampling the ECG signal in the second mode.

Abstract: A glucose sensor includes an insulating metal oxide layer and at least one pair of metallic electrodes arranged on the insulating metal oxide layer and separated by a gap containing the metal oxide layer. In operation, a probe including a voltage supply and current sensor can provide a voltage difference across the first and second metallic electrodes while a sample is present across the gap between the electrodes. A measured current between the first and second metallic electrodes when the voltage difference is provided can be correlated to a glucose level of the sample.

Abstract: Devices, systems, and methods of using one or more memristors as a radiation sensor are enabled. A memristor can be attractive as a sensor due to its passive low power characteristics. Medical and environment monitoring are contemplated use cases. Sensing radiation as part of a security system (at an airport for example) and screening food for radiation exposure are also possible uses. The memristor as a radiation sensor may possibly provide an inexpensive and easy alternative to personal thermoluminescent dosimeters (TLD). Memristor devices with high current and low power operation may be attached with wearable plastic substrates. An example device includes two metal strips with a 50 ?m thick layer of TiO2 memristor material. The device may be made large relative to traditional memristors which are nanometers in scale but its increased thickness can significantly increase the probability of radiation interaction with the memristor material.

Abstract: Data compression using a memristive crossbar is enabled. Conductances of memristors may be set such that the memristors of the crossbar act as coefficients of a wavelet transformation coefficient matrix with respect to voltage signals applied to input rows of the crossbar. The memristors may act as coefficients of the transpose of the wavelet transformation coefficient matrix when voltage signals are applied to input columns of the crossbar. Hence, the memristive crossbar may be used to implement a two dimensional (2D) discrete wavelet transform (DWT) on two dimensional data (e.g., image data) encoded in the voltage signals. The resulting currents in the columns of the memristive crossbar may be integrated and converted to voltage signals that are fed back into columns of the memristive crossbar such that the rows of the memristive crossbar output electronic signals that correspond to the image data compressed in accordance with Haar 2D-DWT image compression.

Abstract: Architecture and a method for maximally stable extremal regions (MSERs)-based detection of exudates in an ocular fundus is disclosed. The architecture includes a communication interface to receive pixels of an ocular fundus image. The architecture further includes processing circuitry that is coupled to the communication interface. The processing circuitry is configured to automatically provide labels for light image regions and dark image regions within the ocular fundus image for a given intensity threshold and find MSERs within the ocular fundus image based on the labels. The architecture also determines MSER regions based on the MSER criteria and then highlights the pixels of the ocular fundus image that are located within MSER regions to indicate the exudates in the ocular fundus. The architecture is further configured to determine MSER ellipses parameters based on MSER regions and MSER criteria and then highlight the locations of the exudates in the ocular fundus.

Abstract: Data compression using a memristive crossbar is enabled. Conductances of memristors may be set such that the memristors of the crossbar act as coefficients of a wavelet transformation coefficient matrix with respect to voltage signals applied to input rows of the crossbar. The memristors may act as coefficients of the transpose of the wavelet transformation coefficient matrix when voltage signals are applied to input columns of the crossbar. Hence, the memristive crossbar may be used to implement a two dimensional (2D) discrete wavelet transform (DWT) on two dimensional data (e.g., image data) encoded in the voltage signals. The resulting currents in the columns of the memristive crossbar may be integrated and converted to voltage signals that are fed back into columns of the memristive crossbar such that the rows of the memristive crossbar output electronic signals that correspond to the image data compressed in accordance with Haar 2D-DWT image compression.

Abstract: An electrocardiogram (ECG) processor is disclosed. The ECG processor includes ECG sampling circuitry configured in a first mode to acquire a continuous ECG sample set from an ECG signal by digitally sampling the ECG signal at a Nyquist rate for a first predetermined number of heartbeats and in a second mode to acquire a non-continuous ECG sample set from the ECG signal for a second predetermined number of heartbeats by digitally sampling active regions of the ECG signal that contain a PQRST complex and not from silent regions between adjacent PQRST complexes. The ECG processor also includes processing circuitry configured to determine from the continuous ECG sample set relative locations of the active regions and provide the relative locations of the active regions to the ECG sampling circuitry for sampling the ECG signal in the second mode.

Abstract: The various embodiments herein provide a Uniform Circular Displaced Sensor Array (UC-DSA) system and method for measuring/estimating Direction of Arrival (DOA) of a wireless signal. The UC-DSA system comprises at least a set of two circular antenna arrays. The two circular antenna arrays have a number of elements. A Radio frequency (RF) receiver captures a wireless signal incident on a circular antenna array. A Direction of Arrival (DOA) estimator processes a received input signal and a Triangulation system provides the exact location of the source of the wireless signal. The two circular antenna arrays with the same number of elements are placed on different radii, and are shifted to have equal separation between inner elements and outer elements.

Abstract: A medical device having automated electrocardiogram (ECG) feature extraction is disclosed. The medical device includes input circuitry configured to receive an ECG signal. Processing circuitry coupled to the input circuitry is configured to identify at least one fiducial point of heartbeat signature of the ECG signal. The processing circuitry is further configured to perform substantially simultaneously both a discrete wavelet transform (DWT) and a curve length transform (CLT) to identify the at least one fiducial point.

Abstract: Architecture and a method for maximally stable extremal regions (MSERs)-based detection of exudates in an ocular fundus is disclosed. The architecture includes a communication interface to receive pixels of an ocular fundus image. The architecture further includes processing circuitry that is coupled to the communication interface. The processing circuitry is configured to automatically provide labels for light image regions and dark image regions within the ocular fundus image for a given intensity threshold and find MSERs within the ocular fundus image based on the labels. The architecture also determines MSER regions based on the MSER criteria and then highlights the pixels of the ocular fundus image that are located within MSER regions to indicate the exudates in the ocular fundus. The architecture is further configured to determine MSER ellipses parameters based on MSER regions and MSER criteria and then highlight the locations of the exudates in the ocular fundus.

Abstract: Methods and systems for detecting and/or tracking one or more objects utilize depth data. An example method of detecting one or more objects in image data includes receiving depth image data corresponding to a depth image view point relative to the one or more objects. A series of binary threshold depth images are formed from the depth image data. Each of the binary threshold depth images is based on a respective depth. One or more depth extremal regions in which image pixels have the same value are identified for each of the binary depth threshold images. One or more depth maximally stable extremal regions are selected from the identified depth extremal regions based on change in area of the one or more respective depth extremal regions for different depths.

Abstract: Hardware architecture for real-time extraction of maximally stable extremal regions (MSERs) is disclosed. The architecture includes a communication interface and processing circuitry that are configured in hardware to receive a data stream of an intensity image in real-time and provide labels for image regions within the intensity image that match a given intensity threshold. The communication interface and processing circuitry are also configured in hardware to find extremal regions within the intensity image based upon the labels and to determine MSER ellipses parameters based upon the extremal regions and MSER criteria. In at least one embodiment, the MSER criteria include minimum and maximum MSER areas, and an acceptable growth rate value for MSER area. In another embodiment, the MSER criteria include a nested MSER tolerance value.

Abstract: A medical device having automated electrocardiogram (ECG) feature extraction is disclosed. The medical device includes input circuitry configured to receive an ECG signal. Processing circuitry coupled to the input circuitry is configured to identify at least one fiducial point of heartbeat signature of the ECG signal. The processing circuitry is further configured to perform substantially simultaneously both a discrete wavelet transform (DWT) and a curve length transform (CLT) to identify the at least one fiducial point.

Abstract: Methods and systems for detecting and/or tracking one or more objects utilize depth data. An example method of detecting one or more objects in image data includes receiving depth image data corresponding to a depth image view point relative to the one or more objects. A series of binary threshold depth images are formed from the depth image data. Each of the binary threshold depth images is based on a respective depth. One or more depth extremal regions in which image pixels have the same value are identified for each of the binary depth threshold images. One or more depth maximally stable extremal regions are selected from the identified depth extremal regions based on change in area of the one or more respective depth extremal regions for different depths.

Abstract: Methods and systems process an MRI image to detect cancer. A method includes forming a series of binary threshold intensity images from an MRI image of a patient. Each of the binary threshold intensity images is based on a respective intensity. The binary threshold intensity images are processed to identify one or more bright extremal regions in which image pixels have the same value, and for which corresponding image pixels in the MRI image have a higher intensity than surrounding image pixels in the MRI image. One or more bright maximally stable extremal regions are selected from the identified bright extremal regions based on change in area of one or more respective bright extremal regions for different binary threshold images in the series. At least one of the selected one or more bright maximally stable extremal regions may be identified as potentially cancerous.

Abstract: Architecture for real-time extraction of maximally stable extremal regions (MSERs) is disclosed. The architecture includes a communication interface and processing circuitry that are configured in hardware to receive a data stream of an intensity image in real-time and provide labels for light image regions and dark image regions within the intensity image that match a given intensity threshold during a single processing pass. The communication interface and processing circuitry are also configured in hardware to find extremal regions within the intensity image based upon the labels and to determine MSER ellipses parameters based upon the extremal regions and MSER criteria. In at least one embodiment, the MSER criteria include minimum and maximum MSER areas, and an acceptable growth rate value for MSER areas. In another embodiment, the MSER criteria include a nested MSER tolerance value.