Abstract: Disclosed is a method for region-of-interest enhanced PET image reconstruction based on multi-task learning, which comprises the following steps: firstly, acquiring a backprojection image of the PET original data, and designing a main task of establishing a mapping between the backprojection image and a reconstructed PET image by using a three-dimensional deep convolution neural network. A new auxiliary task 1 is designed to predict a computerized tomography (CT) image with the same anatomical structures as the PET image reconstructed from the backprojection image, so as to reduce the noise in the reconstructed PET image by using the local smoothing information of the high-resolution CT image.

Abstract: Systems, methods, and computer-readable media are disclosed for systems and methods multimodal indexing of video using machine learning. An example method may include deceiving, by a video encoder of an audio-video transformer neural network comprising one or more computer processors coupled to memory, a first frame and a second frame associated with a first segment of a video. The example method may also include receiving, by an audio encoder of the audio-video transformer neural network, an audio spectrogram comprising first audio data associated with the first segment of the video. generating, by the video encoder, a first video embedding. The example method may also include generating, by the audio encoder, a first audio embedding. The example method may also include determining a fusion of the first video embedding and the first audio embedding using a multimodal bottleneck token. The example method may also include determining an output including the first video embedding and the first audio embedding.

Abstract: A data transfer method, a data transfer device and a computer readable storage medium. The data transfer device comprises at least one first-stage memory, at least one second-stage memory and at least one third-stage memory which are connected in sequence. The data transfer method comprises: receiving a first input comprising a number of input lanes and/or a number of output lanes; selecting a corresponding second-stage memory and controlling a first read signal and a second read signal, according to the number of input lanes and/or the number of output lanes; storing data of the input lanes through the at least one first-stage memory; reading data of the first-stage memory and writing the data into the corresponding second-stage memory when the first read signal is enabled; and reading data of the corresponding second-stage memory and writing the data into the third-stage memory when the second read signal is enabled.

Abstract: A method and an apparatus for controlling read timing, and a computer-readable storage medium. The apparatus for controlling read timing includes an on-chip memory. The method for controlling read timing includes: writing input data into the on-chip memory; performing an end-of-frame detection on the input data; after detecting an end of a frame, counting input data rows and generating a read timing according to a preset Porch parameter; determining whether the on-chip memory will be read to be empty or written to be full according to an input data row count and the generated read timing; and when the on-chip memory will be read to be empty or written to be full, adjusting the read timing until the on-chip memory will not be read to be empty or written to be full, and reading and outputting data of the on-chip memory according to the read timing.

Abstract: The present disclosure provides a display data processing method, including receiving a pixel display data stream, wherein the pixel display data stream includes a plurality of sub-pixel display data corresponding to pixels respectively, converting the pixel display data stream into a pixel island display data stream, wherein the pixel island display data stream includes a plurality of viewpoint display data corresponding to pixel islands respectively, and generating a single-row display datum based on the pixel island display data stream, wherein the viewpoint display data included in the single-row display datum correspond to output channels of a plurality of chips in a one-to-one relationship. The present disclosure also relates to a display data processing device using the method, and a display device based on a multi-viewpoint pixel island architecture which includes the display data processing device.

Abstract: An adaptive optimization method for a 5G weight value includes: mapping 5G MR sample points to grid elements of a 3D grid; identifying a type of each grid element; determining a weight value combination; determining a first coverage value based on a first path loss for a center grid element of the business distribution center grid cluster and each weight value combination, determining a first coverage improvement value based on the first coverage value, determining from the weight value combination an initial weight value combination; determining a second coverage value for a weak depth coverage grid element based on a second path loss and each initial weight value combination, determining a second coverage improvement value based on the second coverage value, and determining, from the initial weight value combination, a final weight value combination.

Abstract: The present disclosure provides a display data processing method, including receiving a pixel display data stream, wherein the pixel display data stream includes a plurality of sub-pixel display data corresponding to pixels respectively, converting the pixel display data stream into a pixel island display data stream, wherein the pixel island display data stream includes a plurality of viewpoint display data corresponding to pixel islands respectively, and generating a single-row display datum based on the pixel island display data stream, wherein the viewpoint display data included in the single-row display datum correspond to output channels of a plurality of chips in a one-to-one relationship. The present disclosure also relates to a display data processing device using the method, and a display device based on a multi-viewpoint pixel island architecture which includes the display data processing device.

Abstract: A data transfer method, a data transfer device and a computer readable storage medium. The data transfer device comprises at least one first-stage memory, at least one second-stage memory and at least one third-stage memory which are connected in sequence. The data transfer method comprises: receiving a first input comprising a number of input lanes and/or a number of output lanes; selecting a corresponding second-stage memory and controlling a first read signal and a second read signal, according to the number of input lanes and/or the number of output lanes; storing data of the input lanes through the at least one first-stage memory; reading data of the first-stage memory and writing the data into the corresponding second-stage memory when the first read signal is enabled; and reading data of the corresponding second-stage memory and writing the data into the third-stage memory when the second read signal is enabled.

Abstract: Provided in the present disclosure is a clock receiving circuit. The clock receiving circuit comprises a common-mode voltage adjustment module, an amplitude amplification module and a level conversion module. The common-mode voltage adjustment module comprises an n-type signal conversion unit, a high-level n-type signal output end, a low-level n-type signal output end, a p-type signal conversion unit, a high-level p-type signal output end and a low-level p-type signal output end. The amplitude amplification module comprises a p-type current source transistor, an n-type current source transistor, a p-type transistor differential pair, an n-type transistor differential pair and a bias control unit. The level conversion module is used for converting, into a CMOS level signal, a CML level signal which is output by the amplitude amplification circuit. Further provided in the present disclosure is an electronic device comprising the clock receiving circuit.

Abstract: A method and an apparatus for controlling read timing, and a computer-readable storage medium. The apparatus for controlling read timing includes an on-chip memory. The method for controlling read timing includes: writing input data into the on-chip memory; performing an end-of-frame detection on the input data; after detecting an end of a frame, counting input data rows and generating a read timing according to a preset Porch parameter; determining whether the on-chip memory will be read to be empty or written to be full according to an input data row count and the generated read timing; and when the on-chip memory will be read to be empty or written to be full, adjusting the read timing until the on-chip memory will not be read to be empty or written to be full, and reading and outputting data of the on-chip memory according to the read timing.

Abstract: Disclosed is a three-dimensional low-dose PET reconstruction method based on deep learning. The method comprises the following steps: back projecting low-dose PET raw data to the image domain to maintain enough information from the raw data; selecting an appropriate three-dimensional deep neural network structure to fit the mapping between the back projection of the low-dose PET and a standard-dose PET image; after learning from the training samples the network parameters are fixed, realizing three-dimensional PET image reconstruction starting from low-dose PET raw data, thereby obtaining a low-dose PET reconstructed image which has a lower noise and a higher resolution compared with the traditional reconstruction algorithm and image domain noise reduction processing.

Abstract: Systems and techniques for retrieving video data associated with a selected attribute are described. The systems and techniques include receiving a multimodal input associated with the attribute for querying a catalog of video data to identify video data including the attribute. A first embedding is determined for the input using a first encoder to map the input to a representation space. A second embedding is determined for the video data to map the video data to the representation space. A similarity score is determined between the video data and the input based on a distance between the embeddings. The video data associated with the attribute may be selected based on the similarity score.

Abstract: Apparatuses, systems, and techniques to perform federated training of neural networks while maintaining control over dissemination of local models of neural networks from which aspects of local training data might be extracted. In at least one embodiment, a neural network is trained on local training data and a local model is provided to be aggregated with other local models into a global model that is in turn used for further local model training, wherein a provided local model or training is adjusted to reduce an ability to extract aspects of local training data therefrom.

Abstract: Disclosed is a method for automatic classification of pathological images based on a staining intensity matrix. This method directly extracts the staining intensity matrix irrelevant to a stain ratio, a staining platform, a scanning platform and some human factors in the pathological image as the feature information of classification, without restoring normalized stained images, while retaining all impurity-free information related to diagnosis. It avoids the phenomenon that the diagnostic effect of the existing computer-aided diagnosis method of pathological images based on the traditional color normalization method changes with the changes of the selected standard pathological sections. Moreover, it avoids the error introduced by the need to restore the stained image, and has a higher diagnostic accuracy and a more stable diagnostic effect. At the same time, the method can realize the diagnosis of pathological images in a shorter time, which is easy to realize and more practical.

Abstract: Apparatuses, systems, and techniques to perform registration among images. In at least one embodiment, one or more neural networks are trained to indicate registration of features in common among at least two images by generating a first correspondence by simulating a registration process of registering an image and applying the at least two images and the first correspondence to a neural network to derive a second correspondence of the features in common among the at least two images.

Abstract: Apparatuses, systems, and techniques to perform registration among images. In at least one embodiment, one or more neural networks are trained to indicate registration of features in common among at least two images by generating a first correspondence by simulating a registration process of registering an image and applying the at least two images and the first correspondence to a neural network to derive a second correspondence of the features in common among the at least two images.

Abstract: Disclosed in embodiments of the present disclosure are a driver chip and a display device. The driver chip includes two opposite long edges, and two opposite short edges connected to the long edges; the driver chip includes multiple output pins and multiple input pins; the output pins are close to one of the long edges; the input pins include first input pins, second input pins, and third input pins; the first input pins are close to the long edge opposite to the output pins; the second input pins are close to the short edges; the third input pins are close to the long edge and located on both sides of the output pins.

Abstract: Disclosed is a CT image generation method for attenuation correction of PET images. According to the method, a CT image and a PET image at T1 and a PET image at T2 are acquired and input into a trained deep learning network to obtain a CT image at T2; the CT image can be applied to the attenuation correction of the PET image, thereby obtaining more an accurate PET AC (Attenuation Correction) image. According to the CT image generation method for attenuation correction of PET images, the dosage of X-rays received by a patient in the whole image acquisition stage can be reduced, and physiological and psychological pressure of the patient is relieved. In addition, the later image acquisition only needs a PET imaging device, without the need of PET/CT device, cost of imaging resource distribution can be reduced, and the imaging expense of the whole stage is reduced.

Abstract: A method for imaging may include directing a PET scanner to perform scans of a subject injected with a tracer during a first time period. The method may also include generating PET images by reconstructing the PET data that are generated by the PET scanner based on the scans of the subject. The method may also include determining a first portion of a blood input function of the tracer in the subject based on the PET images. The method may also include determining an integral of a second portion of the blood input function based on a kinetic model, the PET data, and the first portion of the blood input function. The method may also include determining kinetic parameters of the kinetic model based on the PET data, the first portion of the blood input function, and the integral of the second portion of the blood input function.

Abstract: Provided are a driving method, a gate drive unit and a display touch device. The driving method is applied to the gate drive unit in the display touch device. The gate drive unit includes at least one pull-down circuit. The pull-down circuit is coupled with a pull-up node and a pull-down voltage terminal, and is configured to control connection or disconnection between the pull-up node and the pull-down voltage terminal; one frame of image display time includes a display phase and a touch phase that are alternately set, and at least one of the touch phase is set between two adjacent display phases. The driving method includes: during the touch phase, controlling, by the at least one pull-down circuit, the disconnection between the pull-up node and the pull-down voltage terminal. The problem of a screen flashing phenomenon of existing display touch devices during the touch phase is solved.