Shenzhen Institutes of Advanced Technology

Chinese Academy of Sciences

Yongfeng Yang

Xin Liu

Hairong Zheng

Dong Liang

Zhanli Hu

Na Zhang

Xiaohui Wang

Qi Yang

Hanyu Sun

Qingneng Li

Dongfang Gao

Yingjie Xu

San Wu

Zhonghua Kuang

Chunhui Zhang

Qihang Chen

A low-dose PET image restoration method and system, a device, and a medium are provided. The method includes: S1, performing blocking processing on a training image comprising a low-dose PET image, an MR image, and a standard-dose PET image to obtain a first patch, and performing first preprocessing on the first patch to obtain a second patch; S2, obtaining, according to the second patch, a first joint dictionary by means of sparse coding and dictionary updating; and S3, restoring the low-dose PET image to a restored image of the standard-dose PET image according to the first joint dictionary.

Low-dose PET image restoration method and system, device, and medium

A medical image noise reduction method, system, terminal and storage medium are disclosed. The method includes: obtaining a standard-dose PET image and a constant-value image; inputting the standard-dose PET image and the constant-value image into a decay function to obtain a corresponding low-dose noisy PET image and a noisy constant-value image, respectively; assembling the low-dose noisy PET image and the noisy constant-value image in a width dimension or a height dimension, and then inputting into a trained conjugate generative adversarial network, and outputting a denoised PET image and constant-value image output by the conjugate generative adversarial network.

Medical image noise reduction method, system, terminal, and storage medium

Methods and apparatuses for training a magnetic resonance imaging model, electronic devices and computer readable storage media are provided. A method may include: acquiring a magnetic resonance image data set; constructing a ring deep neural network to be trained; inputting an under-sampled magnetic resonance image and a full-sampled magnetic resonance image respectively to two neural networks included in the ring deep neural network, to generate respective simulated magnetic resonance images; inputting a first simulated full-sampled magnetic resonance image and the full-sampled magnetic resonance image to a pre-constructed first simulated magnetic resonance image class discrimination model, to obtain a first discrimination result indicating whether or not the first simulated full-sampled magnetic resonance image is of a simulated magnetic resonance image class; and adjusting a network parameter of the ring deep neural network based on a preset loss function, to obtain a trained magnetic resonance imaging model.

Methods and apparatuses for training magnetic resonance imaging model

A method for processing positron emission tomography data is provided, this method includes: obtaining a first coordinate and a second coordinate respectively corresponding to two ends of a response line to be processed; determining corresponding dimensional coordinates of the response line to be processed in a sinogram based on the first coordinate and the second coordinate; and generating the sinogram corresponding to the response line to be processed based on the dimensional coordinates. According to this method, the amount of calculation of system matrix is reduced, the accuracy of position information of the generated response line is improved, and the accuracy of generated sinogram is improved accordingly.

Method and terminal device for processing Positron Emission Tomography data

A method, device and equipment for reconstructing a PET image are provided. The method includes acquiring a prior image comprising an anatomical image and an autocorrelation feature image, the autocorrelation feature image being determined based on gray-level co-occurrence matrix of the anatomical image; and acquiring a feature value of the prior image; reconstructing the PET image according to the feature value and an iterative algorithm.

Method, device and equipment for reconstructing PET images

An apriori guidance network for multitask medical image synthesis is provided. The apriori guidance network includes a generator and a discriminator, wherein the generator includes an apriori guidance module configured to convert an input feature map into a target modal image pointing to a target domain according to an apriori feature, and the apriori feature is a deep feature of the target modal image. The generator is configured to generate a corresponding target domain image by taking the apriori feature of the target modal image and source modal image data as an input. The discriminator is configured to discriminate an authenticity of the target domain image outputted by the generator.

Apriori guidance network for multitask medical image synthesis

The present application relates to a method and system for generating multi-task learning-type generative adversarial network for low-dose PET reconstruction, and relates to the field of deep learning. The method includes connecting layers of the encoder with layers of the decoder by skip connection to provide a U-Net type picture generator; generating a group of generative adversarial networks by matching a plurality of picture generators with a plurality of discriminators in one-to-one manner; obtaining a first multi-task learning-type generative adversarial network; designing a joint loss functionfor improving image quality; and training the first multi-task learning-type generative adversarial network according to the joint loss functionin combination with an optimizer to provide a second multi-task learning-type generative adversarial network.

Method and system for generating multi-task learning-type generative adversarial network for low-dose PET reconstruction

A method of processing plaques in magnetic resonance imaging of vessel wall include: step Straining a generative adversarial network and a capsule neural network to obtain a trained generator network and a trained capsule neural network; and step Scascade-connecting the trained generator network with the capsule neural network into a system to recognize and classify plaques in magnetic resonance imaging of vessel wall. In one aspect, the capsule neural network has more abundant vascular plaques characteristic information represented by vector; in another aspect, when the trained generator network and the capsule neural network are cascaded into the system to recognize and classify the plaques in magnetic resonance imaging of vessel wall, an accuracy of recognition and classification may be greatly improved. A device for processing the method as well as a computer for implementing are also disclosed.

Growing community of inventors

Shenzhen, China

Yongfeng Yang