Image remote diagnosis ushered in the 5G era.
80% of medical data is from source medical imaging. Now a patient’s CT scan image can easily reach several hundred megabytes. The most advanced thin-layer scan CT can reach 1G in a patient case. However, such a large-capacity image file will have a significant delay even when transmitted over a hospital wired network; At this stage, with 5G technology, it is possible to achieve fast transmission and real-time analysis of image data, which is also Regional imaging diagnostic collaboration and medical imaging AI bring new opportunities for development.
For this reason, many companies have started commercial exploration in this field, and recently encountered “wings Exhibition Technology (hereinafter referred to as “Wingspan”) is one of them. Yesterday, at the Chinese Medical Association’s 26th National Congress of Radiology (CCR 2019), the company released a new generation of wingspan AI+5G box ( 3.0) (hereinafter referred to as: the wingspan box), it will be possible to use 5G technology to transmit medical images and achieve clinical application.
5G adds new variables to AI medical imaging applications
According to Wingsong CMO Gao Yunlong, Wingspan Box isEdge Computing Smart Gateway Tool with 5G Communication Module It can realize the transmission of data from different areas and a large number of medical imaging equipment to the cloud, and provide real-time quality control, remote monitoring, remote maintenance and fault warning functions, which is convenient for building a large-scale remote medical imaging diagnostic platform.
Wingpan Box
Edge computing means bringing the resources, computing, storage and other capabilities of cloud computing closer to the user’s local edge devices, processing large amounts of data locally and making decisions. For the “AI+Medical Imaging” field, the marginalization of the AI model can significantly reduce the cost of data communication; by processing data locally, it can also avoid large amounts of data.The problem of storing in the cloud. In addition, the introduction of machine learning intelligence into edge computing devices can also meet the growing demand for privacy/confidentiality and low latency as explosive devices grow.
In actual clinical applications, with 5G blessing, experienced experts can provide remote reading, operation, and even diagnosis for primary hospitals in a short period of time, thus effectively promoting graded diagnosis and treatment, benefiting remote areas. The masses.
Gao Yunlong said that the quality of image capture will directly affect the reading effect. However, for a long time, due to the shortage of medical resources and the lack of experience of technicians in township hospitals, the quality of the captured images cannot be guaranteed. It is very troublesome for the doctors at the higher level.
With the wingspan box, the base layer only needs to send the captured image to the wingspan box, and the AI will judge the image quality. If the result is non-A piece, the technician will be directly reminded to shoot again. Compared with the case of no edge calculation, the image has been transmitted to the cloud and processed. The reading doctor finds that the image quality does not meet the diagnostic requirements and then calls back. The feedback in the calculation of the edge of the wingspan box is more timely, saving communication and duplication. The time cost of transmission speeds up the process of remote image diagnosis.
Apply blockchain technology to image transmission cloud platform
In addition to the exploration of 5G applications, Wings Technology has been exploring for a long time in “focusing on improving imaging diagnostic capabilities.”
Established in 2009, Yizhan Technology is a high-tech enterprise that provides intelligent medical imaging solutions and focuses on the development of medical imaging artificial intelligence applications. After 10 years of development, the business map of Wingspan has developed into a remote imaging diagnostics, imaging medical association, imaging doctor group, AI open collaboration platform, medical informationization, medical imaging diagnosis center. The “AI+ Internet” cloud empowerment center.
Wingpan Cloud Shadow
According to Gao Yunlong introduced that the company’s core product is “Cloud Shadow”, which was launched in 2013 and is targeted by “Data + Cloud” The construction of the way of technology + scene”A comprehensive solution for remote diagnosis, remote consultation, distance education, and telemedicine operations, including multiple functions, for multiple application scenarios, as follows:
-
Cloud 3D reconstruction, support 3D MPR, MIP, MinIP, AIP, support layer thickness adjustment and arbitrary angle rotation, provide remote reading environment for video doctors;
-
PET-CT three-dimensional fusion function, showing PET-CT fusion images in 3D view of MPR and MIP, helping doctors to observe tissue changes and accurately locate lesions. Currently, three new pseudo-color schemes have been added;
-
The remote ultrasound consultation function can be used for communication between different offices or different hospitals in the hospital, as well as remote ultrasound solutions for regional medical internet mode.
It is understood that it is mainly to provide professional online reading and remote diagnosis reporting tools for radiologists, that is, the traditional medical imaging islands are connected through the mobile Internet, and the devices and doctors can effectively and quickly connect to improve Accuracy and diagnostic speed. It is reported that the platform can also support online reading, operation and diagnosis of CTA angiography, mammography, fundus images, endoscopy and other images.
After iteration and innovation, Wingspan has launched a remote imaging diagnostic platform integrating nuclear medicine and ultrasound, and a regional intelligent cloud imaging platform based on blockchain technology to ensure that the image report cannot be tampered with. Traceability, in order to maximize patient privacy and data security, accelerate the circulation and exchange of medical image data, improve the efficiency of medical image remote diagnosis, promote the collection and labeling of medical image artificial intelligence training data, and finally expand the distribution The application of medical image artificial intelligence.
Build an open platform and seek AI’s breakout path
Unlike most medical AI companies, the purpose of the wingspan is to ecologically lay out medical images, as evidenced by its current business landscape.
To this end, Wingspan has established an open collaboration platform for medical image artificial intelligence to integrate application tools for different subdivision scenarios and establish a comprehensive platform. Gao Yunlong said that the AIDiscovery platform can be understood as “the App Store in the medical imaging industry”, which provides a platform for application and product commercialization for the global AI enterprise and developer team, and establishes a medical industry ecosystem. In order to optimize the medical imaging diagnosis, it can also cover more medical institutions to help doctors make better clinical judgments.
Open platform for wingspan
As of now, AIDiscovery has cooperated/online applications including computer-generated DR chest radiograph automated diagnostic reporting system, lung disease thermal recognition, CT lung nodule automatic labeling and analysis, and intelligent text correction assistant. Real-time quality control of chest X-ray, diagnostic evaluation of magnetic resonance vascular plaque, and automatic detection of CT cerebral hemorrhage in the brain.
It is worth mentioning that, in addition to building a platform, Wingspan is also independently developing some applications that are urgently needed in the market. For example, Wings has independently developed the Wingspan Rogen Radial DR Chest Automated Reporting System, which has been iterated to version 2.0.
According to Gao Yunlong, The new version of the automated reporting system uses an updated deep learning model, adding nearly 100,000 DR chest radiographs and corresponding reports as training data, based on the original 17 diseases. Increase the identification of rib fractures, the overall model performance is significantly improved, only 0.8 seconds to complete the entire diagnostic report issuance process.
Feedback results show that the AI generated report compares with the doctor’s report, “the overall performance, the readability of the report, the ability to find important information in the image, the ability to describe the lesion and the abnormal site, and the number of lesions. More than 75% of the indicators such as ability, description and abnormal performance ability are comparable to doctors’ diagnosis and even better than doctors.
In the commercial exploration, Gao Yunlong also revealed that the current products or applications of the wingspan have covered nearly 3,000 hospitals nationwide; in addition, the global layout is also underway. Understand that in September this year, Wingspan and Ningxia Weijian Committee and Thailand-Middle East Alliance Economic and Trade Promotion Association jointly signed the “Technical Service Export and “Internet + Medical Health” Model Strategic Framework Agreement”, which will be further Collaborate with the Belt and Road and Southeast Asian countries to export the wingspan medical service capabilities.