Not long ago, the book "undertaking the Mission of Innovation-- ParknShop demonstration Enterprise of Industry-University-Research Cooperation in China", organized and compiled by the China Association for the Promotion of Industry-University-Research Cooperation, was released in Beijing. Jukun Science and Technology was successfully selected into the ParknShop demonstration Enterprise of Industry-University-Research Cooperation in China. The release of the book has been highly valued by more than 100 industry-university-research collaborative innovation platforms and relevant departments belonging to the State-owned assets Supervision and Administration Commission of the State Council, the China Association of Science and Technology and the China Association for the Promotion of Industry-University-Research Cooperation, and has been paid close attention to by the majority of enterprises and industry-university-research circles.
Jukun Technology has independently developed the leading medical AI neural network. Centering on the core technology strategy of "digital human body", it has successfully launched a portfolio of digital doctors, such as "digital heart", "digital brain", "digital chest", "digital abdomen" and "digital muscle bone". It provides intelligent screening, accurate diagnosis and treatment for heart disease, stroke, cancer and other most critical and chronic diseases of mankind.
Today, digital doctor platform products are used in more than 1200 hospitals and are used by clinicians with high viscosity on a daily basis, and many patients may have been treated by "digital doctors". As a clinical tool loved by clinicians, its R & D enterprise has also been listed as a demonstration enterprise of innovation by the China Association for the Promotion of Industry-University-Research Cooperation, so, driven by artificial intelligence technology, how did the digital doctor platform come into being? how does it grow into a cross-era change in the medical field?
With the combination of medical workers and collective wisdom to create the first global heartAI
In 2017, Jukun Technology was founded and began to develop its first product. After in-depth interviews with imaging experts, the company finally decided to start with coronary artery CTA. Coronary artery CTA is positioned as the "gatekeeper" of stable coronary heart disease imaging examination, and there are a large number of clinical examination needs. In the imaging department, it usually takes more than half an hour to process a coronary CTA. Tedious image post-processing and accurate evaluation of each vessel make the reporting time of coronary CTA examination too long, which directly leads to the limited number of patients who are admitted for examination every day, and it usually takes a long time to make an appointment for examination.
The cardiovascular difficulty lies in the lack of public data. First of all, it is difficult to label the data: irregular arteries and veins, small blood vessels, and difficult labeling, and it often takes doctors two or three hours to label the coronary arteries alone. In addition, from the construction of the data set to the artificial intelligence of the whole diagnosis process, including the whole process, cardiovascular needs to do a whole set of things, which involves dozens of algorithms, including both traditional algorithms and deep learning algorithms. Relative to pulmonary nodules, not on an order of magnitude, it is a process from a focus to a disease. After 4 months of polishing, the first model was made.
In December 2017, in order to further verify and iterate the products, Jikun Technology began to cooperate with Beijing Friendship Hospital. And Professor Yang Zhenghan's team continue to polish iterative products. After the trial run of the product prototype in Friendship Hospital, it received feedback from the doctor soon. Doctors hope to identify not only the major vascular branches in the image, but also the 1.5 mm terminal vessels. This requirement is a great challenge to the technology. In order to overcome this problem, Jukun's whole team moved to Friendship Hospital to work, and formed a team of doctors and workers with experts from Friendship Hospital. Study the effect of the algorithm every day, and iterate the algorithm scheme in time after getting the doctor's feedback.
Several Kun science and technology overcame the challenge, before and after the Beginning of Spring, ushered in a change for the better. The cooperation and verification of more than a dozen hospitals, such as Anzhen Hospital, the first affiliated Hospital of Zhengzhou University, Wuhan Central Hospital and so on, make the ability of the model to a higher level. It not only improves the generalization of the model for different types of data, but also absorbs doctors' suggestions on products. Every week, doctors will be invited to give lectures on some basic knowledge and diagnosis of coronary artery. At the same time, it also provides a platform for doctors to demonstrate and use, and doctors provide more and more iterative suggestions.
Compared with the widely used initial image AI products, hundreds of coronary artery CTA images often require higher algorithms, which are more difficult to identify and more time-consuming. However, these difficulties are the strong points of artificial intelligence. For this reason, Jukun Science and Technology cooperated with different hospitals to build a deep neural network model which can deal with images under different equipment and different scanning standards, and based on this, a comprehensive Dukun cardiovascular AI system was built. It can be applied to the vast majority of coronary CTA images to complete standardized and refined vascular reconstruction, and give corresponding diagnostic reports at the same time.
Constantly upgrade and perfect the technology platform of digital human body
When the first-generation AI product of Jukun Technology was released, the AI could only measure the degree of vascular stenosis and carry out corresponding post-processing. In the second generation of AI, functional evaluation such as calcification score and plaque analysis are gradually added, and Sikun technology can give doctors more comprehensive data reference. The next third generation of AI, Jikun Technology tries to extend the role of AI to surgical navigation and consumable selection, completing the leap of artificial intelligence from "diagnosis" to "diagnosis and treatment".
On the one hand, Jukun Science and Technology began to develop a pneumonia diagnosis system to meet the needs of COVID-19 's epidemic situation, on the other hand, its research and development direction expanded from the original cardiovascular to cardio-cerebrovascular, hoping to really solve the problem of stroke screening that plagued Chinese residents.
Sikun Technology then developed the head and neck CTA intelligent diagnosis system, CT cerebral perfusion intelligent analysis system, through scene expansion to explore a more accurate diagnosis. So far, in addition to the existing intelligent blood flow reserve fraction CT FFR measurement system and aortic intelligent assistant diagnosis system, Jukun Technology has developed an intelligent assistant diagnosis system for head and neck CT, which has shifted the scope of AI from cardiovascular to systemic vessels.
In October 2020, the head and neck CTA reconstruction research conducted by Professor Lu Jie of Xuanwu Hospital of Capital Medical University and Jukun Science and Technology was included in the Nature sub-journal. In this study, 18259 head and neck CTA data sets were used to delineate and segment more than 100 million vascular areas by AI. The results show that the qualified rate of reconstructed image quality of AI is 92.1%. CerebralDoc successfully reduced the processing time from 14.22 ±3.64min to 4.94 ±0.36min, reduced the time by about 65.3%, reduced the number of clicks from 115.87 ±25.9to 4, and reduced the number of doctors needed from 3 to 1 after 5 months of use.
In 2021, Jikun Science and Technology launched liver MR artificial intelligence products, which were jointly developed by Professor Yang Zhenghan, director of the Department of Radiology of Beijing Friendship Hospital, and other R & D personnel, with the aim of solving complex liver disease diagnosis problems. Professor Yang Zhenghan pointed out, "globally, from the published papers, there are only a few papers related to liver MRI+AI, and all of them are single-task research." The liver MRI AI system is an overall diagnostic scheme, which is expected to make a breakthrough in the intelligent diagnosis of focal liver lesions with MRI. "
The scientific research achievements of Jukun science and technology are not limited to the extension of modes, but also have made great progress in the abundance of products.
The regional intelligent medical image cloud platform is an important product of Jukun science and technology to promote the reform of China's active preventive public health prevention and control system. For regional residents, there is no need for the first diagnosis of major and minor diseases to go to the central hospital, nearby medical treatment can also enjoy "expert-level medical services."
During the prevention and control of COVID-19 's epidemic situation, Jukun Science and Technology immediately contacted the hospital in Hubei to organize elite research and development force. after more than 10 days of painstaking efforts, COVID-19 intelligent image analysis system was successfully developed. After the completion of the research and development of the system, it was launched in a number of designated treatment hospitals in Hubei, and donated to Leishenshan Hospital. During the critical period of COVID-19 epidemic prevention and control, the Beijing Science and Technology Commission coordinated the Municipal Health Commission and the Municipal Medical Management Center to promote the deployment of COVID-19 AI of Jukun Technology in many designated hospitals, such as Beijing The Temple of Earth Hospital, Beijing you'an Hospital, and Beijing Xiaotangshan Hospital.
Let the digital doctor become the helper of the human doctor.
For engineers and doctors, they are not proficient in each other's industry knowledge, and the research of artificial intelligence algorithms needs the cooperation of two fields of knowledge. To this end, the engineers of Jukun Technology have to go through "hospitalization" before they "start work", so that they can better understand the medical scene and how to maintain better communication with doctors.
Different from traditional equipment, doctors have a stronger sense of participation in the research and development of artificial intelligence, and every slight growth is devoted to the painstaking efforts of Jukun technology and doctors. It is also in this process, the two sides have different growth, in their view, artificial intelligence is not a doctor's assistant, but a colleague in the doctor's work.
In this understanding, the heart of the doctor has risen from the user to the creator, whether in the standard of data, or in the application of the product, all will be strict requirements, so as to make their own recognized products.
In 2020, Jikun Science and Technology launched "Jikun Learning Garden", which provides many young doctors with valuable experience in the understanding and use of artificial intelligence online. It is hoped that through this win-win way, artificial intelligence can better complete clinical application.
The great-leap-forward development of the medical AI industry has found a feasible way to solve the difficulties in the treatment of grass-roots patients and the development of primary health care. Medical imaging AI products can become super assistants of doctors, reduce the burden of clinicians, make up for the shortage of primary medical doctors, and improve the efficiency of medical diagnosis.
In the process of the growth and development of Jukun science and technology, it is everywhere accompanied by the seamless cooperation between the research and development team and doctors. The iteration of the research and development of Sikun science and technology products all stems from the experience and deep learning of doctors and R & D teams in the process of using AI, which also makes Junkun science and technology always be at the forefront of industry-university-research innovation from birth to growth, and constantly work with doctors to create human collective wisdom in the AI era. Let digital doctors always stand at the forefront of industry-university-research medicine.
(responsible Editor: Guo Jiandong)