Recently, the 2022 Medical Artificial Intelligence Conference (CMAI 2022) and the second "China Medical Academic Journal Development" high-end forum were held.
Leifeng.com "Medical and Health AI Gold Nuggets" is a support unit of this conference. Qi Honggang, professor of the School of Computer Science, University of Chinese Academy of Sciences, and Wang Liqun, professor of the School of Science, China University of Petroleum (Beijing) served as the conference host.
This summit forum invited many directors of radiology departments and authoritative experts in artificial intelligence technology from top hospitals to discuss the clinical application and scientific research progress of artificial intelligence technology in medical imaging , and share research experience.
(We will launch a detailed in-depth dialogue and speech content of all speakers in the future, welcome to follow)
Academician of the Chinese Academy of Sciences Huang Wei
Huang Wei, an academician of the Chinese Academy of Sciences and editor-in-chief of Research, delivered a speech on behalf of the CMAI conference. He said that the powerful empowerment role of artificial intelligence on various industries has been revealed. Biomedicine is a data-intensive, brain-intensive, and knowledge-intensive industry that requires relying on strong analytical and processing capabilities for judgment and diagnosis and treatment. It is a very promising field of artificial intelligence application.
htmlFor more than 0 years, my country has intensively introduced a series of policies and regulations on medical artificial intelligence, aiming to establish a fast and accurate intelligent medical system. In the "14th Five-Year Plan", both artificial intelligence and life and health are listed as priority levels in the fields of cutting-edge science and technology, which will surely accelerate the rapid development of my country's artificial intelligence and life and health science. "Although medical artificial intelligence has entered a period of rapid development, it still faces many challenges. We sincerely hope that this meeting will become an opportunity for everyone to collide with ideas and deepen exchanges, and expand future collaboration and cooperation in artificial intelligence and biomedical industries."
Chairman of the Radiology Branch of the Chinese Medical Association Liu Shiyuan
Chairman of the Radiology Branch of the Chinese Medical Association and Director of the Department of Imaging Medicine and Nuclear Medicine of Shanghai Changzheng Hospital, as the first guest of the speech, shared the latest basic situation of industry development based on the upcoming "China Medical Imaging Artificial Intelligence Development Report (2021-2022)".
Liu Shiyuan said that as of May 31 this year, there are more than 30 medical imaging AI registration certificates approved by National Medical Products Administration , covering CT, magnetic resonance, DR and other devices, including cardiovascular and cerebrovascular, chest diseases, diabetes , osteoarticular diseases and children's development assessment.
"We have gone through a calm and rational stage for AI products, from our initial suspicion to overly optimistic longing, and now we have entered a new era of clinical application and commercialization." The approved medical imaging AI products of
can be divided into two aspects: one is a product that optimizes the medical imaging workflow, and the other is a disease-centered diagnostic model.
Specifically for the former, it has become the norm for medical imaging equipment to empower through deep learning technology. It is expected that by around 2023, the penetration rate of AI to CT, will increase to about 50%, and the penetration rate of MRI and ultrasound will increase to about 40%.
The latter is the field where the company invests a lot of energy in research and development, and the most mature products are pulmonary nodules and coronary CTA.
As of 2022, AI products based on disease models have gradually iterated from lesions detection and segmentation to multi-dimensional, multi-functional, and even multi-task models that combine morphological diagnosis and functional diagnosis, forming a platform-based application centered on disease scenarios.
Professor Liu Shiyuan introduced that medical imaging AI products are gradually introduced into hospitals. This year, the AI penetration rate of large hospitals in my country is about 15%, and it is expected to reach more than 30% next year.
It is worth noting that although more than 50% of the AI products used in hospitals are obtained through purchase, more than 94% of patients are free trials. Judging from the remaining 5% charging cases, the main methods of charging include diagnosis, consultation, examination and packaging charges, which have not yet been used as separate charging items.
"This shows that AI products are not mature enough to give patients a strong desire to buy, and the product form and commercialization model of AI still need to be continuously improved.”
Director of Radiology Department of Beijing You'an Hospital Li Hongjun
radiology department, Beijing You'an Hospital, shared the theme of "The Role and Value of Medical Imaging in Medical Big Data".
Li Hongjun introduced the development of medical imaging , the connotation of medical imaging data, the value of medical imaging data, the mining and utilization of medical imaging big data.
Li Hongjun said that in recent years, the application of AI and the upgrade of AI algorithms have driven the entire medical imaging to enter a new stage, including the selection, segmentation and image processing of interest areas have reduced the interference factors of images, and the efficiency of algorithms has been improved, making the results more accurate.
Taking the extraction of coronary artery tree as an example, AI's understanding of global structure, compensation of effective information, and repair of weak signal fractures can achieve the most effective generation, actively remove and repair artifacts, and display the morphology of the entire coronary artery image in a comprehensive and three-dimensional manner.
Li Hongjun believes that the occurrence and development of each disease is not a single data change, but a multiomic change. "Our imagingomics must be integrated with clinical data characteristics, proteomics , genomics, metabolomics , sociomics and other multi-data models, so as to fully and objectively reflect the occurrence, development and prognosis evaluation of individual diseases. "
This also means that the traditional morphological imaging diagnosis model can no longer meet the requirements of precision medicine.
Li Hongjun said that the early AI only warns and predicts diseases based on images and data characteristics, which deviates from the significance of biology. The combination of imaging genomics and AI will be the development and extension of morphological imaging , which can solve diseases that are invisible to the naked eye and achieve the diagnosis of diseases without symptoms and signs.
Professor of Zhejiang University, Changjiang scholar Wu Jian
Professor of Zhejiang University and Changjiang scholar Wu Jian shared the market background, industry status, bottlenecks and difficulties, solutions and stage results of AI ECG diagnosis with the theme of "Artificial Intelligence Electrocardiogram Diagnosis".
Wu Jian introduced that the application demand for electrocardiogram examination in my country is very large, with at least 250 million people undergoing electrocardiogram examinations every year, but they face electrocardiogram Figures and cardiovascular doctors are insufficient and the accuracy of equipment detection is low.
Wu Jian's project research mainly focuses on four major goals, namely, using AI algorithms combined with signal processing methods to perform automatic electrocardiogram analysis, establish an electrocardiogram monitoring model, establish a cardiovascular disease discrimination model, provide doctors' electrocardiogram marking work tools and establish an auxiliary diagnosis platform.
explores AI electrocardiogram assisted diagnosis, Wu Jiantuan The team found that there were bottlenecks in five major aspects: data collection, data cleaning, data annotation, heart beat recognition and model establishment.
In response to these bottlenecks, the algorithm framework developed by Wu Jian’s team has innovative points such as convolution feature description global information, frequency domain analysis feature supplement detailed information, fast and accurate, batch computing, etc.
At present, Wu Jian’s team has obtained more than 2 million ECG data, More than 100 categories of tags were processed, covering 99% of the ECG diagnostic categories. The AI ECG assisted diagnostic platform trained based on these data supports the recognition of the highest 55 types of diagnostic tags, with an overall accuracy rate of 95%, and F1 reached 91%.
In addition, the team has successfully developed an ECG band labeling tool and an intelligent ECG assisted diagnostic system.
Director of the Department of Radiology, Tangdu Hospital, Air Force Medical University Cui Guangbin, director of the Department of Radiology, Tangdu Hospital, Air Force Medical University, shared the theme of "The Current Situation, Challenges and Prospects of AI in Lung Nodules".
Cui Guangbin said that lung cancer is the type of cancer with the most new cases and deaths in my country. In order to reduce the incidence and mortality of cancer, the "Healthy China Action" requires the promotion of early screening, early diagnosis and early treatment of cancer. The configuration and popularization of
CT has enabled my country to have basic hardware conditions for CT lung cancer screening, but there are insufficient imaging doctors in public hospitals of different levels, and lack of experience in reading and diagnosis. Medical imaging AI is an important way to solve these pain points.
With the implementation of medical imaging AI in hospitals, Cui Guangbin found that AI products are somewhat out of touch with the actual clinical application needs. "Artificial intelligence rushed forward, which was very lively, but avoiding the important things. I encountered some practical problems in the work process, but AI companies could not fully meet them."
Take the screening of the new crown CT as an example. Where AI focuses on improvements, it is also a work that doctors can complete with the naked eye. For example, the scope of lesions is not very meaningful. The bedside films taken by X-ray machine for critical cases are overlapping images, and manual inspection will cause many uncertain factors to hinder the diagnosis. This is the field where AI plays its role, but it has not yet been resolved.
Professor and Yangtze River Scholar Peng Shaoliang
Professor and Yangtze River Scholar Peng Shaoliang shared the theme of "Supercomputing-based Digital Therapy and Electronic Medicine".
Peng Shaoliang gave a detailed introduction to the role of digital therapy, as well as multiple digital therapy cases around the world.
Peng Shaoliang believes that digital therapy has many advantages, which can accelerate the effect of treatment, shorten the treatment cycle, and reduce the cost of treatment. Compared with traditional chemical drugs, the research and development of digital therapy is faster.
"The research and development of a new drug takes more than 5 to 10 years, with a minimum expenditure of 1 billion US dollars. On the contrary, digital therapy is a software that does not require such a long time and such a large expenditure. In the future, we only need to verify the effectiveness of data and algorithms."
Peng Shaoliang said that my country's exploration in the fields of metacosmic medical and digital therapy is still blank. He hopes that hospitals, medical companies, IT game companies, etc. will jointly establish the first metacosmic medical and digital therapy alliance in China, focusing on a series of diseases with less international layout such as adolescent depression, Alzheimer's disease , and launching the first domestic digital prescription standard and electronic drugs.
Deputy Dean of Beijing Xuanwu Hospital Lu Jie
Professor Lu Jie is the deputy Dean of Xuanwu Hospital of Capital Medical University and the director of the Department of Radiation and Nuclear Medicine. At the meeting, he shared the topic "Research on the Application of Artificial Intelligence in MRI Imaging of Brain Demyelination Disease".
Lu Jie introduced that multiple sclerosis (MS) and neuromyelitis optic (NMOSD) are common brain demyelinating diseases, and are also common diseases of nervous system that are disabled in middle-aged and young people. There are about 2.8 million patients with multiple sclerosis worldwide, and about 42,000 in China.
Due to similar clinical symptoms, partial overlap in laboratory test results, and long diagnosis of diagnosis cycles, the differential diagnosis of MS and NMOSD is very challenging, especially for primary hospital and low-level physicians.
In clinical practice, nuclear magnetic resonance imaging (MRI) evaluation is an important part of the diagnosis of MS and NMOSD. With the development of artificial intelligence technology in recent years, research on the application of artificial intelligence in MRI imaging of brain demyelination disease has also made great progress.
Lu Jie pointed out in the report that artificial intelligence technology can explore high-level quantitative features that cannot be recognized by the naked eye in image images. The artificial intelligence model based on topology will have important value in predicting the prognosis of brain demyelination disease.
Professor of Beijing University of Posts and Telecommunications Liu Yong
Professor Liu Yong is a professor at the School of Artificial Intelligence of Beijing University of Posts and Telecommunications. He shared the topic "Research on Imaging Omics Characterization of Alzheimer's Disease Based on Magnetic Resonance and PET Imaging".
He said that the machine learning method has been widely carried out on medical images, and researchers have been working hard to explore objective quantitative, repeatable and biologically significant imaging markers of neuropsychiatric diseases.
Liu Yong's research team has been conducting research on how to portray the abnormal representation of brain imaging of Alzheimer's disease (AD) for more than ten years, and explored the feasibility of using magnetic resonance images to study early AD image markers.
"We cannot change our age, family history and genetic gene . One of the things researchers can do is to discover clues as early as possible and provide a little help for early recognition of AD." Liu Yong pointed out at the end of the report, "If we do this, we may be able to bring some benefits to more patients and families."”
Professor of Nanjing University of Aeronautics and Astronautics Zhang Daoqiang
Professor Zhang Daoqiang is a professor at Nanjing University of Aeronautics and Astronautics. He shared the topic "Research Progress in Brain Imaging Intelligent Computing and Several Applications".
Zhang Daoqiang introduced that the best intervention stage of Alzheimer's disease is in the incubation period before symptoms appear and mild cognitive dysfunction stage. Once patients enter the stage of dementia, they will no longer be able to undergo effective treatment. Therefore, early detection and early intervention are particularly important.
Zhang Daoqiang's research is based on brain images to build brain networks, and to mine, analyze and classify the brain network, thereby realizing the technology of diagnosis of Alzheimer's disease.
Among them, the representative "brain connectomics" refers to the discipline that uses multimodal neuroimaging technology and network analysis methods to describe the structure and functional connection patterns of the living human brain. The connection patterns are mainly divided into three types: structural connection, functional connection and effective connection. During the work process, the brain network is first constructed using brain images, then extracted features from the brain network, and finally the extracted features are classified. In the report of
, Zhang Daoqiang also shared the research progress and achievements of his team in the application of brain network classification, imaging genetics, brain cognition and brain decoding.
Professor of the School of Biomedical Engineering, Shenzhen University Raboying
Professor Raboying is a professor at the School of Biomedical Engineering of Shenzhen University. He shared the topic "Intelligent Diagnosis for Clinical Applications".
Raboying's intelligent diagnosis research mainly focuses on two common brain diseases: Alzheimer's disease and Parkinson's disease . Raboying said that in order to improve the diagnostic accuracy of brain diseases, his team proposed to build a regularized longitudinal analysis model of multiple relationships to improve the accuracy of intelligent diagnosis and assist doctors in clinical diagnosis.
in the study, targeting single Time point data, multi-time point data, and multi-template data have different characteristics. Different core methods are adopted to study, and they are applied to the clinical diagnosis of Alzheimer's disease, mild cognitive impairment and autism respectively.
In addition, Raboying's team also explored deep learning in the early diagnosis of Alzheimer's disease. Using the second-order statistical data of MRI, advanced pooling schemes are included in the classifier, and the GAN network combined with tensor training, advanced pooling and semi-supervised learning is used for diagnosis.
Assistant Professor, Hong Kong University of Science and Technology Li Xiaomeng
Professor Li Xiaomeng is an assistant professor in the Department of Electronics and Computer Engineering of the Hong Kong University of Science and Technology. He shared the topic "Empowering Clinical Decision-making by AI-based Medical Image Analysis".
Li Xiaomeng introduced the research content of medical images classification, segmentation and detection, medical image reconstruction, image prediction and other team research through efficient annotation, as well as generalization research using models, such as protecting hospital data privacy through federal learning research.
In view of the situation where there are a large amount of unlabeled data and labeled data at the same time in the training dataset, Li Xiaomeng and his team proposed a semi-supervised learning method to segment medical images based on the self-integrated model of rotation consistency.
In addition, since weakly supervised learning can obtain pixel-level segmentation results in image segmentation , it also has very important application scenarios in medical images, such as gland segmentation in case images. In the sharing, Li Xiaomeng introduced the team's method of weakly supervised learning on natural images.
"We discovered how to use the existing deep learning models on natural images to make them play a greater role in medical images." Li Xiaomeng said.
Associate Researcher at the Institute of Computing of Chinese Academy of Sciences Associate Researcher at the Institute of Computing of Chinese Academy of Sciences Zhao Di
Professor Zhao Di is an associate researcher at the Institute of Computing of Chinese Academy of Sciences and shared it with the title "Neuromorphic Computing for Medical Imaging Analysis".
Zhao Di introduced that in the sixty years since the emergence of artificial intelligence, it has experienced two ups and downs, and now it has entered the eve of its true explosion.
is an important part of the development stage of deep learning. convolutional neural network (CNN) image recognition is one of the main means of medical image analysis. However, as the model becomes larger and larger, the parameter scale continues to grow, new technical means are gradually entering people's attention.
The third generation neural network pulse neural network (SNN) is also called internal brain computing or neuromorphic computing. Compared with CNN, the power consumption of SNN is reduced by orders of magnitude. Zhao Di believes that SNN is a possible direction for the future development of artificial intelligence.
Therefore, pulse convolutional neural network (SCN) that fuses SNN and CNN has great development potential. When the accuracy of classification and object detection segmentation is close, the energy consumption of SCN is much lower than that of CNN.
Zhao Di said that the development of internal brain computing will have a great promoting effect on research in the field of medical and health.
Leifeng.com Leifeng.com
