Direction 1: Mathematical Theories and Methods of Medical Image Analysis and Intelligent Diagnosis
The rapid advancement of S&T and image processing techniques enables the development of medical ultrasound imaging. Ultrasound image analysis has gradually become one of the most common methods for the diagnosis of lesions (nodules or tumors) in organs or tissue. Diagnostic imaging greatly depends on ultrasound imaging which has many benefits, including real-time imaging, convenience, non-invasiveness, and affordability. The accuracy of diagnostic results depends heavily on physicians' experience when the ultrasound practitioner acquires ultrasound cross-sectional images of lesions in real-time and makes preliminary diagnoses of benignity and malignancy based on imaging features, which is not conducive to achieving homogenization of disease diagnosis. The poor signal-to-noise ratio, low contrast, artifacts, and blurry or even missing edges of ultrasound images make it more challenging to automatically detect, segment, and diagnose benign and malignant lesion locations. Novel deep learning algorithms are expected to be built by including rotational invariance and Split Dropout.
Direction 2: Intelligent Robotic Ultrasound
We aim to develop a system for intelligent ultrasound diagnosis that can quickly identify nodules in ultrasound images, automatically mark their precise locations, accurately detect their boundaries, and measure their sizes. The system will also be able to determine whether a nodule is benign or malignant and generate ultrasound reports automatically. Through learning a large amount of image data, the knowledge and expertise of skilled doctors can be transferred to doctors in health professional shortage areas to help in ultrasound diagnosis, easing the challenge of long training time and shortage of doctors. Through learning a large amount of image data, the diagnostic accuracy will be improved by doing objective image feature analysis to prevent being affected by fatigue or other subjective factors. The system is easy to operate and can significantly reduce repetitive work, freeing up doctors to focus on innovative research and clinical skills perfection that will enhance their diagnosis and treatment skills. Furthermore, cloud services enable effective technology implementation in community hospitals and hospitals in rural locations, which will enhance the precision of ultrasound diagnosis.
Direction 3: Key Technologies for Precision Surgery and Robotic Surgery
Surgical navigation technology is a key technology in modern medicine, such as surgical procedures and minimally invasive surgery. It has a direct bearing on the effectiveness of surgery. Intelligent surgical navigation technology based on big data theory and artificial intelligence algorithms, especially multi-modal medical image-guided intelligent surgical navigation technology, has become a research hotspot. Developed countries have invested a lot of human, material, and financial resources in the development of related technologies and equipment. By reducing medical errors and preventing the overuse of healthcare resources, intelligent surgical navigation technology is crucial for benefiting patient care and improving China’s healthcare quality.
[Team member of research program: Kong Dexing, Wu Fa, Xu Lei, Yuan Jing, Huang Shoujun, Xing Yuxin, He Chunlei, Huang Heng]
Direction 4: medical-image-based precision diagnosis and treatment optimization algorithms
The rapid advancement in modern imaging techniques offers reliable technical support in precise organ and tissue lesion diagnosis, precise surgical navigation and quantitative evaluation of surgical outcomes. The three main issues with medical-image-based precision diagnosis and treatment are reconstruction, segmentation, and alignment of medical images. The efficient resolution of these issues can be attributed to the successful resolution of mathematical optimization problems. Thus, mathematical analysis of optimization models and efficient optimization algorithm design are key to solving related problems in medical imaging applications. Unlike the conventional computer image and vision computation problems, medical image analysis calls for accurate, quick and reliable computational results. Also, a large volume of data to be processed poses great challenges for optimization algorithms, while at the same time inspiring new research topics in related disciplines.
Direction 5: medical-image-based surgical navigation theories and algorithms
Medical-image-based surgical navigation is the technical foundation of contemporary precision minimally invasive treatment. The successful application of surgical navigation technology calls for the resolution of complex technical issues, such as preoperative medical image fusion, intraoperative real-time lesion localization and enhanced display, preoperative needle insertion path planning and robot-assisted surgery, as well as the design and implementation of related core algorithms. China is currently behind other countries in the development of medical-image-based surgical navigation research, where skilled doctors and researchers are still in shortage. The research breakthroughs and advances in surgical navigation technology can greatly reduce the workload of doctors, increase hospital efficiency, and effectively meet public demand for modern clinical precision diagnosis and treatment technology.
Direction 6: mathematical modeling and optimization of artificial intelligence in medicine
Optimization theory and statistical data science, as two fundamental disciplines of contemporary artificial intelligence, are important methods of medical AI research. Data science builds pertinent data-driven mathematical models for medical AI, while optimization theory provides powerful tools for an efficient solution to the proposed mathematical models. It is of profound significance to conduct an in-depth investigation into the two fields so as to strengthen the theoretical underpinnings of contemporary medical AI, direct the specific applications of medical AI in precision diagnosis & treatment and surgical navigation, and perfect medical research in China.
[Team member of research program: Yuan Jing, Luo Shousheng, Ge Xinyang, Zhang Jianfeng, Amina Benabid]
Direction 7: Bioelectrical Impedance Medical Imaging Intelligent Algorithm and Equipment
Medical imaging is one of the largest sub-sectors of the medical device industry in China. However, due to a lack of technological competitiveness, China is still heavily dependent on imports for its imaging equipment. The inability to develop key components necessitates technology and product innovation capabilities. A paradigm shift is required in the research and development of novel medical devices, which has traditionally relied heavily on copying. In response to serious diseases like chronic obstructive pulmonary disease and breast tumor caused by the aging population in China, we are motivated to develop autonomous intelligent bio-electrical impedance imaging for long-term dynamic monitoring of disease, to develop high-precision real-time imaging platforms and investigate data acquisition methods, to design and develop new medical imaging devices for the treatment of chronic obstructive pulmonary disease and tumor that can quickly and easily perform dynamic monitoring and assessment as well as predict rehabilitation success and risk.
Direction 8: Intelligent Biosensing and Diagnosis Technology
By using cutting-edge technologies such as bioelectrical impedance, ultrasound spectrum and millimeter wave therapy, we are motivated to investigate new ideas and methods for detecting, extracting, and analyzing effective excitation/physiological signals of tissues. We aim to develop portable diagnosis and treatment technologies and equipment for bio-sensing and physical rehabilitation, targeting serious medical conditions including thrombosis and tumors. We aim to develop POCT (point-of-care testing) diagnostic systems with China’s own intellectual property rights and set related technical standards through scientific research and validation of applied research results to address China’s aging population as well as common but serious medical conditions.
Direction 9: Intelligent Computing and Brain-computer Interface Technology
We are motivated to investigate artificial intelligence algorithms and intelligent analysis of EEG signals. By applying wireless technology and DC-coupled analog-front-end techniques to hardware optimization, we aim to develop autonomous miniaturization techniques in wearable EEG equipment and cutting-edge brain-computer interface technology. By communicating with the brain at the brain region/neurons, we aim to develop innovative non-invasive/invasive brain-computer interface devices to meet the demands of long-term EEG signal collecting, real-time dynamic monitoring, and intelligent precision measurement. By using impedance detection technology and online signal quality monitoring technology, we seek to guarantee the dependability and safety of system performance and provide reliable technical support for the clinical treatment of patients with brain diseases and serve those with special needs.
[Team member of research program: Wen Jianming, Li Hua, Li Gang, Wang Xiaolin, Ge Xinyang]