Exclusive Interview with Professor Yang Guanyu from Southeast University: Artificial Intelligence Has Played an Important Role in Medical Imaging Diagnosis
Yang Guanyu is Vice Dean, Professor, and Doctoral Supervisor at the School of Computer Science and Engineering, School of Software, and School of Artificial Intelligence of Southeast University, as well as a Senior Member of IEEE. He also serves as a Committee Member of the Medical Imaging Professional Committee of the Chinese Society of Image and Graphics. Professor Yang holds a PhD in Biomedical Engineering from Southeast University, a PhD in Signal and Image Processing from the University of Rennes 1 (France), and has worked as a Postdoctoral Fellow at the Image Processing Laboratory of Leiden University Medical Center (LUMC, Leiden University, the Netherlands). He has long been engaged in research on medical artificial intelligence, image processing and analysis, and computer-aided diagnosis and surgery. He has undertaken or participated in more than 10 projects, including the National Key R&D Program, National Major Science and Technology Projects, National Natural Science Foundation of China, and Jiangsu Provincial Natural Science Foundation. He has published over 60 papers in top journals and conferences such as IEEE TIP, IEEE TMI, Medical Image Analysis, ECCV, and MICCAI, and holds more than 10 authorized national invention patents.
With the continuous development of artificial intelligence (AI) technology, it has begun to play an important role in medical imaging diagnosis. In recent years, a growing number of studies have shown that using AI technology to assist medical imaging diagnosis can improve the accuracy and efficiency of doctors' diagnosis, and provide patients with better medical services. AI technology can use deep learning and computer vision algorithms to conduct rapid and accurate analysis and diagnosis of medical images, which greatly improves the work efficiency and diagnostic accuracy of doctors while reducing medical costs. Nanjing Innovation Investment Group has invited Professor Yang Guanyu from Southeast University to share with us the application of AI technology in medical imaging.
What are the latest cutting-edge developments of AI technology in the medical field currently?
Let me share a few pieces of news. First, a study on skin cancer diagnosis from Stanford University pointed out that the level of deep learning models in skin cancer diagnosis is comparable to that of doctors. Second, an AI algorithm using non-contrast-enhanced cardiac MR cine sequences can effectively diagnose chronic myocardial infarction while avoiding the use of contrast agents. Third, the AlphaFold team has used deep learning models to achieve high-precision and rapid protein structure prediction.
What are the main technologies of AI in medical imaging diagnosis?
The main technologies are as follows:
(1) Image Recognition: AI technology can automatically process and analyze medical images, helping doctors identify and analyze details and features in the images. For example, in images of suspected lung nodules, AI can automatically identify and classify different types of nodules, and provide doctors with detailed analysis results and suggestions. In breast cancer screening, AI technology can automatically identify differences between tumors and normal tissues, assisting doctors in quickly and accurately diagnosing tumors.
(2) Image Segmentation: Medical images often present challenges such as blurriness, low contrast, and noise. These issues tend to make it difficult for doctors to analyze and diagnose images. AI technology can help doctors solve these problems and improve the quality of medical images. Among these technologies, image segmentation is highly useful—it can divide medical images into different regions and assign each region to a specific tissue, structure, or organ.
What are the applications of AI technology in medical imaging diagnosis?
There are mainly three aspects of applications:
(1) CT and MRI Image Analysis: CT and MRI are commonly used imaging examination methods in clinical medicine, but analyzing these images often requires a lot of time and effort. AI technology can automatically identify tissues and organs in the images and analyze the images in a relatively short time. For instance, in CT scans, AI can help doctors automatically identify bone structures, blood vessels, tumors, etc., and provide diagnostic recommendations. This greatly reduces the workload of doctors and improves the accuracy and speed of diagnosis.
(2) Mammography: AI technology is also widely used in mammography. Mammography is currently one of the main methods for early detection of breast cancer, but analyzing mammographic images requires a high level of skill from doctors. By using AI technology, doctors can be assisted in quickly and accurately analyzing images and identifying potential breast cancer lesions.
(3) Ultrasound and Cardiac Imaging Analysis: Ultrasound and cardiac imaging are crucial for the diagnosis of cardiovascular diseases. AI technology can help doctors make diagnoses by automatically identifying cardiac structures, functions, and blood flow in the images. For example, in ultrasound images, AI can automatically calculate the size and movement range of the heart and generate 3D images of the heart, helping doctors diagnose conditions more accurately.
What practical cases have there been recently in China regarding AI + medical imaging?
A recent relevant research case involves the Jiangsu Provincial People's Hospital, which proposed a new method for treating kidney cancer—laparoscopic partial nephrectomy with segmental renal artery occlusion. This surgical technique is highly difficult and requires a precise personalized preoperative surgical plan. Our team, in collaboration with the Department of Urology and the Department of Radiology of Jiangsu Provincial People's Hospital, used AI technology to develop a series of key preoperative image processing algorithms based on 3D convolutional neural networks. These algorithms enable fully automatic segmentation of surgery-related organs (such as the kidney, tumor, arteries, and veins) in preoperative CT images, as well as quantitative evaluation of segmental artery blood supply. This creates conditions for the design of precise laparoscopic partial nephrectomy plans. The relevant results have been published in top journals and conferences in the fields of medical image processing and AI, including Medical Image Analysis, IEEE JBHI, MICCAI, and ICAI.
Could you share with us the future development trends of AI technology in medical imaging diagnosis?
I believe there are four main future trends:
(1) Higher Precision: With the continuous advancement of deep learning technology, the accuracy of AI algorithms will become higher, and diagnostic results will be more accurate. Currently, AI algorithms can already perform complex image analysis and diagnosis on medical images, but there are still cases of misdiagnosis in some details. In the future, as deep learning technology further improves, the accuracy of AI algorithms will be higher, and diagnostic results will be more precise.
(2) Personalization: Future medical imaging diagnosis will be more personalized. AI technology can analyze each patient's imaging data and provide more accurate diagnosis and treatment plans based on the patient's individual differences. For example, using the patient's genomic data and imaging data, AI technology can predict the patient's risk of developing a certain disease or provide personalized drug treatment plans.
(3) Telemedicine: With the popularization of 5G technology, the application of AI technology in medical imaging diagnosis will become more widespread. 5G technology can achieve faster and more stable data transmission, sending medical imaging data from the patient's location to remote doctors. This greatly improves the efficiency and quality of medical services.
(4) Education and Training: In the future, AI technology will also be applied to medical research and education in the field of medical imaging diagnosis. AI technology can use a large amount of medical imaging data for analysis and research, providing more data support and methodologies for medical research. At the same time, AI technology can be applied to medical education, providing medical students with more realistic and intuitive learning experiences, and improving doctors' diagnostic capabilities and work efficiency.
Thank you for Professor Yang's sharing. We believe that as the role of AI technology in medical imaging diagnosis becomes increasingly prominent, the efficiency and quality of medical services will surely improve rapidly in the future!
Source: Zhang Surong (Fourth Investment Department), Yang Guanyu
Reviewer: Xue Yao
Publisher: You Yi
