What revolutionary changes has AI brought to the medical field?

May 5, 2019 / BIOON / - when Google deepmind's alphago beat legendary go player Lee sedol in 2016, the terms of artificial intelligence (AI), machine learning and deep learning were pushed into the mainstream of technology AI is usually defined as the ability of computer or machine to display or simulate intelligent behavior, such as Tesla's app and le's Siri This is a booming field, and it is also the focus of many research and investment Machine learning is the ability of artificial intelligence system to extract information from original data and learn prediction from new data Deep learning combines artificial intelligence with machine learning The algorithm it focuses on is inspired by the structure and function of brain, which is called artificial neural network Photo source: https://cn.bing.com recently, in-depth learning has attracted widespread attention in the consumer world and the entire medical community Alex net, a neural network designed by Alex krizhevsky, won the 2012 Imagenet large scale visual recognition challenge, an annual image classification competition Another relatively new development is the use of graphical processing units (GPUs) to support deep learning algorithms GPUs is good at deep learning the computation (multiplication and addition) required by the application, thus reducing the processing time of the application At the University of Saskatchewan laboratory, researchers are conducting interesting in-depth learning studies related to healthcare applications - led by Professor seokbum Ko, a professor of electrical and computer engineering In health care, the use of artificial intelligence or machine learning for diagnosis is a new technology, and exciting and promising progress has been made It is an important method for the diagnosis of diabetes and heart disease to extract the vascular and retinal vascular abnormalities in the eyes In order to provide reliable and meaningful medical information, doctors must extract retinal vessels from retinal images for reliable and meaningful interpretation Although manual segmentation is possible, it is a complex, time-consuming and tedious work, which requires high professional skills The team has developed a system that can segment retinal blood vessels by reading original retinal images It is a computer-aided diagnosis system, which reduces the workload of eye care experts and ophthalmologists, improves the speed of image processing by 10 times, and maintains a high accuracy Image source: https://cn.bing.com detection of lung cancer computed tomography (CT) is widely used in the diagnosis of lung cancer However, due to the similar visual effects of benign (non cancerous) and malignant (cancerous) lesions in CT scan, CT scan can not always provide a reliable diagnosis Even a chest radiologist with years of experience With the rapid development of CT scan analysis, it is urgent to use advanced computing tools to assist radiologists in screening Photo source: https://cn.bing.com in order to improve the diagnostic ability of radiologists, researchers proposed a deep learning solution According to the results of their study, their solution is superior to that of experienced radiologists In addition, the use of deep learning based solutions improves diagnostic performance overall, with less experienced radiologists benefiting the most from the system Limitations and challenges although deep learning algorithms have shown great prospects in various tasks of Radiology and medicine, these systems are far from perfect Obtaining high quality annotated datasets remains a challenge for deep learning training Most computer vision research is based on natural images, but for healthcare applications, we need large annotated medical image data sets From a clinical perspective, another challenge will be to test how deep learning techniques perform compared to human radiologists More cooperation is needed between doctors and machine learning scientists The high complexity of human physiology will also bring challenges to machine learning technology Another challenge is to validate the need for deep learning systems for clinical implementation, which may require multi agency collaboration and large data sets Finally, an efficient hardware platform is needed to ensure the fast processing of deep learning system In the complex field of healthcare, artificial intelligence tools can support human practitioners to provide faster services and more accurate diagnosis, and analyze data to identify trends or genetic information that may lead to a specific disease When saving time means saving lives, artificial intelligence and machine learning may have a revolutionary impact on medical workers and patients Reference materials: [1] faster, more accurate diagnoses: healthcare applications of AI research [2] when Siri met Hal [3] zhexin Jiang et al Retain blood vessel segmentation using full revolutionary network with transfer learning Computerized medical imaging and graphics Https://doi.org/10.1016/j.computerag.2018.04.005