Xinzhiyuan participated in the Smart Future Medical Artificial Intelligence Summit on March 25. The event was jointly sponsored by Huiying Huiying and Intel Medical. Among them, Huiying Huiying was established in 2015. Its founder, Chai Xiangfei, returned from Stanford University two years ago. At that time, he thought that Dr. Chunyu had done a very good job, and he saw great innovations in the business model of medical imaging. Therefore, he and Guo Na founded Hui Medicine Huiying, dedicated to discovering the value of medical image data, and using the most advanced cloud computing, big data, and artificial intelligence technologies to create intelligent medical imaging platforms and tumor radiotherapy platforms. In this process, Intel Medical provided a lot of support. Huiying Huiying also became a member of Intel's Joint Lab.
The guests participating in this medical AI summit included Li Yadong, general manager of Asia Pacific of Intel Medical and Life Sciences Department, Chai Xiangfei and Guo Na, founders of Huiyin Hui, Chen Weiguang, partner of Lanchi Ventures, and Xing Lei, a tenured professor of radiotherapy at Stanford University. Zhang Qin, deputy general manager of Wanfang Data Co., Ltd. and Wang Xinjun, secretary of the Party Committee of the Zhengda Five Affiliated Hospital and a representative of a famous equipment manufacturer. This article synthesizes the conference presentations, forums, and Xin Zhiyuan’s exclusive interviews with Stanford medical AI expert Xing Lei, allowing us to break the superstition and follow the AI ​​technology and the healthcare industry to unveil the current status of medical AI in China.
Strengthen deep learning: the most promising technology for medical AI?
Since 2012, deep learning technology has been introduced into the image recognition data set ImageNet (as a test standard), and its recognition rate has reached new heights in recent years, and it has reached the human level in certain fields such as image classification. Deep learning technology, combined with years of accumulated data in the medical imaging field, has brought a surprising breakthrough in this area.
Xinzhiyuan has reported that researchers at Stanford released a study on Nature. CNN diagnosed skin cancer and compared it with 21 dermatologists. The result was a system that was as accurate as human doctors (“at least†91%). There is also the diagnosis of diabetic retinopathy using CNN published on JAMA. The results show that the performance of the algorithm is consistent with the performance of ophthalmologists.
The application of CNN in medicine can be described as being applied to top-level publications. Is deep learning technology the best technology in the field of medical imaging?
Professor Xing Lei of Stanford University told Xinzhiyuan that deep learning and enhanced in-depth learning represent the latest technologies. They can solve many problems that could not be solved before and push medical AI to a new climax.
Deep learning
Intensive deep learning AlphaGo played against Li Shizhen in early 2016. AlphaGo learns the game to a certain extent, and can use reinforcement learning to further improve it in a large number of chess games that rivals and opponents. This is arguably the key to its continuous surpassing itself and ultimately overcoming the human championship. As an effective machine learning method, reinforcement learning mainly studies the operation mode in a specific situation or environment, so that the reward signal is maximized. Also in the decision-making process of medical AI, a program operation will often affect the data it receives. In different operations, the program will receive different input information. Reinforcement learning can find the best plan for a decision or operation to get the maximum reward.
Analysis of mammography of molybdenum target
Dr. Xing Lei introduced that in fact, before the deep learning, in the 90s, many people already had computer-aided diagnosis (CAD). Before the neural network was not deep, now that there is a new type of computer and deep learning, deep networks can be realized. However, taking the diagnosis of skin cancer as an example, there is currently no clinically significant large-scale application. Diagnosis based on deep learning is still in the research and development stage. However, with the current R&D speed, these new technologies are not far away from clinical applications.
Cerebral hemorrhage area is automatically marked
In addition, deep learning is not required in all scenarios. This is related to the specific problems encountered, and sometimes general machine learning techniques are sufficient. The new algorithm can be said to be endless, with each passing day. Combining different algorithms is also a common method in the field of AI.
1 2 3 Next>- The Description of GPS Antenna
- GPS is a terminal for positioning or navigation by receiving satellite signals. An antenna is needed to receive the signal. GPS satellite signals are divided into L1 and L2 with frequencies of 1575.42MHz and 1228MHz respectively. L1 is an open civil signal and the signal is circularly polarized. The signal strength is about -166dBW, which is a relatively weak signal. These characteristics decide to prepare special antennas for the reception of GPS signals
- The use of GPS in cars will become more and more common. But the automobile shell, especially the automobile explosion-proof film produces the serious obstacle to the GPS signal. An external antenna with a magnet that attaches to the roof of the car is necessary for a GPS in the car.
-
The GPS antenna plays a pivotal role in ensuring reliable reception of these satellite signals, even in challenging environments. It is designed to be highly sensitive to the specific frequencies of GPS satellites, namely L1 and L2, enabling it to capture the weak signals emitted by the satellites effectively. The antenna's design focuses on enhancing signal capture while minimizing interference from external sources, such as the metallic body of a vehicle or the shielding effects of an automobile's explosion-proof film.
For automotive applications, GPS antennas are often equipped with a strong magnetic base, allowing them to be securely attached to the roof of a car or any other metallic surface. This placement is strategic as it positions the antenna away from the obstructive materials inside the vehicle, providing a clearer line of sight to the sky and the orbiting GPS satellites. By doing so, it significantly reduces signal attenuation, ensuring accurate and consistent positioning data for the vehicle's navigation system.
Moreover, these external GPS antennas are designed to withstand various weather conditions, including extreme temperatures, moisture, and vibrations, ensuring reliable performance over time. Some advanced models also incorporate features like multi-path mitigation, which helps to filter out reflected signals that can cause positioning errors, further enhancing the accuracy of the GPS system.
In summary, the GPS antenna is a crucial component that enables efficient and accurate reception of satellite signals, especially in automotive applications where signal obstruction is a common challenge. Its specialized design and strategic placement contribute significantly to the overall performance and reliability of GPS-based navigation systems.
- The Picture of GPS Antenna
gps marine antenna,high gain gps antennas,gps car antenna,gnss & gps antenna,gps antenna tracker
Yetnorson Antenna Co., Ltd. , https://www.yetnorson.com