AI image reading highlights the advantages of technology in medical treatment

In the early stage of the new crown pneumonia epidemic, in terms of nucleic acid detection, due to the limited experience of medical staff in the use of kits, it was difficult to collect saliva samples near the upper respiratory tract, resulting in false negatives of patients, resulting in limited nucleic acid detection to a certain extent.

The sudden outbreak of the new crown epidemic has made AI film reading a lot of use.

AI image reading is the use of AI for medical imaging diagnosis. The outbreak of the epidemic has made the medical system see the advantages of AI technology.

In the early stage of the new crown pneumonia epidemic, in terms of nucleic acid detection, due to the limited experience of medical staff in the use of kits, it was difficult to collect saliva samples near the upper respiratory tract, resulting in false negatives of patients, resulting in limited nucleic acid detection to a certain extent.

With the accumulation of clinical diagnostic data, the imaging big data characteristics of COVID-19 have gradually become clearer, and CT imaging diagnosis has been included in the “Health and Health Commission Guidelines for the Diagnosis and Treatment of COVID-19 (Fifth Edition)”. Compared with nucleic acid detection, CT imaging images are clearer. It has a high degree of sensitivity and can diagnose early mild lesions, so CT imaging has also become one of the ways to detect new pneumonia.

Front-line battlefields such as Huoshenshan Hospital, Leishenshan Hospital, Wuhan Tongji, Xiehe, and Zhongnan Hospital have been deployed and put into use. With the accumulation of clinical data, AI + cloud will also play a greater role in the diagnosis of new coronary pneumonia.

In the past two years, AI has become more and more prominent in the application of the medical and health industry, including medical robots, intelligent drug research and development, intelligent diagnosis and treatment, intelligent image recognition, intelligent health management and other directions. These applications are used in the prevention and treatment of this epidemic. It played an important role.

Difficulties in whether AI film reading can stand for a long time

① Pneumonia is divided into many types, viral, bacterial, fungal, mycoplasma, chlamydia, allergic and so on. Among them, the imaging manifestations of viral pneumonia are different from other types of pneumonia. The current difficulty is that CT cannot accurately determine whether a patient is carrying the new coronavirus.

That is to say, there may be inconsistencies between CT and nucleic acid test results. In this case, the epidemiological history and clinical manifestations of the patient can be combined to diagnose new coronary pneumonia more accurately.

②AI can effectively identify nodules that are easily missed, such as solid nodules under 6 mm and ground glass nodules. But in the face of the new crown pneumonia virus, there is still more room for AI.

It is still very difficult to identify which virus is viral pneumonia by CT images alone. If AI can make achievements in this regard, it will be a great breakthrough, but it is very difficult.

③In the early stage of the epidemic, although there were many suspected CT data of new coronary pneumonia, there was a lack of effective labeling. CT photos seem to be massive, but for AI, it is still not enough. To make AI smarter and smarter, it needs to “feed” a large amount of data.

However, in medical scenarios, no matter how large the total number of patients is, it is difficult to obtain a large amount of data training in a certain hospital, and the individual patients vary greatly, and each person’s disease course, condition, and disease types will be different.

④ From a national perspective, the problem of unbalanced regional distribution of human resources in imaging departments is prominent. It is difficult to say whether primary medical units can use AI medical imaging diagnosis systems like large hospitals.

⑤ At present, AI still fully relies on the support of data authenticity and quality in the imaging diagnosis of the nervous system. In the field of diagnosis of difficult and rare diseases that lack the support of big data, there is still a gap between AI and professional doctors.

AI image reading highlights the advantages of technology in medical treatment

Commercial development of AI medical industry

From 2013 to 2017, China’s medical artificial intelligence industry received a total of 241 financings. Among them, in 2017, nearly 30 financing events were announced in the domestic medical artificial intelligence industry, with a total financing of more than 1.8 billion yuan.

In 2018, the enthusiasm of capital for the medical artificial intelligence market is still unabated. In the first half of 2018 alone, 18 companies have been invested, with a total amount of more than 3.1 billion yuan. As of June 2018, a total of 89 medical artificial intelligence startups in China have received investment, with a total amount of about 21.938 billion yuan.

At present, most start-ups in China focus on auxiliary diagnosis, and the specific business is mainly imaging, intelligent auxiliary diagnosis system and speech recognition, covering a wide range of diseases, but most of them focus on imaging and pathological pictures based on image recognition. Identified diseases, such as lung cancer, cervical cancer, etc.

However, although the enthusiasm of capital has pushed medical artificial intelligence startups to the forefront, for startups, how to achieve commercialization is a huge problem. At the same time, the competition of peers and the annexation of giants also bring certain difficulties to these enterprises.

The biggest problem of AI in the medical field may be that there is no very good business model, or there is no real closed loop to realize self-hematopoietic ability. The main reason is that a particularly good application may not be found, and it is currently in the stage of continuous trial and error.

The end point of AI imagery is at the grassroots level

The top three hospitals are the main positions of medical imaging AI. Large hospitals are overcrowded and doctors are overwhelmed. The emergence of AI can relieve the work pressure of doctors, free up time for doctors to think about problems at a higher technical level, and allow doctors to return to medical treatment itself.

The role of AI at this stage is mainly an auxiliary tool, and it has an obvious role in a few medical diagnosis processes. In addition, AI should not stop at large hospitals, and grassroots hospitals also need the help of AI. Whether it is AI software or system equipment, there is a lot to do at the grassroots level.

For grassroots medical institutions, medical resources are insufficient. Because grassroots doctors do not have the opportunity to contact and study a large number of different cases, they lack corresponding experience in reading images. AI can assist primary doctors in making a diagnosis.

Both large hospitals and grassroots hospitals need AI, but the needs are different. With the development of technology, AI will present greater social and commercial value at the grassroots level.

A new medical technology is often passed from high-level to low-level. AI technology is polished, piloted, and mature in large hospitals, and then sinks to grass-roots hospitals after being recognized. This development path is different from AI technology in other industries.

Overseas will become the future main market for medical AI

At present, the epidemic situation in my country has entered the final stage, and the focus of epidemic prevention is also to prevent imported infections. Outside the country, the number of patients with new coronary pneumonia has risen sharply. In this case, AI companies have also taken advantage of the situation to go abroad and expand overseas markets.

At present, overseas countries are advocating nucleic acid testing, while my country is focusing on the combination of nucleic acid testing and imaging targeting.

The fate of Chinese medical AI companies is to go overseas. Even if they are “eaten” by overseas leading companies, they must compete in the global landscape. Competitors of Chinese medical AI companies should come from overseas markets.

In fact, the United States, Israel, and India are all powerful competitors. The advantage of the United States is that the commercialization path is very smooth.

The medical industry is huge, and the society encourages the implementation and charging of innovative technologies. In countries such as Israel, the government will take the lead to help companies obtain scientific research data and ensure the accuracy of their products, so as to go abroad and get a share of the international market.

In short, with the gradual maturity of artificial intelligence technology, a trillion-level artificial intelligence health industry is about to emerge. The vertical extension to the therapeutic field is a major trend of artificial intelligence. At present, the application of artificial intelligence technology still has a long research path to go. Only from medical assistance to the stage of influencing treatment decisions can we explain the true trend of “AI + medical care”. Mature.

The Links:   NL6448BC26-09C BSM200GB120DN2

Scroll to top