5 minutes to complete the MRI check Facebook's AI technology is reliable?

Intelligent relativity: Deeply digging artificial wells, judging saltiness, telling black and white, and telling the depth.

Recently, Facebook's artificial intelligence lab has teamed up with New York University School of Medicine to try to increase the speed of magnetic resonance imaging (MRI) inspections by a factor of ten. MRI examinations typically range from half an hour to an hour. If the rate of inspection is increased tenfold, future radiologists are expected to complete the test in just a few minutes. And what does this seem like little technical cooperation bring to the current medical situation?

AI accelerates MRI examination and blows a new medical wind

No matter how much people talk about artificial intelligence to defeat human Go masters, game masters, etc., as long as the practicality of artificial intelligence is mentioned, it will be a bit scornful. Although there are huge slogans everywhere telling us that the world has entered the era of intelligent development, even if we take the most powerful smartphones, the best thing to do is the AI ​​photo, which is still the same for thousands of years. disappointing.

Therefore, if artificial intelligence wants to be recognized by the public, it still needs to bring a new wind to the public's life. So, let's take a look at this Facebook's AI accelerated MRI to see if this new wind is still comfortable.

1. Technology is more adaptable to human nature and optimize medical experience

If the time of MRI can be shortened to a few minutes, and no matter how advanced its medical significance, it is a great boon for ordinary patients. Because, first of all, when performing MRI, the patient not only needs to lie still in the examination equipment, but also requires control of breathing, or even swallowing saliva, otherwise it will easily affect the examination results.

In addition, in the claustrophobic inspection space, the inspection machine will continue to produce awkward noise, which is unbearable. However, due to the long time of traditional MRI, this further aggravates the pressure on patients due to MRI. Therefore, completing such an examination is a very difficult thing for most people, especially children and the elderly.

However, if the MRI examination time becomes 5 minutes, then the painful medical experience in the past will be improved, and the technology adapted to human nature can also benefit more people.

2, the public medical examination needs are more comprehensive coverage

Between June 2014 and the end of July 2016, there were about 40 investment M&A events in the medical imaging field, which was very popular with capital. Behind this is the surge in the demand for medical imaging inspections brought about by the simultaneous rise of China’s total population and aging population, as well as the rise in overall social income.

In the past 20 years, among the population structure of our country, the age group with high incidence of diseases has continued to increase, resulting in an increase in per capita medical imaging demand. The increase in per capita GDP has led to an increase in consumption, which has led to an increase in public interest in distributing income in the medical field. Especially in the face of medical imaging examinations, its own preventive significance is quite important for the current population.

However, back to the current field of medical imaging inspection, the traditional CT examination has a short inspection time but brings about a large radiation damage. Traditional MRI tests have no radiation damage, but the time required is too long. This means that in the face of some emergency patients, the hospital can only use CT and cannot use MRI. In addition, due to the long inspection time of nuclear magnetic resonance, the number of patients who can be scheduled for examination every day is very limited, and the consequences are the imbalance in the supply of imaging examinations and the waiting time for a large number of patients in imaging examination. Too long.

Therefore, when the accelerated MRI enters the hospital, it will be able to more comprehensively cover the current medical examination needs from the public, thereby alleviating the current situation of supply imbalance caused by the currently too tight medical resources.

But across the 1 hour and 5min "the long river of time", you still need to trek

However, the 5-minute check-up time is just a slogan from Facebook. While applauding this swearing, we still need to hear the screams from all sides.

1. The underlying building of AI may be unstable

The reason why MRI time is so long is its physical principle. During the inspection process, different atoms are excited in the magnetic field under different imaging. The time required for the atom to recover from the excited state to the initial state determines the length of the scan time. However, different atom recovery times are different, and the recovery time is not changed by the blessing of external software.

5 minutes to complete the MRI check Facebook's AI technology is reliable?

This shows that the deep learning technology used by Facebook does not substantially shorten the inspection time of MRI. In fact, Facebook's new bottle of old wine this time is only an extension of machine deep learning in the image field. To put it simply, Facebook mainly uses AI's deep learning technology to transform low-resolution medical image restorations formed in a short period of time into high-definition images that can be diagnosed by doctors.

However, this is back to the issue of AI data, the underlying building. First of all, the medical data currently available is risky. It is not known whether the medical image data used by AI for deep learning has been approved by the patient and whether the patient's privacy is exposed. In the future, if the research and development results are commercialized, it will inevitably face privacy questioning from the public.

Moreover, the quality of medical data is still difficult to guarantee at present, the accuracy of medical data needs to exceed the data required in any field, and the personnel who identify medical data are currently under-resourced, the level of young doctors is not enough, and highly qualified doctors It seems to be overkill, and how to resolve the contradiction between them is probably one of the key factors for improving the quality of medical data in the next step.

2. Technology popularization is difficult to get manufacturers nod

Shortening the time of MRI examination is undoubtedly a good thing for the medical practitioners, but it is still unknown whether this good thing has a wide range of conditions. At present, the average possession and per capita possession of medical magnetic resonance equipment in China are at a low level. However, as the consumption level of residents continues to rise and health awareness continues to strengthen, the market demand for medical magnetic resonance equipment will be large in the future. The increase in magnitude, which is the three major manufacturers of high-end magnetic resonance equipment, General Electric, Siemens and Philips was originally a signal of rising profits.

However, if the AI ​​technology shortens the time required for MRI, the number of inspections that a magnetic resonance device can perform in a single day will increase by a factor of ten. Returning to the problem of imbalance between supply and demand, it will be found that the increase in the number of demand side does not have to be solved by increasing the quantity on the supply side, and improving the quality of the supply side can also alleviate the problem.

However, for the three major players, they may not be so easy to nod the popularity of this technology. First, they have no incentive to carry out product AI. Looking at the global MRI market, Siemens, Philips and General Electric currently have a market share of 80%, which is basically a monopoly. Take the domestic situation as an example, these imported brands have already occupied 70% of China's high-end medical device products, while most domestic brands are concentrated only in the low-end market. Therefore, the three major manufacturers without external threats do not have any pressure to have to carry out AI-level blessing and product reform.

Second, they are not willing to be profited. The cost of embedding AI technology in magnetic resonance equipment will increase the cost of magnetic resonance equipment by increasing the cost of production. But with the increasing openness of AI technology, what we can see is that the AI's blessing may not increase the price of magnetic resonance equipment. Therefore, the three major manufacturers may need to carefully consider the popularity of this technology.

3. Hype ingredients dilute the public's recognition of artificial intelligence

Whether it is 5 minutes to complete the MRI check, or more than 90% of the AI ​​"reading rate", or artificial intelligence defeated human Go players and game masters. These media assertions have touted artificial intelligence technology far beyond the current level of artificial intelligence in real life. This is like the story of "Wolf is coming". A simple scientific research cooperation, an ordinary scientific research result, may become an explosive science and technology news under the second interpretation of the media, and this is not the original intention of researchers.

Moreover, too much media hype is easy to kill the public's expectation of artificial intelligence, and will have an indelible negative impact on the promotion of artificial intelligence technology that will truly benefit the society in the future. It is envisaged that if the entire artificial intelligence industry is caught in the “Taxi trap”, then the first one to suffer the trust will be the medical industry.

At present, completing the MRI in 5 minutes is only a research goal, but the media often uses a positive tone to attract readers to click. But in the face of such rigorous news reports, the public should learn to stay calm. Can the magnetic resonance image scanned in 5 minutes reach the resolution for the doctor to diagnose? Is the learning data of the rare disease enough? If the image clarity is not enough, can the tiny lesion be discovered? How many years does this technology take? This is the question that the public should ask themselves when facing such information.

Nowadays, the gimmicks of artificial intelligence are too large, but the lack of realistic landing scenarios has led the industry to use technology marketing to supplement market confidence and attract capital. But after all, this is not a long-term solution. We must start to pay attention to whether the public's confidence can still be sustained under the artificial intelligence hype.

Activated Carbon Rotary Kiln

JHR series Activated Carbon Regeneration equipment is a series of equipment for activated carbon fire regeneration developed by Changchun Gold Research Institute Co., Ltd. in the 1990s and put on the market. After more than 30 years of application and continuous development, the series of equipment has been changed according to different heat sources. Three series of products are derived, the JHR-O series with diesel-flavored heat source, the JHR-A series with natural gas as the heat source, and the JHR-E series with electric energy as the heat source.

This series of equipment can regenerate various types of activated carbon such as coal, wood, nut shell, mineral (petroleum), or powder, granule, column, or primary, extruded, and re-formed activated carbon, according to the characteristics of activated carbon fire method regeneration treatment , This series of equipment is equipped with unique flue gas treatment equipment to ensure the green and environmental protection of the activated carbon fire method regeneration process.

Activated Carbon Regeneration Kiln,Kiln Furnace,Activated Carbon Regeneration,Elution Of Gold From Activated Carbon

Changchun Gold Research Institute Co.LTD , https://www.changchungoldresearch.com