Numerical Lymic Stool Type Features

AI Ultrasound: AI imaging field of up-and-coming.

AI ultrasound receives far less attention than AI CT images in the various subdivisions of artificial intelligence (AI) medical imaging. In early 2020, the FDA approved the first AI-assisted ultrasound diagnostic software, Caption Guidance, which received more attention as a major breakthrough in the field of AI ultrasound.

Improve diagnostic accessibility.

Ultrasound testing is a safe and inexpensive medical diagnostic method, but in the actual diagnosis and treatment process, the use of ultrasound is not high. Currently, there are about 50 million doctors worldwide, but only 2% have the skills to perform ultrasound scans. In addition, according to the China Medical Equipment Association statistics, China's ultrasound equipment ownership is not low, but the number of hospitals at all levels is not balanced. By the end of April 2018, China's 2427 tertiary hospitals have 24,270 color super equipment, with an average of 10 color super-equipment, while the average number of secondary and first-level hospitals has 5 color super equipment, the gap is more obvious.

Ultrasound detection has the characteristics of radiation-free, repeatable diagnosis, with the progress of technology, ultrasonic testing costs are becoming lower and lower, equipment is gradually miniaturized. Only by lowering the threshold for use can it become a truly inclusive, portable diagnostic tool.

Ultrasound diagnostics are different from radiology diagnostics - radiologists can diagnose by still images, ultrasound doctors need to collect dynamic images of different faces for real-time diagnosis, and the acquisition and diagnosis of ultrasound images rely heavily on the experience of the doctor. AI is mounted on ultrasound equipment and can help solve two problems: how to better obtain images and how to better analyze them.

AI needs to complete three steps to capture and analyze images in a short time. In the case of Caption Guidance, the software first uses AI to guide doctors to obtain images, and non-professional doctors can capture ultrasound images through real-time guidance of AI; In general, doctors who work on interpreting and analyzing ultrasound imaging require years of learning, while In-depth learning, Guidance can automatically measure blood score and assist doctors in assessing patients' conditions.

AI makes ultrasound testing easier and more accessible, and ultrasound is more conducive to abetter AI value. In general, areas of economies of scale are more likely to generate AI value depressions. Ultrasound has more applications than CT and nuclear magnetism, so the prospect of commercialization in the field of AI-assisted ultrasound diagnostics is more attractive.

"One big, one small" two paths.

The combination of AI and ultrasound is becoming a rising star on the AI imaging circuit. In addition to helping with better diagnosis, AI can also perform automated image quality evaluation, image standardization, image sketching, automatic measurement, and more in ultrasonic images. These functions cannot be implemented in ultrasound through a common scheme, so AI ultrasound takes two completely different paths.

One route is in the traditional ultrasound department, AI makes large ultrasonic equipment more intelligent, so that ultrasonic equipment is no longer just an imaging product, but become a collection of data collection, management, analysis in one, into the deep learning of the intelligent terminal. In 2019, General Medical has launched a LOGIQTM E20 equipped with a cSound-TM image generator in China, which can realize the functions of tissue organ structure screening, intelligent lesions segmentation, intelligent measurement and so on through the perception of images, and help doctors get rid of the many redundant image optimization and measurement work and focus on clinical diagnosis and treatment. The device is mainly used in intervention, thyroid, breast, muscle, pediatrics, heart and other clinical fields, to assist clinicians with accurate diagnosis.

Of course, the role of AI is currently just the icing on the cake for large devices, but it is foreseeable that AI will play an increasingly important role in the future. At the same time, compared with the hardware ultrasonic equipment, AI software iteration is faster, software and algorithms are expected to become the mainstream research direction in the field of ultrasound in the future. Highly digital devices generate large amounts of data, and how to connect and consolidate data is key to research.

The other route is applied to the primary care scenario. China has nearly 900,000 primary medical institutions, medical, pharmaceutical, inspection of these three links, to crack the medical architecture contradictionis is essential to increase investment in the inspection of this link. With the help of portable handheld ultrasound equipment, it is a feasible route for ultrasonic equipment to empower primary medical institutions.

At present, the layout of the primary hospital market is mainly start-up companies, these companies are mainly a use of AI technology in handheld ultrasound equipment, more use of them are ultrasound testing experience of doctors. In the past, ultrasound diagnosis needed to rely on professional doctors through the eyes to identify the anatomical structure in the image, and AI through intelligent recognition, can automatically find the best image and assist diagnosis, so that ultrasonic test operators are not limited to professionally trained doctors, more primary medical institutions of General Practitioners can also carry out ultrasound diagnosis.

The track's "late maturity" stems from technical barriers.

Comparing the AI CT images, the AI ultrasound track is not crowded, with only a few start-ups involved. So why is AI ultrasound a late-maturing track in the field of AI imaging? The main reason is that AI ultrasound technology is more challenging than other imaging fields.

The main technical challenges of AI for ultrasound are three:

The first is real-time diagnostics. Unlike the static images of CT and MRI, ultrasonic images are dynamic real-time images, and the difficulty of ultrasonic detection lies in the simultaneous completion of image acquisition and reading. The collection of IMAGES such as CT, nuclear magnetic and X-ray is completed by technicians, while the reading film is completed by a radiologist, and ultrasound testing requires image acquisition and reading, which puts forward higher requirements for auxiliary diagnostic technologies such as algorithms and calculation sourcing.

Secondly, in the data, because of the special data browsing, processing and storage habits of ultrasound image, the image data is more difficult to obtain than CT image, the size of the database is limited. In addition, the standardization of ultrasound imaging is low, image clarity mainly depends on the operator of the ultrasound doctor and the equipment model, AI ultrasound needs to be cleaned and analyzed by a strong team of experts.

Finally, the limitations of the algorithm framework. For AI ultrasound enterprises, it is very important to have their own algorithm framework. However, the vast majority of enterprises are using open source algorithms, with very few independent algorithms. Unlike radioaIta, in order to ensure the accuracy and real-time of the analysis, it relies heavily on the self-developed algorithm framework, which can lead to slow processing of the device if the algorithm is too long. Ultrasound detection requires high real-time, producing dozens or even hundreds of frames in a second, and without powerful algorithms, it is impossible to process such a large amount of data.

In addition to the above three points, if you want to carry AI technology on the handheld ultrasound, you also need to solve the problem of force limit. Because the handheld ultrasound equipment is much smaller than the traditional ultrasonic equipment, equipped with AI software is very testing the calculation power of AI. Not only do businesses need a very accurate model to analyze ultrasound video, but on this basis, they must also ensure that the model works effectively with the limited resources of a tablet or mobile platform.

In the process of the continuous evolution and progress of ultrasonic technology, hardware and software capabilities are very important. At this stage, AI ultrasound software system technical barriers are higher, in the hardware relative homogenization of ultrasonic equipment market, whether to develop a good AI software, to a certain extent determines the application space of AI ultrasound producthardwy. At the same time, for primary hospitals, portable or small devices are cheaper than large devices, so AI handheld ultrasound is more suitable for promotion than large devices such as AI CT. How to realize the intelligent process of AI in the palm of the hand to better meet the application needs of general practitioners in primary hospitals is a key problem that enterprises need to solve.