With all the diseases and landings, how can Voxel Technology obtain nearly $100 million in financing?
From the reaction of investment institutions, it is obvious that the difficulty of financing in the second half of 2018 has become an indisputable fact. However, emerging technology fields have never lacked layouts. Huihui Huiying and Pushing Technology have released a new round of financing in the cold winter. Technology has more to harvest the largest amount of 50 million US dollars, the total financing amount will soon break billions of dollars.
The harvest is always related to the effort. In 2018, the voxel developed a progressive and dense V-type deep network and agent supervised learning algorithm, which solved the problem of organ positioning, navigation, segmentation and weak data utilization in 3D images, and released the "Study of Body Skin". More than 200 skin diseases were analyzed using a multitasking model.
Recently, at the "2018 Future Medical Top 100" forum held by the Arterial Network, the reporter had the opportunity to interview Ding Xiaowei, the CEO of Voxel Technology, to try to understand the logic behind the huge financing.
Ding Xiaowei, founder and CEO of Voxel Technology
Under the pressure of economic downturn, what do investors value?
Whether it is possible to build core competitiveness in a limited time is the topic that investors are most concerned about. Yu Jia, Managing Director of Sequoia Capital, said: After the industry concept period and capital winter, technology and product barriers have become the main focus of investors.
The objective factors that measure technical barriers include both performance and team. The existing achievements reflect the current level of technology research and development of the company, while the team ability implies the development potential of the company.
In the case of unknown commercialization, choosing a technology-competitive enterprise is a good way to avoid risks, and a quality entrepreneurial team is the core of a long-term enterprise. For these two points, voxel technology has undoubtedly handed over a good answer.
On the one hand, the addition of Prof. Demetri Terzopoulos and Professor Eric Topol, founder of Cleveland Clinic Lerner Medical School in Cleveland Medical Center, infused the talents into the voxels; on the other hand, the simultaneous development of multi-product lines allowed the coronary, fundus and skin of voxels The sensitivity and specificity of such products continue to improve. Taking the fundus as an example, the sensitivity of the whole retinal solution to the presence or absence of sugar nets has reached 97% and the specificity has exceeded 90%.
Today, voxels are trying to liquidate products from multiple sources, all of which are voxel technology valuations.
Design details, what product philosophy should we follow?
The role of medical artificial intelligence products is more complicated. Generally speaking, the products of enterprises are designed for doctors, and the actual paying party is hospitals, but the direction of services is “patient-centeredâ€. The misplacement of payments, services and services has led to conflicts of interest.
From the current form, most companies are designed around doctors. This is due to the fact that at the current stage of landing, doctors have a greater say in deciding whether to accept AI products.
Nowadays, with most of the head enterprises crossing the B round of financing, the idea of ​​polishing products only for the doctor's preference may need to be changed, and companies generally face the pressure of liquidation.
Under this circumstance, Ding Xiaowei believes: "The appeal of different objects is different. The hospital cares about the rationality of resource allocation in the whole hospital. In the department, doctors are more concerned about the performance of products. Among doctors, seniors and low years There is a big difference in the demand for AI products among doctors. The senior doctors hope that AI can solve the efficiency problem. The low-grade doctors hope that the products you design can make reminders to increase the confidence of diagnosis and prevent missed diagnosis. It will also react to product differences, as in the United States, where the use of new technologies is best for medical insurance."
Therefore, when designing and improving products, voxel technology often follows the bottom-up and top-down approach, collects multiple opinions, synthesizes them, finds the largest public area of ​​the audience, and calculates the actual possibility. Increased efficiency. In the face of geographical and national differences, separate product lines need to be designed separately.
From single disease to whole disease, the process of quantitative to qualitative change
To date, most AI imaging products have focused on single heavy tasks such as pulmonary nodule detection and fractures. As a small tool, such artificial intelligence products are sufficient, but for most scenarios, the value of a single disease analysis is very limited.
Therefore, voxel technology puts the core of research on the exploration of all-species products, Ding Xiaowei said: "The so-called whole disease does not mean that we can analyze all the diseases of all organs in human beings. But in a given In the image protocol, such as thin-layer plain chest CT, the whole disease can be analyzed not only for one type of lesion, but also for all visible abnormalities in the current scan, compared with a single heavy task. Clinicians often rely on the complaint. And other examination results for the patient to issue an imaging checklist, the purpose is to find out what is the abnormal disease; healthy people participate in the physical examination, the purpose is to find the potential risk of unknown and no feeling in the body. These two groups are not making disease assumptions. In the case of the fundus, for example, an ophthalmology or endocrinologist can see a lot of information from a fundus photograph, such as age-related macular degeneration, suspicious glaucoma, and sugar net disease. However, when the patient has symptoms such as decreased vision If the doctor suspects that it is a fundus lesion, it is necessary to determine what kind of lesion has occurred in the fundus through fundus examination. AI-assisted products naturally need a product that can detect all visible abnormalities in the fundus, rather than selecting a product that can only remind the disease of the sugar net; on the other hand, patients may have multiple diseases, and the detection of a single disease may be distorted. Doctor's evaluation."
Voxel Technology Retina Whole Disease Solution
In the future, Voxel Technology will enable AI products to comprehensively analyze the patient's image in response to the needs of the doctor, so that the image can play its true value.
*The picture in the text is provided by the respondent.
ABS Flood Control Damper, Flood Defenses, Doorway Flood Barrier, Devices For Flood Prevention, Aluminum flood barrier
Denilco Environmental technology(Suzhou)Co., Ltd. , https://www.wflood.com