For the giants who started with “hardware”, their core competence is still digital signal processing, and there is no need to spend a lot of manpower and material resources to develop a new software department, so working with Medical AI has become their best choice. At the same time, it not only enriches the function of the product, but also increases the competitiveness of the equipment itself.
Philips and Burke Medical Cooperation ISD is only a starting point, I believe that in the future there will be more device manufacturers and medical AI companies to work together, so that equipment manufacturers involved in more medical scenarios to meet different needs.
Robert Axelrod, the member of National Academy of Sciences (NAS) and well-known expert of behavioral analysis and game theory, once said in his book The Evolution of Cooperation, “What really makes you strong is not to conquer others, but to fully guide the cooperation.”
This word is also applicable for Philips and Perception Vision Medical Technology (PVmed).
After 2016, Philips Greater China officially carried out a digital transformation, focusing on “health technology”. To this end, Philips has made a range of choices. For business activity, it determined to abandon ones that they were not good at, and to invest more in the field of expertise to become bigger and more professional; for medical industry, it decided to withdraw from the inferior field, and to focus on medical AI with accumulated experience.
In 2017, just one year later, Philips released IntelliSpace Discovery 2.0 (ISD2.0), the AI platform, at the RSNA. It serves as a platform and research tool for radiologists, and in a word, it is the “App Store” in medical AI.
It is worth noting that the first product in this “app store” came from a startup company, PVmed. The automatic contouring system for nasopharyngeal carcinoma (NPC) radiotherapy which based on the technology of CT and MR, is the only commercial and mature AI one that examined strictly by top medical device manufacturers.
Smooth Cooperation in Only One Month
In May 2017, Lu Yao and his college schoolmate Shen Shuo founded PVmed.
Lu Yao once served as a researcher at the University of Michigan Medical School. He returned to China due to “The Recruitment Program of Global Experts” of the Central Organization Department in 2013 and was employed as the professor in the School of Data and Computer Science and director of the Institute of Data Science at the Sun Yat-Sen University.
It only took them over one month to embed the software into ISD platform since the first time they contacted each other. And that was the first cooperation between two teams.
Lu Yao recalled 4 months after the establishment of PVmed, Zhou Zhenyu, the senior director of the Clinical Science Department of Philips Total Solution Center, went to Shenzhen for a business trip, and then he visited Guangzhou PVmed by the way. Before the visit, Philips and Sun Yat-sen University Cancer Center had signed an agreement on comprehensive strategic cooperation to jointly advance cutting-edge researches and clinical practices from imaging diagnosis to treatment planning.
After the introduction to automatic contouring system for NPC radiotherapy, Frans Van Houton, Global CEO of Philips, He Guowei, CEO of Philips Greater China, and Zhou Zhenyu fully affirmed the promising future of the integration of PVmed with Philips ISD platform. The conversation lasted for two hours, in which Lu Yao and Zhou Zhenyu shared PVmed products and their prospects.
Looking back, Zhou Zhenyu said at that time, there were more than 100 AI medical startups in China. The reasons why PVmed was the first one selected by Philips:
First, PVmed technologies are produced from profound cooperation with Sun Yat-sen University Cancer Center, which owns Chinese top radiotherapy department.
Second, PVmed is special. It did not choose to research pulmonary nodule, sugar diabetic retinopathy and other fields as other AI medical startups did, but focused on a more difficult filed, automatic contouring system of NPC radiotherapy.
Third, the application scenario is particularly efficient, shortening Intelligent Radiotherapy Contouring Solution (IRCS) for specialist from 3-5 hours to tens of seconds.
Taken all of these into full consideration, Philips started to coordinate with Guangzhou PVmed to embed this product into ISD platform after the Chinese National Day (1, Oct.).
Communication is the most time-consuming. The team responsible for ISD platform interface was located in Germany and there was time difference of six hours between two teams. Besides, the pre-release version of ISD platform and its interface was not mature enough. In order to get through the interface between two systems, more time was required for teams on both sides to get the details done.
Lu Yao said that this was not easy, but it had to be done in the early stage for a startup. “After all, the only way to survive in this competition is being more active and vigorous.”
A month later, at the end of November 2017, Philips officially released ISD platform at the North American Radiology Annual Conference (RSNA). PVmed then successfully became the first supplier on the platform.
Is Platformization a Necessary Road for Medical Device Supplier?
Why did Philips choose to work with Medical AI company? In fact, this is a microcosm of the trend that the entire medical device industry is seeking for transformation.
In recent years, “fusion of the soft and the hard” has become the key word for device manufacturers. At the CMEF, GE, Philips, Siemens and other industry leaders invariably put forward the soft — “platformization”, besides displaying the hard — core products.
For example, “Edison” platform of GE Healthcare enables seamless integration with existing AI partners and related products. Based on “Edison” platform, clinical partners will develop algorithms, and technology partners can apply the latest data processing to applications and smart devices based on the platform.
Siemens Healthineers is the subsidiary of Siemens, a German electrical and electronic engineering company, has more than 170 years of history. At the CMEF booth this year, Siemens Healthineers reserved a place for two Chinese AI partners of Team Play of its medical digital ecosystem.
For example, it has worked with QED Technique to advance the research and implementation of its AI-aided diagnosis on Alzheimer’s disease (AD). The AI image recognition-related products for lung nodules, cerebral stroke, breast cancer and other diseases, launched in cooperation with Deepwise, also appeared at the exhibition.
Lu Yao said that, in fact, it was easy to understand transformation ideas of device manufacturers. “Device manufacturers provide services in a platform-based manner to meet soft needs of users. Clinical training, scientific research will be the focus of their development.”
Business giants starting from selling “hardware” still make most of their profits from this now, and their software sectors are relatively weak. With increasing demands for precision and personalization, it is impossible for software sectors of one equipment manufacturer to meet all the needs of the hospital.
Take ISD platform as an example. ISD was designed to solve haunting pain points in medical imaging research. There are some long-standing problems such as prognostic analysis and self-adaptive radiotherapy in radiotherapy. However, it is difficult for doctors to transform them into scientific research achievements.
This is because doctors, poor at programming and software development, can not immediately turn their precious experience and abundant data into new products or applications.
What makes medical AI companies valuable is that they help medical equipment manufacturers meet clinical needs in a more flexible, fast and professional way. “Medical AI startups can be wholeheartedly committed to a very small field of technology, which is hardly seen in equipment makers. Second, the use of platform can expand the product line of equipment makers at low cost and popularize their products.”
We know that PVmed has two main product lines: Intelligent Radiotherapy Contouring Solution (IRCS) and Intelligent Medical Image Processing Solution with a series of radiotherapy tools embedded, covering automatic segmentation, target volume contouring, data annotation, radiomics analysis and quantitative analysis.
For example, PVmed’s intelligent contouring system provides multiple automatic contouring and modifying tools for various radiotherapy target volumes and organs at risk (OAR). The contouring covers both gross target volume (GTV) and clinical target volume (CTV), which was greatly reliant on doctors’ clinical experience previously. Now, products from PVmed can finish the contouring automatically with accuracy no less than 80 percent. With these products, doctors only need to do some modifications before putting it into practical use.
OAR contouring was a labor-intensive and time-consuming process. But, now, PVmed can finish the automatic contouring of up to 43 OARs in NPC radiotherapy within 20 seconds, which reduces the contouring time from 1 hours to 10 minutes (modification time included).
These tools not only complement the software department of traditional medical equipment makers, but also speed up the launch and operation of their products.
ISD was first applied in The First Bethune hospital of Jilin University in August 2018, followed by Peking Union Medical College Hospital, Beijing TianTan Hospital Affiliated to Capital Medical University, Xuanwu Hospital Affiliated to Capital Medical University and West China Hospital Affiliated to Sichuan University. It is predicted that more than 50 hospitals in China will be equipped with PVmed products by 2021.
In the long run, the ISD platform with different applications promises Philips a brighter future. It can incubate and sell new technologies and products, which will undoubtedly add more possibilities to the single hardware sales mode of medical device manufacturers.
All this can only be achieved through the cooperation of Philips and other medical AI companies. After all, it is difficult to maintain a “App Store” without “Apps”.
According to Leiphone.com, the application software derived from the two solutions of PVmed will be released on the ISD platform this year.
PVmed also welcomes partnership with other equipment manufacturers, and it will offer imaging AI technology services to medical service providers.
Lu Yao said in an interview, “We hope that the counterpart will have a good ability of technology fusion and channel expansion. In addition, the agreements we have signed with partners are all non-exclusive, so there are a large number of manufactures we can cooperate with, maintaining a balanced ecosystem with mutual benefits.
Earlier, in an interview with Leiphone.com, He Guowei, CEO of Philips Greater China, said that Philips is open to the world at multiple levels, that is, it seeks cooperation with institutions in the entire medical process.
Because of the particularity of the medical industry, the AI-centered medical ecosystem needs to cover every link of the whole industrial chain, including a large number of subdivision areas and application scenarios.
In other words, because equipment manufacturers are unable to get involved in all scenarios, platformization and open cooperation are necessary to their further development.
PVmed, as a medical AI enterprise, with high-quality data and evolving algorithms at its core, will join hands with traditional medical device manufacturers and medical service providers to form a new ecological and sustainable structure in medical industry.