Written by
Share article:
To date, most of the academic research and commercial activity in the field of machine learning for medical imaging has been driven by the US. Several of the leading US academic hospitals have established teams to develop and commercialise artificial intelligence solutions for the detection and diagnosis of diseases. In the commercial world, around half of the nearly $200 million of venture capital invested in medical imaging AI start-ups has gone to US companies, with most of the remainder going to companies in Europe and Israel (click for our analysis of funding for medical imaging AI start-ups). Not to mention the huge investments in healthcare AI by leading healthcare technology vendors, including IBM and GE Healthcare. Moreover, the US was the early adopter of computer-aided detection (CADe) solutions, most notably for use in breast cancer screening, and today accounts for around two-thirds of the global CADe market. Whilst China is late to the party, it is poised to emerge as a leading global player, both in terms of technology development and market demand.
China’s Internet Giants Enter Medical AI
In March of this year, Alibaba Cloud, the public cloud division of Alibaba Group, announced ET Medical Brain, a suite of deep learning solutions for medical imaging, drug development and hospital management. The suite also includes a virtual medical assistant for use by patients. In one project, Alibaba Cloud partnered with a hospital in Zhejiang province to automatically identify thyroid cancer from ultrasound scans. According to Alibaba Cloud, about 20,000 images were used to train its system, which was 85% accurate in trial tests. The company is also developing a solution to detect lung cancer in chest CT scans.
Alibaba’s healthcare unit, Ali Health, which is primarily involved with the distribution of pharmaceutical and health products in China, has also recently revealed an AI solution for medical imaging, called Doctor You. The initial applications for Doctor You are the early detection of lung cancer and diabetes. Ali Health entered the medical imaging market in March 2016 through a $35 million investment to acquire a 25% stake in Wan Li Yun Medical Information Technology (majority owned by Wandong Medical Equipment), which markets a cloud-based medical imaging platform called Wlycloud. Ali Health claims over 1,600 hospitals are using the Wlycloud platform, which gives the company an effective route to market for its Doctor You solution.
In August of this year, Tencent, one of the world’s largest internet providers and gaming companies, announced it had developed a deep learning solution for the detection of oesophageal cancer. The company also plans to develop solutions for the detection of lung cancer, breast cancer, diabetic retinopathy and other diseases. However, Tencent has stated that its medical AI solutions are not yet ready for commercialisation.
Chinese Medical Imaging AI Start-ups
Worldwide there are over 50 start-up companies developing machine learning solutions for medical imaging, most of which are based in the US, Europe and Israel. In 2016, the first Chinese start-ups entered the market – DeepCare and Infervision. DeepCare has raised nearly $1 million in angel funding and is developing deep learning solutions for medical device OEMs. Infervision raised $7.2 million in a Series A funding round led by Sequoia Capital and has developed solutions to detect lung cancer on CT scans and cardiothoracic lesions on x-ray scans. The company claims to have established partnerships with around 20 tier 3 hospitals in China, including Peking Union Medical College Hospital and Shanghai Changzheng Hospital.
Chinese Research Initiatives
In July of this year, the National Clinical Research Center for Cancer (NCRCC) announced it had signed a deal with the Institute of Computing Technology, part of the Chinese Academy of Sciences (CAS), to develop artificial intelligence solutions for medical imaging. The initial focus is to develop solutions to read ultrasound breast scans and mammograms. The aim of the agreement is to improve diagnosis accuracy and encourage breast cancer screening in regions of high prevalence and in rural areas, where experienced medical professionals are in short supply.
Chinese Market Demand
The world’s most populous nation suffers from a shortage of skilled medical professionals such as radiologists and oncologists, particularly in rural areas. According to World Health Organization (WHO) there are only 1.5 doctors for every 1,000 people in China, notably lower than many other countries (e.g. 2.5 per 1,000 in the US). Moreover, cancer rates in China are on the increase and last year more than four million Chinese people were diagnosed with the disease. Cancer has been the leading cause of death in China since 2010, with lung cancer causing the most deaths. Hardly surprising given the high prevalence of smoking in China – more than half of Chinese men are smokers. The incidence of breast cancer in China has more than doubled over the last 30 years and continues to rise. With China’s population ageing at a faster rate than the global average, the burden of cancer on China’s healthcare system is expected to grow in the coming years.
The China National Lung Cancer Screening Guideline recommends annual lung cancer screening with LDCT for high risk individuals aged 50-74 years who have at least a 20 pack a year smoking history and who currently smoke or have quit within the past 5 years. However, the cost of population-based screening is prohibitive, not to mention the shortage of healthcare professionals. This is where AI has a part to play. In recent years the Chinese government has initiated several large-scale lung cancer screening programs. As far as Signify Research is aware, computer-aided detection was not used, but as screening programs are scaled-up to cover larger proportions of the Chinese population, the case to use the technology will grow stronger. Although details of future lung cancer screening initiatives have not been announced, there will likely be more large-scale, regional and multi-site screening programs in the coming years.
Between 2008 and 2011, over 1.2 million women were enrolled in the Chinese National Breast Cancer Screening Program (CNBCSP). Ultrasound was used for the primary imaging exam and all women with a suspicious lesion found on the ultrasound study underwent an additional mammography exam. Ultrasound is preferred over mammography, particularly in rural areas, as it is more cost-effective and clinically more effective for women with dense breasts. Again, computer-aided detection has not been routinely used to date, but this is expected to change in the coming years, particularly as 3D automated breast imaging becomes more commonly used.
The use of AI in diagnostic imaging will not be limited to cancer screening initiatives. For example, in April of this year, Enlitic announced that is had executed a Memorandum of Understanding with Beijing Hao Yun Dao Information and Technology Co., Ltd (“Paiyipai”) to provide Enlitic’s deep learning solution to Paiyipai for diagnostic imaging in health check centres across China. Approximately 300 million Chinese people use health check services. Most patients receive annual chest x-rays and basic check-up services, with options to include MRIs, CT studies and genetic testing.
No Baggage
Another factor that will accelerate the uptake of medical imaging AI solutions in China is customer acceptance. In the US, many radiologists see AI as a threat or are sceptical of the capabilities of AI solutions based on their experiences with early generation CADe solutions, which suffered from low sensitivity and/or specificity levels. Vendors are likely to encounter less resistance to AI solutions from Chinese healthcare professionals, who have skipped the early generations of machine learning technology and jump directly to the latest deep learning solutions.
The Government is the Gatekeeper
The pace of healthcare reform and the modernisation of the Chinese healthcare system, coupled with increasing domestic research and development activity, would suggest that China will be an early adopter of AI in medical imaging. With over 2,200 tier 3 hospitals (general hospitals in large cities with over 500 beds) and nearly 8,000 tier 2 hospitals (medium-size city, county or district, with up to 500 beds), the market opportunity for medical imaging AI in China is vast. However, ultimately government policy will dictate the market.
Related Report from Signify Research
“Machine Learning in Medical Imaging – 2017 Edition” provides a data-centric and global outlook on the current and projected uptake of machine learning in medical imaging. The report blends primary data collected from in-depth interviews with healthcare professionals and technology vendors, to provide a balanced and objective view of the market.
For further information about purchasing this report please click here or contact simon.harris@signifyresearch.net.