This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy
How digital innovation is transforming pathology at Ruijin Hospital
For most diseases, pathological diagnosis is recognized as the gold standard of clinical medicine. If a patient is diagnosed with a tumor, determining whether the tumor is benign or malignant is critical.
Unreliable pathology reports can result in patients being over- or undertreated. For cancer patients in particular, this can be a matter life and death.
Ruijin Hospital, affiliated with Shanghai JiaoTong University School of Medicine, integrates medical care, teaching, and scientific research. Its pathology department receives diagnoses from all over China and livestreams discussions on difficult diagnoses every two weeks.
Dr. Wang Chaofu, the chief of the pathology department of Ruijin Hospital and head of department of pathology of Shanghai Jiao Tong University School of Medicine, is a renowned pathology expert in China.
In 2010, he saw a 32-year-old male initially diagnosed with a benign tumor in his kidney, but upon closer inspection discovered that the tumor was, in fact, malignant. There was no corresponding kidney tumor in the World Health Organization tumor classification. Dr. Wang relied on his experience and intuition to make the correct diagnosis. Doing so made him the first doctor to discover this type of kidney cancer in China, when only 17 cases had been reported worldwide.
Not every patient is lucky enough to come across an expert like Dr. Wang. Many are misdiagnosed or “missed diagnosed,” meaning the doctor fails to spot a disease that is already present.
Now, even patients who don’t have the benefit of Dr. Wang’s experience can be confident that their conditions will be diagnosed correctly.
In February, Ruijin Hospital released a large pathology model that is expected to significantly improve diagnostic efficiency and accuracy, accelerating medical intelligence in pathology diagnosis scenarios.
The model, called Ruipath, is based on Huawei's DCS AI solution, a suite of technologies and services designed to enhance the efficiency, performance, and intelligence of data centers through the integration of artificial intelligence.
Developed over the past three years, RuiPath processes different types of data in areas including clinical, imaging, and pathology. It covers 90% of all types of cancer found in China.
Huawei DCS AI provides the ModelEngine AI end-to-end toolchain, which streamlines the lifecycle of AI model development, from data preparation and model training to deployment and inference, automating the entire process, from data cleaning, labeling, to AI training and optimization. This improves model training efficiency by 70%. The time needed to generate a pathology diagnosis report is shortened from 40 minutes to 10 minutes.
"RuiPath is a clinical-grade multimodal interactive pathology model," said Ning Guang, president of the hospital and academician at the Chinese Academy of Engineering. “It was built on the comprehensive business processes of the digital smart pathology department of Ruijin Hospital.”
This is not the first time digital technology has helped Ruijin improve its patient care by improving the management of diagnostic and other data.
For example, the hospital takes in thousands of specimen slides per day and pathologist doctors need to read them carefully. In addition to being overwhelmed by the sheer volume of work, pathologists must deal with problems related to the storage of medical images. Slides tend to fade and become unusable after several years. Outdated practices such as using handwritten labels compound the problem.
Digital technology promises a solution. Data can be permanently stored in a digital library, solving problems related to the storage, fading, damage, loss, and retrieval of glass slides. Doctors can read any area of a digital slide image at any magnification ratio they need, anytime and anywhere.
But it also faces new challenges. For example, the generation, storage, and reading of massive digital slide images pose challenges to data storage system. The data volume of a single digital slide image is equivalent to that of an HD movie. Tens or even hundreds of thousands of pathological slide images can be generated daily, especially in larger hospitals.
Data must be stored long-term to meet regulatory compliance requirements, burdening the storage resources of hospitals.
Ruijin Hospital chose OceanStor Pacific scale-out storage to build a data lake for pathology, reading 1,000 slide images per second, 80 times faster what was possible using traditional storage methods, compressing the size of the data files up to 45%. The high-density design of OceanStor Pacific scale-out storage enabled the hospital to use less space to store more data, for a longer time, saving 73% space in the hospital equipment room.
Through long-term technological innovation, Huawei integrates technologies with healthcare applications, making breakthroughs in fundamental digital pathology technologies and streamlining the last mile of hospital digital transformation. Its work with Ruijin Hospital demonstrates the company’s commitment to saving lives by providing innovative technology.
