The Hematology Hospital of the Chinese Academy of Medical Sciences is the world’s top hematology specialist hospital. Every day, the hospital is crowded with patients from all over the country who come to seek medical treatment. DeepcytoCEO Wang Zhigang, as the father of a child with leukemia, has a personal experience of this scene. He said: “Due to the difficulty in diagnosis of blood diseases, the lack of primary medical resources and the long time required for leukemia treatment, many patients from other places have to leave home for long-term treatment, which brings a heavy burden of disease.”
Hematological pathology is a relatively independent subject and a high-precision project of hospitals. At present, only more than 800 hospitals in the country have the ability to carry out blood pathological diagnosis. Moreover, due to the complex classification of blood diseases and the difficulty of pathological diagnosis, even doctors in large hospitals are faced with the problems of low accuracy and slow efficiency. At the grassroots level, blood pathological diagnosis is almost a blank. my country urgently needs to improve the accuracy and efficiency of blood pathological diagnosis and promote the sinking of high-quality medical resources.
Wang Zhigang believes that AI technology is a breakthrough point for improving the efficiency of blood pathological diagnosis and alleviating the problem of tight medical resources. This is also the entry point for Wang Zhigang to establish Deepcyto.
Wang Zhigang holds a bachelor’s degree in engineering mechanics and a master’s degree in computational fluid mechanics from Tsinghua University, and a master’s degree in computer and aerospace engineering from Pennsylvania State University. He has served as a senior R&D engineer and management position in famous companies such as IBM, Cadence, Mentor Graphics, and Siemens for many years. He also founded technology The company BioCAX cooperates with NIH to develop medical images, human 3D scanning imaging and smart wearable devices.
In 2016, with the vision of “realizing the intelligent comprehensive diagnosis of blood pathology and promoting the sinking of high-quality medical resources”, Wang Zhigang began to develop blood pathology AI diagnostic products in the form of scientific research cooperation. “At that time, we estimated that the accuracy rate could reach 70%-80%, but we didn’t expect the accuracy rate to exceed 90%.” Wang Zhigang was greatly encouraged by this. He returned to China in 2018 and founded Deepcyto, focusing on AI comprehensive diagnosis of blood pathology. Soon after its establishment, the company received angel round investment from SoftBank China and Yuansheng Venture Capital. In 2019, the company completed the pre-A round of financing for Northern Light Venture Capital.
Based on the world’s largest integrated MICM database of blood pathology, Deepcyto integrates machine learning, deep learning and big data mining technologies, and has launched a variety of AI solutions for blood cell morphology diagnosis, flow cytometry diagnosis, and cytogenetic karyotype analysis. Among them, the cell morphology AI diagnostic product Deepcell can identify more than 40 common cell morphologies in leukemia, lymphoma and other diseases, quickly and accurately interpret cell morphology, with an accuracy rate of 97.5%, and a 90% reduction in interpretation time; DeepFlow It is the world’s first flow cytometry AI cloud diagnosis system developed by Deepcyto. The accuracy of the diagnosis of acute leukemia is as high as 95%, which is about 100 times faster than manual diagnosis. DeepKaryo is the first domestic chromosome karyotype AI analysis system that can automate the entire process from chromosome scanning, analysis to report issuance.
Complete cell morphology scan in 5 minutes, which can process peripheral blood and bone marrow samples at the same time
Cell morphology is the gold standard for diagnosing blood diseases, but the diagnosis of cell morphology in my country is in a very “primitive” state. It depends on manual diagnosis. Each case requires manual classification and counting of 200-500 cells. Manual adjustment and manual counting are required during the counting process. , The labor intensity is large, the subjective deviation of personnel is large, and the repeatability and testability are poor. Moreover, due to the long training cycle of cell morphological diagnosis doctors and many years of experience, my country is facing a huge gap in cell morphological diagnosis personnel.
The DCS-1000 developed by Deepcyto is currently the only fully automated cell morphology microscopic scanning system in the country that can process peripheral blood and bone marrow samples at the same time. It can automatically scan and analyze peripheral blood and bone marrow cells to achieve double-precision scanning and fully automatic scanning. The film is dripped with oil, and the entire film is scanned within 5 minutes. The supporting DeepCell artificial intelligence cell morphology analysis software can automatically issue a report.
Deepcyto is cooperating with the Hematology Hospital of the Chinese Academy of Medical Sciences, Peking University Hospital, Jinyu Medical, Beijing Chaoyang Hospital, etc. for clinical verification. The data shows that DCS-1000 can be used with DeepCell to identify more than 40 kinds of cell morphologies that are common in leukemia, lymphoma and other diseases, quickly and accurately interpret the cell morphology, with an accuracy rate of 97.5%, and a 90% reduction in the interpretation time compared to the same period last year.
According to Wang Zhigang, DCS-1000 will soon obtain a Class II certificate in China and is preparing to submit a 510(k) certification application to the US FDA. “It is worth noting that DCS-1000 does not need to apply for new charging items and can be quickly put into the market after approval.”
The world’s first flow cytometry AI diagnosis platform, which can complete the analysis in 5-10 seconds
Flow cytometry is a commonly used technique for blood pathological diagnosis, and it is also the focus of Deepcyto’s layout. Wang Zhigang said: “There is no shortage of instruments and reagents for clinical diagnosis of flow cytometry, but people with data analysis capabilities are lacking. Therefore, Deepcyto’s AI analysis software will greatly expand the application of flow cytometry in clinical diagnosis.”
Flow cytometry is a high-throughput technology that has the advantages of simple operation, rapid analysis, objective accuracy and so on. However, the diagnosis of cell morphology faces a similar dilemma, and the analysis of flow cytometry data in my country still relies on manual labor. Manual analysis has large subjective factors, low sensitivity, and is easy to miss a very low density of residual cancer cells after chemotherapy. In addition, manual analysis is time-consuming and labor-intensive, usually more than 10 minutes or even half an hour. Taking leukemia as an example, leukemia immunoassays need to analyze 30-40 markers and as many as one million cells, and the manual analysis workload is huge. In addition, the learning curve for manual analysis of flow cytometry data is long, and there is an extreme shortage of talents with flow cytometry analysis capabilities in my country.
DeepFlow is the world’s first flow cytology AI fully automated diagnostic platform developed by Deepcyto. It is compatible with all mainstream flow cytometers, provides automated flow cytology data cloud management, storage and diagnostic workflow, and automatically determines common blood diseases and Immune abnormalities, it only takes 5-10 seconds to complete the analysis. It not only eliminates the subjective error introduced by manual gating interpretation, but also greatly improves the efficiency and accuracy of flow cytometry diagnosis.
Now, DeepFlow has carried out clinical trials in Peking University Hospital, Hematology Hospital of Chinese Academy of Medical Sciences, Zhongshan First Hospital, MD Anderson Cancer Center in the United States, and Oregon Medical College Affiliated Hospital. At the same time, Deepcyto has signed a strategic partnership agreement with Cytek, a US flow cytometer manufacturer, to provide customers with full-process solutions.