2019 AACR Annual Meeting
Recently, the 110th American Association for Cancer Research (AACR) was held in Atlanta, USA. The American Association for Cancer Research (AACR) is the world's oldest and largest scientific organization dedicated to cancer research. This year's theme covers the latest discoveries in cancer research – from population science and prevention to cancer biology, transformation and clinical research. The survival and advocacy of patients has attracted the attention of experts, scholars and business people all over the world, and they have come to participate in the exchange of new knowledge and new ideas.
(AACR project poster showcase)
AACR focuses on the work and contributions of research and medical elites from institutions around the world: new immunotherapeutics, new targets, new joint programs, numerous big data on various cancers, and emerging new informatics diagnostic techniques It also adds brilliance to the conference... It's just a 1300-page conference summary that's fascinating.
(Intellitronic Technology AACR Conference Summary, Poster Display)
Located in Exhibit Hall B (Permanent Abstract Number: 1629), DM Intelligence Ltd. has attracted many cancer diagnostic intelligence individuals with breast cancer and breast cancer histopathology AI research. The research project demonstrated by Zhirui Technology is based on the theme of "Classification of cancer histology images using deep learning".
At the AACR meeting, Dr. Xie Weidong, the founder of Zhirui Technology, personally participated. In the face of the vast number of scholars who are thirsty for knowledge, Dr. Xie gave a brief speech on the project. Starting from the predicament of cancer pathology, Dr. Xie Weidong gave a brief introduction on the technical points, specific operation plans, results prediction and clinical value of the team's research: cancer histopathological diagnosis is the "gold standard" for judging cancer, but in clinical practice. There is a problem of backward technology and a serious shortage of pathologists. The deep learning algorithm uses digital pathology to predict the degree of carcinogenesis of histopathological slices, which is expected to reduce the burden on pathologists, improve work efficiency and save costs. The project passed the collection, labeling and pretreatment of 45,000 samples of breast cancer and breast cancer in the hospital for the training of the company's algorithm platform. At present, the prediction accuracy of the second classification (normal, benign VS. malignant breast cancer, malignant breast cancer) can reach 99.9% (whole WSI). At present, the company has carried out clinical transformation of the results and accelerated the pace of its application.
"Being a pioneer in the diagnosis of cancer pathology AI." When talking about the original intention of this research project, Dr. Xie Weidong said.
(Dr. Xie Weidong, founder of Zhirui Technology, introduced the poster)
Image analysis of cancer pathology tissue has long been a time-consuming, error-prone and often subjective process. Computational new technologies, especially artificial intelligence algorithms, can solve a series of problems faced by pathology: completely manual operation, using more than 100 years of old technology, has been unable to analyze and process the vast amount of information we can now get from patients. Massive information processing is the unique advantage of computers. The use of computers for pathological detection has become an indispensable helper in cancer diagnosis. In the research of cancer pathology AI recognition, following the "Alphago", Google artificial intelligence for analyzing breast cancer pathological sections can significantly improve the diagnostic accuracy (89%). In 2017, the results of the joint research team of Stanford University published in "Nature" also showed that the accuracy of skin cancer diagnosis is comparable to the artificial intelligence of human doctors. Of course, although the research results in this area exist, the results actually applied to the clinical are rare, especially in the AI-assisted diagnosis of cancer histopathology.
(The poster attracts domestic and foreign experts to stop and communicate)
It is reported that the project is jointly researched and developed by Zhirui Technology and the First Affiliated Hospital of Sun Yat-sen University. The pathologists will scan and mark the data of the hospital case database, and integrate machine vision, deep learning and big data mining technology through the information team. , carry out digital pathological section artificial intelligence analysis and auxiliary diagnosis, and achieve quantitative analysis of cancer. At present, using the team's research and development products and platforms, the pathologist samples the slides, scans by digital pathology scanner and small box analysis (integrated automatic drawing system, pathological recognition system, etc.), and can obtain AI diagnosis opinions on the platform. The pathologist only needs to review and print out the report. For difficult problems, the pathologist can trace all the results, even on the server to obtain WSI (full-field digital slice) for MTD multidisciplinary diagnosis or remote analysis, in order to truly achieve clinical application and market transformation.
(Dr. Xie Weidong and the experts seriously discuss the poster)
Of course, like other automated systems, the main role of this project is to help pathologists make accurate diagnoses instead of replacing them. Pathologists can shift their focus to more complex tasks, such as combining the findings of patient tissue sections with other diagnostic analyses to achieve a more targeted treatment. Artificial intelligence is expected to provide pathologists with a pair of eye-catching eyes to interpret pathological patterns that are invisible to the naked eye.
