Xinhua News Agency, Nanjing, February 24th (Reporter Wang Yan) The reporter learned from the Suzhou Medical Institute of the Chinese Academy of Sciences on the 24th that researchers from the institute cooperated with 8 hospitals such as the Jinshan Hospital of Fudan University to develop a new non-invasive ovarian cancer diagnosis method. Using this method, medical personnel can accurately determine the classification of ovarian cancer using only the patient's MRI images.
This time, the research team cooperated with eight tertiary hospitals in East China, South China, and North China to take the lead in applying artificial intelligence technology to the imaging diagnosis of ovarian cancer patients. Researchers collected MRI data of 501 ovarian cancer patients in two major categories, screened features and constructed models through machine learning methods, and finally formed a new set of intelligent non-invasive diagnostic methods for ovarian cancer. Comparing the machine diagnosis conclusions obtained by the new method with the diagnoses of 6 imaging doctors who have been in the business for 2 to 13 years, the results show that the average diagnosis accuracy of the doctor is 79.5%, and the average diagnosis accuracy of the machine is 91.7%. The diagnostic accuracy of the new method is significantly higher than that of manual.
"The combination of artificial intelligence and imaging diagnosis has changed the thinking of manual interpretation of the radiographs in the past. The new method can not only assist doctors in clinical practice, improve the accuracy of ovarian cancer diagnosis, but also suggest which elements are most valuable for the diagnosis. Help doctors improve the efficiency of imaging examinations. "Said Gao Xin, the research leader and researcher at the Suzhou Medical Institute of the Chinese Academy of Sciences.
Related results were published in the Journal of Magnetic Resonance Imaging, an authoritative journal in the field of radiology in February.