Shortage of Medical Manpower Draws the Coming of AI in the Field of Mammography.KVGH Works with III to Develop “Mammography AI Assistant”.
Caption: In recent years, III has applied the gradually mature AI model technologies to the detection and identification solution for breast lesions.
The picture shows that artificial intelligence (AI) assists in the process of mammography interpretation.
Breast cancer ranks first in terms of incidence rate among all cancers affecting female citizens in our country. More and more women reaching the age of high incidence would have a mammography. According to statistics from the Health Promotion Administration, Ministry of Health and Welfare, the number of breast cancer screenings in 2018 has exceeded 860,000 tests, which has also caused great pressure on the physicians responsible for the screening reports. In order to reduce the workload of medical staff and provide better quality of medical services, Kaohsiung Veterans General Hospital cooperates with the Digital Transformation Institute (DTI) of Institute for Information Industry in the field of the interpretation of mammography with the help of artificial intelligence (AI). Currently, the artificial intelligence training model testing has been completed and it is expected that, by the end of this year, the number of 3,000 cases verified by AI will be reached.
Kaohsiung Veterans General Hospital, whose high-quality mammograms and standardized structural reports are accumulated for years, has been working with III since 2018. By applying the III-developed medical image marking system, the project team could establish a complete breast disease data set for training artificial intelligence model. At the same time, physicians can also use the medical image marking feedback loop system developed by the III to provide feedback and make corrections to the artificial intelligence model, adjust the preliminary interpretation of the artificial intelligence model, and gradually improve the accuracy of the AI breast screening assistant.
Supported by the special project proposed by the Department of Industrial Technology (DoIT), Ministry of Economic Affairs, in recent years, III has applied the lately mature artificial intelligence model to the detection and identification solutions of breast lesions. Adopting deep learning technologies for development, “Mammography AI Assistant” learns from radiologists to identify suspicious symptoms in mammograms, such as lumps, calcifications, and micro-calcifications and gives scores to each mammogram according to the symptoms listed by the Breast Imaging Reporting and Data System.
Dr. Tseng-Lung, Yang, one of the mammography experts of Kaohsiung Veterans General Hospital and also its Innovation Director of Research and Innovation Center, indicates that about 90% of mammography results are normal negative cases (that is, there are no signs of breast cancer symptoms). In the future, it is expected that the “Mammography AI Assistant” of III will be able to screen out patients at high risk according to their mammograms. Physicians can therefore prioritize their work orders to focus on and interpret these 10% images which appear to be at high risk so that the output of screening reports could be expedited. “Mammography AI Assistant” can detect the location and volume of the tumor within a few seconds, which assists physicians in interpreting images, saves their time, and increases the accuracy of the whole process.
Dr. Tseng-Lung, Yang also points out that deep learning AI technologies play a significant role in various applications of medical image analysis. With the help of AI technologies, the current medical structure could be modified. In addition to improving the accuracy of diagnosis, medical efficiency could also be improved. The technologies could save a lot of time for doctors interpreting the mammograms so that they can focus more on communicating with patients. III “Mammography AI Assistant” can correct errors with the assistance of a physician and relearn several times, which helps the AI assistant continuously improve the accuracy of the interpretation and approach the judgment level of a senior physician.
In the future, with the coming trend of AI clinics, patients will have their mammography at Kaohsiung Veterans General Hospital and all mammograms will be screened by the AI breast lesion detection and identification solution developed by the Institute for Information Industry. This solution is expected to shorten the interpretation time for medical practitioners, speed up the output of screening reports, give physicians more time to explain imaging symptoms, and then help doctors and patients communicate in depth. Through this cooperation case between the medical and technology industry, not only could the burden on physicians be lightened, but also the accuracy of diagnosis could be improved and the quality of medical treatment for the public could be considerably leveled up.
Planning & Promotion Division(III) Joy Yen Tel: (02) 6631-8635 firstname.lastname@example.org
Ying Shen Tel: (02) 6631-8643 email@example.com