Your browser does not support JavaScript!



[2019/06/03] Recognition by International Journals: Professor Chung Directs Doctoral Students Who have Developed an AI Disease Diagnosis System

Professor Chung from the NYMU Institute of Biomedical Informatics (left), doctoral Student Cheng (middle) and Dr. Liao from the Chang Gung University AI Research Center 


Professor Chung from the National Yang-Ming University Institute of Biomedical Informatics, his student Dr. Cheng at the Chang Gung Memorial Hospital, Linkou (Trauma and Emergency Department), and Dr. Liao of the Chang Gung AI Research Center have developed a “Pelvic cavity X ray femoral neck fracture detection model” that uses a deep learning convolutional neural network; this will benefit doctors when diagnosing fractures. The results emphasize “the importing of AI/deep learning into disease diagnosis and hospital systems” and has been published in the journal European Radiology (IF=4.027, rank 20/129). This is also the first time that Taiwan has published a paper on AI and a disease diagnosis system. 


Femoral neck fractures are one of neglected areas of fracture diagnosis and these types of injuries occur after an accident. In order to assist doctors with the diagnosis of femoral neck fractures and reduce the burden on radiologists, Dr. Cheng identified a group of radiologists, surgeons, orthopedic surgeons and emergency physicians who were willing to uses their ability to read X ray results and then used the resulting information to produce a ROC curve (see below) for analysis; this was then used to assess the efficacy of the AI model system. The detection model was found to have a very similar effectiveness to assessment by a doctor and thus should be able to help doctors with the diagnosis of this type of fractures. 


The “Femoral neck fracture detection model” compared to the ability of doctors to read x-ray scans (blue dots: radiologists; green dots: surgeons; yellow dots: orthopedics; red dots: trauma and emergency doctors; black line: model)


 Prof. Chung remarked that AI has value when making quick predictions based on unknowns using the data that is available and its present application in medical imaging involves this approach. The “Femoral neck fracture detection model” is being used in hospitals and is being patented. The practical application of this AI system makes it an excellent computer assistant when carrying out a diagnosis after accidental injury. Doctors in the trauma and emergency department will be able to diagnose based on the predictions of the detection model and inform patients that they might have a fracture. This will result in the best treatment being available as early as possible, which will greatly improve the quality of the treatment provided.


 With the help of the AI model, doctors no longer need to wait for a radiologist to issue a formal report and are able to make an immediate diagnosis. This will also reduce the incidence of radiologists missing a fracture, which obviously will be helpful clinically.


The model established by Dr. Cheng and Prof. Chung laboratory is only able to be used to detect femoral neck fractures, but these researchers are hoping to improve its imaging interpretation ability, which broaden the scope of the model and further improve the quality of medical care in Taiwan.