Development and implementation of an AI system for detecting fatal active bleeding due to trauma using whole-body search type medical image analysis AI technology
As the population ages and the number of patients increases rapidly, Japan is facing a serious shortage of doctors. Emergency care facilities, with the number of emergency transports increasing dramatically, are already over capacity, and there are patients whose lives are difficult to save when staff is thin outside of normal operating hours. A prime example of this is patients with severe trauma. The leading cause of death in severe trauma patients is blood loss. Therefore, whether or not a patient can be saved depends entirely on whether the bleeding point can be quickly identified and the bleeding can be stopped. In emergency care facilities, it is not possible to determine where the bleeding is coming from just by examining the surface of the body, so a device called a CT scan is used to create detailed images of the inside of the body from head to toe, and based on this, the source of the bleeding is identified. While this CT scan is very useful, it has two major problems. It takes time and can lead to overlooking something. Our goal in this research is to develop an AI system that can quickly find bleeding points from CT images, thereby enabling optimal trauma rescue in any situation, regardless of the time of day or size of the hospital, and realizing a world without preventable trauma deaths (trauma deaths that account for up to 20% of all trauma deaths and are estimated to be saveable with appropriate treatment).
