Congratulations Tayem Zgool on the publication of his article “Point-of-Care Ultrasound Imaging for Automated Detection of Abdominal Haemorrhage: A Systematic Review”
Abdominal haemorrhage is one of the most dangerous consequences of traumatic injury. It needs to be detected quickly, but the standard tool, ultrasound, depends heavily on the operator skill, which makes it unreliable outside of hospitals or emergency departments. Medical teams often look at a specific area in the abdomen called Morison’s pouch, where fluid tends to collect, but relying only on this region can mean missing bleeding elsewhere.
Our paper reviewed the latest progress in combining ultrasound with artificial intelligence (AI) to make detection faster, easier, and more reliable. We found seven research studies that tested AI systems on thousands of ultrasound images from trauma patients. These systems, which used advanced computer models like YOLOv3, U-Net, and ResNet50, were able to detect internal bleeding with very high accuracy up to 98% sensitivity and 99% specificity. This suggests that AI could help clinicians, paramedics, and even non-specialist operators use ultrasound effectively in high-pressure situations such as ambulances, battlefield medicine, or rural clinics.
However, we also identified important gaps. Most of the research still relies on 2D ultrasound scans and often tests AI systems only on the same data used for training. Only one study checked its results on patients from a different hospital. To make these systems truly dependable in real-world trauma care, future research needs to move toward 3D imaging and test across multiple hospitals and patient populations. Our review highlights both the promise and the current limitations of AI-driven ultrasound, pointing the way toward technologies that could save lives by enabling earlier detection and treatment of abdominal haemorrhage.
You can find Tayem’s paper here: DOI 10.1016/j.ultrasmedbio.2025.07.024
