Japan: An artificial intelligence (AI) algorithm can improve fracture detection on whole-body trauma computed tomography (CT) exams performed on patients in emergency departments, researchers state in a recent study published in Scientific Reports.
Following training in a convolutional neural network (CNN), the researchers found that it remarkably increased the sensitivity of orthopedic surgeons for detecting rib, pelvic, and spine fractures. CNN model application may reduce missed fractures from whole-body CT images, lead to faster workflows, and improve patient care through efficient diagnosis in polytrauma patients.
The emergency department is an environment with a possible risk for diagnostic errors during trauma care, specifically for fractures. CNN deep learning methods are used widely in medicine as they improve diagnostic accuracy and efficiency and decrease misinterpretation. The research team investigated whether automatic localization and classification utilizing CNN could be applied to the spine, rib, and pelvic fractures. They also examined if this fracture detection algorithm could be helpful for physicians in diagnosing fractures.
For this purpose, Satoshi Maki, Center for Frontier Medical Engineering, Chiba University, Chiba, Japan, and colleagues used a total of 7664 whole-body CT axial slices (pelvis, abdomen, chest) from 200 patients. The calculation of precision, sensitivity, and F1-score were done to evaluate the CNN model's performance.
The study's key findings were as follows:
For the grouped mean values for pelvic, spine, or rib fractures, the sensitivity was 0.786, precision was 0.648, and F1-score was 0.711.
With the assistance of the CNN model, surgeons showed improved sensitivity for detecting fractures, and the time of reading and interpreting CT scans was reduced, especially for less experienced orthopedic surgeons.
AI also sharply decreased the time to diagnosis, dropping from 278.4 seconds to 162.3 seconds for surgeon 1, from 205.2 seconds to 144.5 seconds for surgeon 2, and from 233.7 seconds to 155.5 seconds for surgeon 3. All differences were statistically significant.
"CNN model application reduced missed fractures from whole-body CT images, led to faster workflows, and improves patient care through efficient diagnosis in polytrauma patients," the researchers wrote in their conclusion.
They noted that CNN could serve as a triage system in a busy emergency department.
"If CNN can diagnose fracture before orthopedic surgeons or radiologists have a prospect for image reviewal, the suspected fracture could become a high priority in the worklist," the researchers wrote. "If the interpretation of images can be prioritized with potentially positive findings, diagnosis delays can be minimized, therefore improving patient care."
Reference:
Inoue, T., Maki, S., Furuya, T. et al. Automated fracture screening using an object detection algorithm on whole-body trauma computed tomography. Sci Rep 12, 16549 (2022). https://doi.org/10.1038/s41598-022-20996-w