Artificial spine fractures: detection with helical and conventional CT.

Abstract

To compare the accuracy of detection of artificial spine fractures with helical computed tomography (CT) versus conventional CT. Twenty-six motion segments from 15 human cadaver thoracolumbar spines were imaged with helical and conventional CT both before and after fractures were artificially induced. The vertebrae were scanned with different collimations, reconstruction indexes, and exposure doses. A total of 900 images were analyzed by four radiologists, and a receiver operating characteristic (ROC) analysis was performed. ROC analysis showed a large area under the curve for conventional CT (0.913) than for helical CT (0.844) when 3-mm collimation was used. The ability to detect fractures with helical CT increased when collimation was decreased and exposure dose was increased. Conventional CT allows more accurate detection of artificial spine fractures than does helical CT. Helical CT requires thinner collimation for fracture detection comparable with that of conventional CT.