Wu, Kaicheng. Image-based screening methods for reducing misdiagnosis of carpometacarpal fracture and dislocation. Retrieved from https://doi.org/doi:10.7282/t3-ep8t-ms25
DescriptionUlnar-sided carpometacarpal joint dislocation has been commonly missed on initial examination. The traditional diagnosis procedure manually determines the injuries based on the angles between the index metacarpal and the fourth or fifth metacarpal on the posteroanterior or lateral radiographs. This procedure is limited since metacarpal bones are difficult to identify due to osseous overlaps. This thesis develops an image-based screening method for reducing misdiagnosis of ulnar-sided carpometacarpal (CMC) fracture and dislocation.
The screening method uses a computer-aided computational framework to calculate our proposed angles for diagnosing purpose. Each step in the computaional framework is strictly defined to ensure high re-producibility and inter- and intra- observer reliability. We develop two similar frameworks for calculating 2D and 3D angles, while 3D version is only for comparison purpose.
We propose the two sets of capiate-based angles for diagnosis purpose: 1) the angle from lateral view; 2) the surrogate spatial angle calculated from lateral view angle and posteroanterior (PA) view angle:
ˆθ = √θ2lateral + θ2PA. We show that by using the computational framework to calculate our proposed angles, the sreening method can effectively diagnose ulnar-sided CMC fracture and dislocation. Moreover, compared to traditional method, it can reduce the confusion associated with carpometacarpal injuries and have better sensitivity and specificity. Notably, the surrogate spatial angle further out-performs the lateral view angle in terms of differentiating between injury and non-injury cases, and the ability to detect extreme cases.