Login / Signup

Accuracy in prediction of long-term functional outcome in patients with traumatic axonal injury: a comparison of MRI scales.

Marleen Maria van EijckMartin Willy HerklotsJo PelusoGuus Geurt SchoonmanAnnemarie Wilhelma OldenbeuvingJolanda De VriesJoukje van der NaaltGerwin Roks
Published in: Brain injury (2020)
Purpose: Functional outcome prediction for patients with traumatic axonal injury (TAI) is not highly related to the MRI classifications. The aim of this study was to assess the accuracy in predicting functional outcome in patients with TAI with several MRI scoring methods and to define the most accurate method.Methods: Patients with TAI (2008-2014) confirmed on MRI <6 months after injury were included in this retrospective study. Long-term functional outcome was prospectively assessed using the Glasgow Outcome Score Extended. The Gentry classification is most used in clinical practice. This method was compared to methods that score lesion load, lesion locations, and to modified Gentry classifications. The area under the curve (AUC) was calculated for the scoring methods.Results: A total of 124 patients with TAI were included, medium follow-up 52 months. The AUC for the Gentry classification was 0.64. All tested methods were poor predictors for functional outcome, except for the 6-location score (area under the curve: 0.71). No method was significantly better than the Gentry classification.Conclusion: The Gentry classification for TAI correlates with functional outcome, but is a poor predictor for the long-term functional outcome. None of the other tested methods was significantly better.
Keyphrases
  • spinal cord injury
  • machine learning
  • deep learning
  • magnetic resonance imaging
  • contrast enhanced
  • clinical practice
  • computed tomography
  • high resolution
  • optical coherence tomography