Background and purpose: As magnetic resonance imaging (MRI) signs of normal pressure hydrocephalus (NPH) may precede clinical symptoms we sought to evaluate an algorithm that automatically detects this pattern.
Methods: A support vector machine (SVM) was trained in 30 NPH patients treated with ventriculoperitoneal shunts and 30 healthy controls. For comparison, four neuroradiologists visually assessed sagittal MPRAGE images and graded them as no NPH pattern, possible NPH pattern, or definite NPH pattern.
Results: Human accuracy to visually detect a NPH was between 0.85 and 0.97. Interobserver agreement was substantial (κ = 0.656). Accuracy of the SVM algorithm was 0.93 and AUROC 0.99. Among 272 prespecified regions, gray matter and CSF volumes of both caudate, the right parietal operculum, the left basal forebrain, and the 4th ventricle showed the highest discriminative power to separate a NPH and a no NPH pattern.
Conclusion: A NPH pattern can be reliably detected using a support vector machine (SVM). Its role in the work-up of asymptomatic patients or neurodegenerative disease has to be evaluated.
Keywords: Artificial intelligence; CSF shunt; Machine learning; Normal pressure hydrocephalus; Support vector machine.
© 2021. The Author(s).