Computerized predictive model for differential diagnosis of normal pressure hydrocephalus based on complex analysis of MRI image

Abstract

Background. The main problem is to differentiate idiopathic normal pressure hydrocephalus (iNPH) with other neurodegenerative diseases accompanied by similar clinical presentation. For this purpose, a number of neuroimaging and invasive methods were proposed. The study objective is to assess most reliable magnetic resonance (MR) symptoms of iNPH and to create a prognostic model for differential diagnosis of iNPH by these criteria. Materials and methods. MR-data of 213 iNPH, 144 neurodegenerative (Alzheimer disease, Parkinson disease and early mild cognitive impairment) patients and 79 healthy age-matched controls were retrospectively analyzed. Following criteria were assessed: Evans Index, size of the third ventricle, size of temporal horns, disproportionately enlarged subarachnoid space hydrocephalus, focal convexital sulci enlargement, callosal angle, periventricular hyperintensities, focal subcortical gliosis and perivascular spaces enlargement. One-way ANOVA was used to explore differences between groups. Classification matrix was formed by discriminant analysis. Results. Differential model was established and Excel-tab-calculator was created for complex computed assessment of MR data and differential diagnosis of iNPH patients. Prognostic accuracy of model for iNPH reaches 99.5 %, for degenerative disorders – 89 %, common diagnostic accuracy is 92 %. Conclusion. Advantages of complex approach for assessment of MR-data in iNPH patients were underlined. Prognostic model for prediction of iNPH based on MR criteria was established.