Application of run length matrix to magnetic resonance imaging diagnosis of Alzheimer-type dementia

Abstract

To examine the possibility of diagnosing Alzheimer-type dementia, we studied this condition using the run length matrix, on head MR images of 29 Alzheimer-type dementia patients (8 men, 21 women, 78.7 +/- 6.7 years) and healthy elderly controls (10 men, 19 women, 72.3 +/- 8.7 years) . The results showed that differences in GLN (gray level nonuniformity) and RLN (run length nonuniformity) were statistically significant. Furthermore, discriminant analysis based on GLN and RLN showed a rate of sensitivity of 69.0%, specificity 86.2%, and correct classification 77.6%. Although this rate of correct classification is inferior to the planimetric and volumetric methods, run length matrix is only one method of texture analysis. The results of this study indicate the possibility of MR imaging-based diagnosis of Alzheimer-type dementia with texture analysis including a run length matrix.