Using neuroimaging to classify aphasia

Fridriksson, Julius and Fridriksson, Astrid and Yourganov, Grigori and Rorden, Chris (2014) Using neuroimaging to classify aphasia. [Clinical Aphasiology Paper]

PDF 510-939-1-RV(Fridriksson-Fridriksson-Yourganov).pdf Download (38kB)

Abstract

The proper classification of aphasia based on clinical symptoms has been debated for well over a century. Much of the early debates centered on relating localized brain damage to a constellation of speech and language impairments. The premise behind much of this work was based on the notion that lesion-symptom mapping could reveal how language was organized in the brain (Broca, 1861, 1865; Dejarine, 1906; Marie, 1906). Although the principle for classifying aphasia based on specific symptoms has been fervently challenged (e.g. Head, 1926) it is still customary to report aphasia types in clinical studies of aphasia. Similar symptoms in sub-groups of patients suggests a similar pattern of brain damage. Nevertheless, it remains unclear if specific aphasia types can be diagnosed simply based on the location of cortical damage. One way to examine this issue would be to relate lesion patterns to aphasia types using multivariate pattern analysis (MVPA). MVPA of neuroimaging data has been successfully used to diagnose diseases such as dementia, schizophrenia, and Parkinson's disease (Orru et al., 2012). In the present study, we demonstrate how MVPA can be used to predict aphasia type in persons with chronic stroke. Unlike previous studies that perform the analysis on voxels (using MRI scans), we trained a classifier on the proportional damage to brain areas (defined with a brain atlas). In addition, we computed the loadings that reflect the contribution of each brain area to classification.

Actions (login required)

View Item