SubCMap: Subject and Condition Specific Effect Maps
share
Most widely used statistical analysis methods for neuroimaging data identify condition related structural alterations in the human brain by detecting group differences. These methods can construct detailed maps showing population-wide changes due to a given condition of interest using the data of an appropriate cohort. Although very useful, they are not designed to identify subject-specific structural alterations in a diagnostic fashion, i.e. for a new subject without information on the condition's presence. Existing methods, such as multivariate predictive models, can be modified for this task, however, modifications can fall short in detection accuracy due to the design of the original methods. In this article, we propose SubCMap, a novel method to detect subject and condition specific structural alterations. SubCMap is designed to work without information on condition presence in order to provide diagnostic value. Its detections are condition-specific and can be used to study the effects of various conditions. The method combines techniques from predictive modeling and image restoration. It is univariate in nature and thus allows localised interpretations. Experimental evaluation is performed on synthetically generated data as well as data from the ADNI database. Results on synthetic data demonstrate the advantages of SubCMap compared to population-wide techniques and higher detection accuracy compared to outlier detection and other alternatives based on predictive modeling. Analysis with the ADNI dataset show that SubCMap detections on cortical thickness data well correlate with non-imaging markers of Alzheimer's Disease (AD), the Mini Mental State Examination Score and Cerebrospinal Fluid amyloid-beta levels. Furthermore, repeatability experiments on the ADNI dataset show that SubCMap detections are repeatable across longitudinal images to a large extent.
READ FULL TEXT
