Abstract
Purpose:
To develop a functional MRI (fMRI) signal based model that can evaluate depression severity in a numeric form; therefore, depressed patients can be identified during the course of illness, independent from symptoms.
Materials and Methods:
Data from 20 medication‐free depressed patients and 16 healthy subjects were analyzed. The event‐related fMRI scanning features under sad facial emotional stimuli were extracted as model inputs. Fuzzy logic and a genetic algorithm were used to provide suitable model outputs for numeric estimations of depression.
Results:
The correlation value r between the model estimations and the professional Hamilton Depression Rating Scales (HAMD) was 0.7886 with P < 0.00016. A typical tracking history for a particular subject has also promised the possibility for early disease warning, when the clinal symptoms are ambiguous or recessive.
Conclusion:
A numeric and objective estimation for the course of illness can be provided. The model can be used by psychiatrists to track the recovery process. As a simple extended application, the proposed model can be applied to classify subjects into different patterns: major depression, moderate depression, or healthy. J. Magn. Reson. Imaging 2010;31:1067–1074. © 2010 Wiley‐Liss, Inc.