Measuring complexity of lateral ventricle from MR images for schizophrenia

摘要

This paper mainly contributes to measure the complexity of lateral ventricle's surface of the brain and reveal a lineament for identification of schizophrenia. To measure complexity of surface of lateral ventricle modified box counting was implemented using image processing toolbox of MATLAB. Magnetic Resonance images of 67 subjects were selected and subjected to weighted median filter. Lateral ventricle region was extracted in semi automatically and boundary of segmented region was detected traditionally. Then it was exposed for measurement of fractal dimension using modified box counting method. Satisfied results were obtained. It conveyed that greater complexity is in healthy controls than in patients. The results confirmed significant difference in Fractal Dimension values between patients and controls. However, any significant differences in lateral ventricle's volume or surface area between groups were not found, which might reflect superior sensitivity of fractal dimension measures over more traditional measures like area and volume. It is concluded that diminution of complexity gives an evidence of enlargement of ventricle in schizophrenia.