Applied Mathematics & Information Sciences

Author Country (or Countries)



An automatic method is presented to extract edema around spontaneous intracerebral hemorrhage (SICH). A new way to cluster edema based on region growing is proposed, with seeds derived from expectation-maximization algorithm, local grayscale mean derived from adaptive local thresholding with varied window sizes, and growing rules that combines local grayscale mean and grayscale information in the form of two-dimensional entropy. The algorithm has been validated on 36 patient datasets to achieve a Dice coefficient of 0.79 in less than 3 minutes. It may provide a potential tool for neurosurgeons to quantify edema and guide therapy of patients with SICH.

Suggested Reviewers