Abstract #1012

Brummer M, Sun B, Giddens D, Oshinski J

We have developed a novel MAGIC-FCM algorithm that mitigates tissue misclassification due to intensity variations in traditional FCM algorithm. This model takes into consideration of both spatial constraint and inhomogeneity field characteristics when performing the classification. The proposed method can be used implicitly or explicitly to achieved the improved classification comparing to traditional FCM technique. It effectively suppresses the intensity variations in medical images and can be easily implemented. Improved segmentation results were achieved using the proposed algorithm with fixed parameters compared to uncorrected classification or correction by existing methods.