Semi-automatic Epileptic Hot Spot Detection in ECD brain SPECT images

A method is proposed to process ECD brain SPECT images representing epileptic hot spots inside the brain. For validation 35 ictal interictal patient image data were processed. The images were registered by a normalized mutual information method, then the separation of the suspicious and normal brain areas were performed by two thresholdbased segmentations. Normalization between the images was performed by local normal brain mean values. Based on the validation made by two medical physicians, minimal human intervention in the segmentation parameters was necessary to detect all epileptic spots and minimize the number of false spots inside the brain.

• Type: Conference
• Year: 2009
• Where: BILDMED
• Authors: Laszlo Papp, Maaz Zuhayra, Eberhard Henze
• Keywords: BILDMED 2007, BILDMED 2009, Ecd Brain Spect, Epileptic Hot Spots, Normal Brain