Implementation and Test of B.R.A.S.I.L: An Epilepsy Computer-Aided Diagnosis Toolkit

It may be difficult to detect the epileptic focus through a simple visual interpretation of ictal and periictal SPECT images, placed side-by-side, especially in patients with extra-temporal or non-localizable intractable epilepsy. Surgery is an important option in the treatment of these epileptic patients. Our methodology aids planning the surgical process. It is a variation of SISCOM methodology. The ictal and periictal images are subtracted, after the registration with MR. The next step is calculate the mean and standard deviation in subtracted SPECT image, the voxel values below the mean plus two standard deviations are ignored (zeroed). We developed 400 phantoms to test this methodology with 4 spheres that suffered alteration in theirs voxels values. Alterations had varied from 5% up to 25%, with steps of 5 units. Results have shown that our method had 71.60% and 93.16% of average sensitivity in alterations of 5% and 25%, respectively. Moreover, our technique has been preliminary validated by clinical and surgical outcome of epileptic patients and with other neurological disorders and the results corroborate the potential of the methodology in clinical routine.