CLINICAL USEFULNESS OF SUBTRACTION IMAGES OF ICTAL-INTERICTAL SPECT USING STATISTICAL COUNT NORMALIZATION IN TEMPORAL LOBE EPILEPSY
Abstract number :
2.294
Submission category :
Year :
2004
Submission ID :
783
Source :
www.aesnet.org
Presentation date :
12/2/2004 12:00:00 AM
Published date :
Dec 1, 2004, 06:00 AM
Authors :
1Joong-Koo Kang, 2Ki-Chun Im, 2Yu-Jung Yang, 1Sang-Ahm Lee, 1Ji-Hyun Kim, 3Jung-Kyo Lee, and 2Jae-Seung Kim
For the subtraction of interictal from ictal SPECT coregistered to MRI (SISCOM), count normalization is an essential procedure and may affect the outcome of SISCOM. Although whole-brain activity is widely used to normalize the count of ictal and interictal SPCET, the statistical normalization method has been reported to be shown good results on various simulate images. We compared the sensitivity and usefulness of whole brain activity normalization (WN) SISCOM, statistical normalization (SN) SISCOM, and visual assessment of ictal and interictal SPECT images in the patients with unilateral temporal lobe epilepsy (TLE) to determine whether SN SISCOM method is more useful than WN SISCOM and visual analysis. Twenty TLE patients (M/F=10/10, age=32[plusmn]7.7, right=8, left=12) with successful outcome after surgery were included in the study. Ictal and interictal SPECT was performed using Tc-99m ECD and triple head gamma camera. For SISCOM, the count normalization of ictal and interictal SPECT images was performed by both methods (WN and SN). While the count scaling factor of WN was obtained by the linear regression between ictal and interictal counts in the whole brain, that of SN was obtained by the linear regression of ictal and interictal voxels within one standard deviation of mean value in whole brain. Two reviewers, who were blinded to clinical data, assessed these three image sets (WN SISCOM, SN SISCOM, and ictal/interictal SPECT images) using three grading scale: good (possible localization); fair (possible lateralization); and bad (non-lateralization). SN SISCOM had good (45%) or fair (38%) in 16.5 (83%) of 20 patients, whereas WN SISCOM had good (35%) or fair (38%) in 14.5 (73%) patients and visual analysis showed good (50%) or fair (25%) in 15 (75%) patients. The concordance rates of two reviewers in SN SISCOM, WN SISCOM, and visual analysis were 85%, 85% and 45%, respectively. Diagnostic sensitivities of SN SISCOM, WN SISCOM, and visual analysis for the correct local/lateralization were 75% (15/20), 60% (12/20), and 40% (8/20), respectively. There were 2 (10%) false positives for the local/lateralization in all methods. This study suggested that SISCOM is more sensitive and objective method than visual analysis of ictal/interictal SPECT to localize the epileptogenic area in TLE patients. SN method may improve the diagnostic sensitivity of SISCOM and make interpretation easier than WN method.