Under construction....

SNM2000:

Epilepsy related perfusion changes in the peripheral cortex

Rik Stokking, Diagnostic Radiology, Yale University School of Medicine

Objective: Tc99m-HMPAO SPECT difference analysis (ictal minus interictal perfusion) is a routine part in the presurgical evaluation of patients with epilepsy at Yale [Zubal JNM95, Avery EJNM99]. Perfusion changes not only show up in the epileptogenic area in a large number of cases, but also in other areas of the brain. Our aim was to evaluate the epilepsy related perfusion changes for the peripheral cortex using a 3D integrated visualization method called Normal Fusion (NF) [Stokking JNM99]. Methods: 9 patients from the Yale Epilepsy Program were selected with a confirmed epileptogenic area in the mesial temporal lobe, presurgical interictal and ictal Tc99m-HMPAO SPECT and high resolution MRI, and video-EEG monitoring during the ictal HMPAO injection. SPECT difference images were calculated and registered to the MR (positioned in Talairach space). The NF method produced 3D visualizations of the brain with the perfusion changes over a depth of 0-10 mm below the cortical surface mapped and color encoded onto the surface. Results+conclusions: Distinct perfusion patterns were found that correctly lateralized the epileptogenic area in all cases. Also, these patterns appeared highly correlated with time of injection from seizure onset and unrelated to seizure duration. This will considerably improve the analysis of SPECT difference images and subsequently the presurgical evaluation of patients with epilepsy.

Extended version of work in progress:

Epilepsy affects approximately 1% of the population and it is estimated that approx. 500.000 patients with epilepsy in the USA would benefit from removal of their epileptogenic region [Spencer&Ward].  A prerequisite for successful removal of the abnormal brain tissue is the localization of this region. The functional imaging modality SPECT has proven of considerable importance in the pre-surgical evaluation of patients with medically refractory epilepsy since it is a comparatively practical technique to acquire information on blood perfusion not only between, but also during seizures.

The technique of SPECT uses a radioactive tracer (here: 99mTc-HMPAO) injected into a vein. Upon arrival in the brain the tracer is trapped in brain cells. Subsequent scanning basically presents a 3D snapshot image of the brain perfusion right after the injection. Since the blood flow is known to increase during a seizure in the epileptogenic region, comparing SPECT images obtained during a seizure (ictal scan) to images obtained in between seizures (interictal scan) are considered helpful in the localization of the epileptogenic region. Furthermore, perfusion changes related to the seizure will also show up in other areas of the brain, e.g. as a result of seizure spread. An understanding of the seizure propagation patterns is likely to (further) improve localization accuracy [see EvansJCBFM91, A69-78].

Results show that the comparison between the interictal and the ictal SPECT scans is best performed using sophisticated computer techniques to accurately register, normalize, and subsequently subtract the patient's interictal from the ictal SPECT scan and overlay the results onto the corresponding anatomical information provided by the (registered) MR image data [Zubal95,Spanaki99,Spencer..].

The current study aims to evaluate the 3D cortical perfusion patterns in epilepsy using the Normal Fusion approach. We wanted our analysis to be as uniform as possible, which is why we decided to use only patients with mesial temporal lobe epilepsy confirmed by excellent surgical outcome (category ?? according to [Engel00]) and pathology of the resected tissue. Additional selection criteria were the availability of an ictal and interictal SPECT scan, a high resolution  MRI scan (slice thickness <= 3mm), no gross abnormalities of the brain, no previous surgery, and continuous video-EEG monitoring during the ictal 99mTc-HMPAO injection.  This rigorous selection resulted in a total of  ?? patients.

The whole procedure from the original image datasets till the 3D visualizations involves several registration steps using the Mutual Information technique (ictal to interictal SPECT, interictal SPECT to MRI, MRI to Montreal Brain Phantom)[Maes??,Studholme], normalization of the ictal to the interictal SPECT using [Studholme], a brain segmentation using MBRASE [Stokking, 2000], and 3D visualization using VROOM [Zuiderveld]. All of the separatesteps are fully automatic. The operator is only required to verify all of the results which is, of course, of the upmost importance. The final result is a mosaic image with 8 views of the brain of the patient color encoded with the respective local functional information from the difference SPECT data. Six views show each of the orthogonal views of the brain and the other 2 views are so-called split-brain visualizations i.e. views of the medial surfaces of  both hemispheres.

Fully automatic segmentation of the brain from the MRI data was performed using the MBRASE software program [Stokking, 2000]. This is a simple program based on principles from mathematical morphology [see also Hohne92]. All registrations (ictal to interictal SPECT, and interictal to MRI) were performed using the Mutual Information technique, the current standard for these types of registrations [Maes, West, Studholme]. We also performed a non-rigid registration of all the MR datasets to the Montreal Brain Phantom, which is in Talairach space. We applied the obtained translation and rotation components in the 3D visualization step effectively presenting all the brains in the same reference frame.  The visualizations were performed using the software package VROOM [ZuiderveldPhD].