Abstracts

RATIONALE: Hippocampal atrophy (HA) in mesial temporal lobe epilepsy (MTLE) is frequently associated with refractory seizures and excellent pot-operative prognosis. There are some reports of HA in patients with MTLE and good outcome, but it is not known if the hippocampal abnormalities in this group is different from that observed in refractory patients. The objective of this study is to describe hippocampal texture (Txt) analyses and volumetry in patients with MTLE and benign clinical course.
METHODS: We identified patients with MTLE that had a benign clinical course, defined as a maximum of 1 complex partial seizure per year. All patients underwent high resolution Magnetic Resonance Imaging (MRI) with T1 and T2-WI in three orthogonal planes. Coronal T1-IR 3mm slices were used for Txt analyses, using the software MAZDA[reg], that generates 250 parameters. Region of interest was defined as the hippocampal head, and data was compared to values obtained in a group of healthy adult volunteers. In addition we performed hippocampal volumetric studies in all patients and in a group of normal controls. For statistical analyses we selected those parameters that better discriminated normal hippocampi from pathologically proven mesial temporal sclerosis in a previous study. All data were analyzed using SYSTAT9. We performed ANOVA and regression analyses between hippocampal volumes and Txt values.
RESULTS: A total of 22 patients (16 women) were studied, with mean present age of 42 years (18-68). Control group was composed by 20 adults (15 men), with mean age of 30 years (21-62). There were 16 patients with hippocampal atrophy determined by volumetry: 7 right, 3 left, 6 bilateral. ANOVA showed a significant difference among Txt values from hippocampi of controls, patients[ssquote] normal volume hippocampi and patients[ssquote] atrophic hippocampi. Post-hoc comparisons showed a significant difference between controls[ssquote] versus patients[ssquote] normal volume hippocampi (p=0.03 to 0.001), as well as controls[ssquote] versus patients[ssquote] atrophic hippocampi (p=0.004 to 0.00006). However, the difference between patients[ssquote] normal volume hippocampi and atrophic hippocampi did not reach statistical significance. Regression analysis showed no association between volumes and Txt values.
CONCLUSIONS: Txt analysis is a new post-processing technique that allows quantification of digital imaging characteristics, invisible to human eye. In our series of well controlled patients with MTLE, Txt parameters that could best differentiate atrophic from controls hippocampi were the histogram based parameters. These are correlated to the grey tone dispersion that may be correlated to structural complexity. In these patients, the reduction of these parameters indicates a simplification in hippocampal head formation. The significant difference between controls and patients[ssquote] normal volume hippocampi is indicative that texture analysis may be of potential use for detection of subtle abnormalities, not necessarily associated with reduction of hippocampal volumes.
[Supported by: FAPESP, S[atilde]o Paulo, Brazil]