Reliability of amygdalohippocampal volume measured directly on MRI films

Volume measurement of the amygdalohippocampus on MRI may provide important information for epilepsy surgery. This study investigated the reliability of direct volume assessment (DVA) versus computer-assisted measurement (CAM) of amygdalohippocampus. DVA was carried out by counting the number of squares occupied by amygdalohippocampus in a transparent mm2 overlay used on both sagittal and coronal films. CAM was achieved using the 'thresholding and tracing' automated technique. Seventeen patients undergoing language-dominant left temporal lobectomy were studied. Mean volume as determined by the CAM was 4.99 +/- 1.31. In DVA, the mean volume was 5.05 +/- 1.23 (r = 0.974). Reproducibility of the volume measurement was tested by repeating measurements 5 times on 5 patient samples of amygdalohippocampal complex in DVA, and 3 times on 4 samples in CAM. The mean coefficient of variation for amygdalohippocampal volume was 6.9 +/- 3.2% in DVA and 3.9 +/- 2.0 in CAM. We also tested a more simplified method of assessing amygdalohippocampal volume by calculating the left to right volume ratio from MRI scans (2-40 films for each sample). Volume ratios determined directly from MRI films made in different months (14 samples) and of different sections (coronal and sagittal series, 18 samples) correlated well with each other (p < 0.01). Direct assessment of amygdalohippocampal volume can be used effectively for preoperative evaluation of the patient with epilepsy.