Jacobus F.A. Jansen1, Cecile R. Jeukens2, Marielle C. Vlooswijk2, H J. Majoie2, Marc C. de Krom2, Albert P. Aldenkamp2, Paul A. Hofman2, Walter H. Backes2
1Department of Medical Physics & Radiology, Memorial Sloan-Kettering Cancer Center, New York, NY, USA; 2Maastricht University Hospital, Maastricht, Netherlands
This work provides a detailed reliability analysis for hippocampal volumetry of T1-weighted MR images acquired at 3T. Two readers performed hippocampal volumetry on MR images of 40 patients with cryptogenic localization-related epilepsy and 20 healthy control subjects. Reliability measures and criteria to assess and ensure sufficient reliability were defined. Also, the validity of the correction for total intracranial volume size was critically assessed. The interreader reliability values were determined: intra-class-correlation-coefficient, ICC = 0.86 (left) and 0.86 (right), percentage volume difference VD = 7.4 5.8 % (left) and 6.2 4.8 % (right), and overlap ratio OR = 0.82 0.04 (left) and 0.082 0.03 (right). The positive Pearson correlation between hippocampal volume and total intracranial volume volumes was found to be low: r = 0.48 (p = 0.03, left) and r = 0.62 (p=0.004, right), showing the limited benefit of the brain size correction.