Intraobserver and interobserver variability in measuring changes in lesion volume on serial brain MR images in multiple sclerosis.

PURPOSE We evaluated the intraobserver and interobserver variability in measuring long-term changes in the volume of brain lesions on 5- and 3-mm-thick MR sections in patients with multiple sclerosis. METHODS Eighteen 18 patients were scanned on two separate occasions with a mean interval of 16.4 months between the two examinations. In each session, a scan with 24 contiguous 5-mm-thick axial sections and another with 40 contiguous 3-mm-thick axial sections was acquired consecutively without moving the patient. We assessed MR lesion load by using a semiautomated local thresholding technique. RESULTS Lesion volume was significantly higher on images with 3-mm-thick sections than on those with 5-mm-thick sections both at baseline and at follow up. Significant increases in total lesion volume were observed during the follow-up period on images obtained with both 5- and 3-mm-thick sections. The intra- and interobserver variability in measurements of changes in lesion volume was significantly higher on images with 5-mm-thick sections than on those with 3-mm-thick sections. CONCLUSION Our data indicate that the acquisition of thinner sections increases the reliability of the assessment of changes in brain lesion load on MR images in patients with multiple sclerosis.