The influence of slice thickness on the volume measurement accuracy of 3-D MR reconstructions of acrylic phantoms: a precursor to knee imaging

Abstract

Purpose Publications regarding volume measurements derived from reconstructed three-dimensional (3-D) images have arisen from several research centres, yet only one example could be found of previous work specifically addressing the effect of acquisition slice thickness on the accuracy of such volume measurements in a musculoskeletal setting. Given the relative lack of previous work, as a precursor to clinical tissue trials, a phantom study was undertaken to examine the accuracy with which volume measurements could be made from 3-D reconstructed magnetic resonance (MR) images. We also wished to quantify the effect of acquisition slice thickness on resultant volume measurement accuracy. Methods A series of acrylic phantoms of varying known volumes (range: 990–21,450 mm 3) was scanned using T1-weighted MR images at slice thicknesses of 1.0, 2.0, 3.0, 4.0 and 5.0 mm. Four lesions (range: 61–212 mm 3) were milled into the surface of the largest phantom. Using a reconstructive software package, the scans obtained at each slice thickness were then used to create a 3-D computer model of the phantoms and lesions from which direct volume measurements were recorded. These measurements were then compared to the known volumes of the phantoms and a percentage measurement error (PME) was determined. Results The mean PME values recorded in the assessment of the phantoms decreased as acquisition slice thickness was reduced (10.84%, 8.27%, 4.86%, 2.80% and 0.59% for 5.0, 4.0, 3.0, 2.0 and 1.0 mm slices, respectively). An association between slice thickness and accuracy of volume measurement was shown (one-sample t-test: P<0.05). Discrete surface lesions were confidently recognised at all acquisition slice thickness although quantification at each level showed considerable variance. Conclusion The results of this study support the accepted inverse association between slice thickness and volume measurement accuracy. If a 5% level of PME is considered acceptable, the findings suggest that slice thicknesses of 3.0 mm or less be employed where analysis of lesions estimated to fall within a volume range of 1000–20,000 mm 3 is being considered.