Reproducibility of 2D and 3D fractal analysis techniques for the assessment of spatial heterogeneity of regional blood flow in rectal cancer.

Sanghera, Bal; Banerjee, Debasish; Khan, Aftab; Simcock, Ian; Stirling, J James; Glynne-Jones, Rob; Goh, Vicky

PURPOSE To characterize the two-dimensional (2D) and three-dimensional (3D) fractal properties of rectal cancer regional blood flow assessed by using volumetric helical perfusion computed tomography (CT) and to determine its reproducibility. MATERIALS AND METHODS Institutional review board approval and informed consent were obtained. Ten prospective patients (eight men, two women; mean age, 72.3 years) with rectal adenocarcinoma underwent two repeated volumetric helical perfusion CT studies (four-dimensional adaptive spiral mode, 11.4-cm z-axis coverage) without intervening treatment within 24 hours, with regional blood flow derived by using deconvolution analysis. Two-dimensional and 3D fractal analyses of the rectal tumor were performed, after segmentation from surrounding structures by using thresholding, to derive fractal dimension and fractal abundance. Reproducibility was quantitatively assessed by using Bland-Altman statistics. Two-dimensional and 3D lacunarity plots were also generated, allowing qualitative assessment of reproducibility. Statistical significance was at 5%. RESULTS Mean blood flow was 63.50 mL/min/100 mL ± 8.95 (standard deviation). Good agreement was noted between the repeated studies for fractal dimension; mean difference was -0.024 (95% limits of agreement: -0.212, 0.372) for 2D fractal analysis and -0.024 (95% limits of agreement: -0.307, 0.355) for 3D fractal analysis. Mean difference for fractal abundance was -0.355 (95% limits of agreement: -0.869, 1.579) for 2D fractal analysis and -0.043 (95% limits of agreement: -1.154, 1.239) for 3D fractal analysis. The 95% limits of agreement were narrower for 3D than 2D analysis. Lacunarity plots also visually confirmed close agreement between repeat studies. CONCLUSION Regional blood flow in rectal cancer exhibits fractal properties. Good reproducibility was achieved between repeated studies with 2D and 3D fractal analysis.