Assessment of the spatial pattern of colorectal tumour perfusion estimated at perfusion CT using two-dimensional fractal analysis

• Published in: European radiology Vol. 19; no. 6; pp. 1358 - 1365
• Main Authors: Goh, Vicky, Sanghera, Bal, Wellsted, David M., Sundin, Josefin, Halligan, Steve
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer-Verlag 01.06.2009; Springer Nature B.V
• Subjects: Aged; Aged, 80 and over; Algorithms; Artificial Intelligence; Colorectal Neoplasms - blood supply; Colorectal Neoplasms - complications; Colorectal Neoplasms - diagnostic imaging; Computed Tomography; Diagnostic Radiology; Feasibility Studies; Fractals; Humans; Imaging; In Vitro Techniques; Internal Medicine; Interventional Radiology; Male; Medicine; Medicine & Public Health; Middle Aged; Neovascularization, Pathologic - complications; Neovascularization, Pathologic - diagnostic imaging; Neuroradiology; Pattern Recognition, Automated - methods; Perfusion Imaging - methods; Radiographic Image Enhancement - methods; Radiographic Image Interpretation, Computer-Assisted - methods; Radiology; Reproducibility of Results; Sensitivity and Specificity; Tomography, X-Ray Computed - methods; Ultrasound; Perfusion CT; Dynamic contrast-enhanced CT; Fractal analysis; Colorectal cancer
• ISSN: 0938-7994; 1432-1084
• DOI: 10.1007/s00330-009-1304-y

The aim was to evaluate the feasibility of fractal analysis for assessing the spatial pattern of colorectal tumour perfusion at dynamic contrast-enhanced CT (perfusion CT). Twenty patients with colorectal adenocarcinoma underwent a 65-s perfusion CT study from which a perfusion parametric map was generated using validated commercial software. The tumour was identified by an experienced radiologist, segmented via thresholding and fractal analysis applied using in-house software: fractal dimension, abundance and lacunarity were assessed for the entire outlined tumour and for selected representative areas within the tumour of low and high perfusion. Comparison was made with ten patients with normal colons, processed in a similar manner, using two-way mixed analysis of variance with statistical significance at the 5% level. Fractal values were higher in cancer than normal colon (p ≤ 0.001): mean (SD) 1.71 (0.07) versus 1.61 (0.07) for fractal dimension and 7.82 (0.62) and 6.89 (0.47) for fractal abundance. Fractal values were lower in ‘high’ than ‘low’ perfusion areas. Lacunarity curves were shifted to the right for cancer compared with normal colon. In conclusion, colorectal cancer mapped by perfusion CT demonstrates fractal properties. Fractal analysis is feasible, potentially providing a quantitative measure of the spatial pattern of tumour perfusion.