Fractal analysis of microvascular networks in malignant brain tumors.

Brain tumors are characterized by a microvascular network which differs from normal brain vascularity. Different tumors show individual angiogenic patterns. Microvascular heterogeneity can also be observed within a neoplastic histotype. It has been shown that quantification of neoplastic microvascular patterns could be used in combination with the histological grade for tumor characterization and to refine clinical prognoses, even if no objective parameters have yet been validated. To overcome the limits of the Euclidean approach, we employ fractal geometry to analyze the geometric complexity underlying the microangioarchitectural networks in brain tumors. We have developed a computer-aided fractal-based analysis for the quantification of the microvascular patterns in histological specimens and ultra-high-field (7-Tesla) magnetic resonance images. We demonstrate that the fractal parameters are valid estimators of microvascular geometrical complexity. Furthermore, our analysis allows us to demonstrate the high geometrical variability underlying the angioarchitecture of glioblastoma multiforme and to differentiate low-grade from malignant tumors in histological specimens and radiological images. Based on the results of this study, we speculate the existence of a gradient in the geometrical complexity of microvascular networks from those in the normal brain to those in malignant brain tumors. Here, we summarize a new methodology for the application of fractal analysis to the study of the microangioarchitecture of brain tumors; we further suggest this approach as a tool for quantifying and categorizing different neoplastic microvascular patterns and as a potential morphometric biomarker for use in clinical practice.