Fractal Dimension Assessment of Dibenz[a,l]pyrene Oral Carcinogenesis

Oral squamous cell carcinoma (OSCC) is diagnosed in over 270,000 patients worldwide every year. Histopathologic diagnosis relying upon a change of pattern in tissue architecture may be compromised by disagreement among pathologists, leading to significant consequences for patient treatment and prognosis. Syngeneic Syrian hamsters were treated with dibenz[a,]pyrene (DBP), a polycyclic hydrocarbon and a powerful carcinogen, in concentrations ranging from 0.00025 nM to 0.025 nM over a period of up to 40 weeks. A fractal analysis was performed on histology images of random treated tissue samples using ImageJ image processing software. This analysis calculated fractal dimension (FD), lacunarity, and the presence of inflammatory infiltrate for each sample. In addition, each image was categorized histopathologically according to an established pattern of tumor invasion. By performing advanced statistical analysis, the categorization created via fractal analysis was compared to the established histopathology criteria. Samples categorized as normal tissue or those that showed signs of mild dysplasia were found to have typically lower FD and higher lacunarity when compared to samples categorized as showing hyperplasia or OSCC. The process of two-step clustering produced separate categories that correlated to the pattern of invasion criteria, and their statistical significance was confirmed by area under the curve tests (based on p < .05). These results provide evidence that fractal dimension analysis can categorize various stages of carcinogenesis, displaying the potential to improve the overall accuracy of diagnosis.