A Review on Image Processing and Fractal Analysis in Oral Potentially Malignant Disorders

AIM: The clinical evaluation of patients with oral potentially malignant disorders is primarily based on physical examination and observable clinical features. Clinical photographs play a key role in patient monitoring and help identify signs that may indicate malignant transformation. These images can be accessed by artificial intelligence techniques and fractal analysis, and image processing and segmentation are crucial for these tasks. This review discusses image processing techniques applied to clinical photographs of oral potentially malignant disorders, with emphasis on fractal analysis. METHODS: Studies investigating fractal dimension and lacunarity as measures of lesion complexity and homogeneity were reviewed, along with research on segmentation and binarization. RESULTS: A search of scientific databases identified 10 studies reporting the access of texture parameters on clinical images to measure fractal dimension and lacunarity. CONCLUSIONS: The reviewed studies showed that fractal dimension and texture parameters were useful for evaluating lesions and distinguishing normal and pathological tissues; however, there is a lack of methodological standardization. Although fractal dimension was assessed in all articles, lacunarity was examined in just one. Binarization and segmentation impact fractal dimension, lacunarity, and other texture measurements, which can be significantly enhanced by appropriate image processing. In contrast, poorly executed preprocessing may lead to misinterpretation of results.