Related Experiment Videos
Fractal and integer-dimensional geometric analysis of pigmented skin lesions
S S Cross1, A J McDonagh, T J Stephenson
1Department of Pathology, University of Sheffield Medical School, England.
The American Journal of Dermatopathology
|August 1, 1995
Summary
Fractal geometric analysis of pigmented skin lesions was studied to improve melanoma diagnosis. The fractal dimension of lesion boundaries did not significantly differentiate malignant melanomas from benign lesions.
Area of Science:
- Dermatology
- Medical Imaging
- Geometric Analysis
Background:
- Accurate diagnosis of pigmented skin lesions is crucial for identifying malignant melanomas and avoiding unnecessary excisions of benign lesions.
- Current clinical diagnostic accuracy rates for pigmented skin lesions are approximately 65%.
Purpose of the Study:
- To investigate if fractal geometric analysis of pigmented skin lesions can enhance diagnostic accuracy.
- To determine the utility of fractal dimension as a morphometric discriminant for classifying skin lesions.
Main Methods:
- Photographed and excised 42 pigmented skin lesions (15 malignant melanomas, 21 melanocytic nevi, 6 basal cell papillomas).
- Measured the fractal dimension of lesion boundaries using a box-counting method on a microcomputer-based image analysis system.
- Performed multivariate linear discriminant analysis using fractal and Euclidean geometric parameters.
Main Results:
- All lesions exhibited a fractal dimension greater than their topological dimension, indicating fractal characteristics.
- Multivariate analysis correctly classified only 45% of lesions, with a kappa statistic of 0.33.
- No significant difference was found in the fractal dimension between melanocytic nevi and malignant melanomas (p = 0.18).
Conclusions:
- Pigmented skin lesions possess fractal elements in their structure.
- The fractal dimension of lesion boundaries is not a reliable morphometric tool for distinguishing malignant melanomas from benign melanocytic nevi.