Software development for fractal analysis
of histopathologic images
Pavlopoulos PM, Kavantzas N, Agapitos E, Patsouris E
Department of Pathology, Medical School, National University of Athens, Greece
Introduction: Considering the multiple applications of fractal analysis in the field of histopathology, we developed "Fractalyser", a specialized fractal analysis program for Windows operating system.
Material and Methods: The application was developed in Visual Basic 6.0 environment. The program was designed with multiple document interface and Windows XP compatibility. It can open, edit and analyze the most common image file types. The program has some basic image processing functions such as invert, horizontal and vertical flip, grayscale conversion and crop. The user can precisely select the area of interest applying parameterized threshold or edge detection algorithms in the color image, while getting real time graphical feedback on the process. The produced binary image, which is projected as a red layer on the original image, can be further optimized before fractal analysis with "dilation", "erosion" or several noise reduction filters. The estimation of the fractal dimension of the image is carried out by using the "box-counting" algorithm. Results: The accuracy of the application was assessed by analyzing images with theoretically known fractal dimension. The errors ranged from 0.0 to 5.8%, being conversely proportional to the number of grid offsets. Repeated comparisons of the statistical results of the program with the corresponding data of a commercially available statistical package confirmed their validity and precision.
Conclusion: As it has been observed in thorough testing, the application presented here constitutes a stable, relatively fast, reliable and easy to use tool for fractal analysis. "Fractalyser" is provided as freeware to researchers.
Êey words: Fractal analysis, fractal dimension, image analysis.
|