Fractal image analysis for automated kidney microscopy

Abstract

This paper describes the implementation of an image analysis system for automated kidney microscopy. Autofocused images of kidney tissue sections, which differ in texture according to the severity of tissue damage, are evaluated. It is shown that fractal dimension varies according to the state of the kidney tissue, and is significantly different between normal and abnormal states. A linear classifier is designed based on the statistical properties of the experimental data. Satisfactory classification of normal and damaged kidney tissue is achieved.