Abstract
A novel Fourier-based image analysis method for measuring fractal features is presented which can significantly reduce artifacts due to non-fractal edge effects. The technique is broadly applicable to the quantitative characterization of internal morphology (texture) of image features with well-defined borders. In this study, we explore the capacity of this method for quantitative assessment of intracellular fractal morphology of mitochondrial networks in images of normal and diseased (precancerous) epithelial tissues. Using a combination of simulated fractal images and endogenous two-photon excited fluorescence (TPEF) microscopy, our method is shown to more accurately characterize the exponent of the high-frequency power spectral density (PSD) of these images in the presence of artifacts that arise due to cellular and nuclear borders.
Highlights
The analysis of biomedical images is critical for detection of abnormalities and disease, but it is often subject to the interpretation of a medical professional
To demonstrate the baseline sensitivity of the power spectral density (PSD)-based approach for recovering power-law exponents (β) that describe the fractal character of images, we evaluate the PSD of square fractal images generated as described in Methods
To assess whether variations in cell-shaped features impact PSD-based outcomes, we generated binary image models of simulated cell objects (SCOs) that consist of uniform circles on the size order of cells and nuclei
Summary
The analysis of biomedical images is critical for detection of abnormalities and disease, but it is often subject to the interpretation of a medical professional. Optimization and development of these methods is still underway, and more groups are recognizing the utility of these techniques for extracting patterns and information from biomedical images Uncovering this image information is likely to lead to the discovery of novel and objective diagnostic criteria, improving diagnostic sensitivity and enabling earlier disease detection. In this way, widespread application and optimization of quantitative image analysis techniques has great potential to impact the performance of clinical diagnostics and basic research that relies on interpretation of biomedical images. Fourier-based techniques have wide-range applications in signal and image assessment and are gaining a more critical role in tissue characterization. Fractal can be considered self-affine if variation in one direction scales differently than variation in another direction [4,5,6]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
