Feedback-Driven Design of Normalization Techniques for Biological Images Using Fuzzy Formulation of a Priori Knowledge

Abstract

In digital imaging, a normalization procedure is an important step for an efficient and meaningful analysis of any random image dataset. The original intensity information in a digital image is mostly distorted due to imperfect acquisition conditions resulting in the shading phenomenon. Additionally, the high contrast of gray values present in an image also imparts a bias to retrieved gray values. Consequently, image processing goals such as segmentation and cell classification are adversely affected by aforementioned factors. In many microscopic imaging applications, retrospective shading correction methods are more commonly used as opposed to prospective methods in order to remove unwanted shading effects. The objectives of a normalization process, for one, can be rescaling of pixel values to a desired range while disregarding outliers and noisy background pixels. To counter shading effects, robust normalization techniques based on the adaptation of normalization parameters should be devised. We propose a feedback-based automatic image normalization technique that incorporates the evaluation criterion for its effectiveness based on image processing goals such as segmentation. Such a technique employs surface fitting of the available image pixel values to structures of a given family of function (such as polynomials) describing the spatial intensity variation of that image. It incorporates fuzzy formulation of criteria for normalization evaluation as an internal consistency check, while including post-segmentation results based on a priori segmentation knowledge at the same time. Results from a biological dataset consisting of images showing normal and dying cells are included to elucidate the effectiveness of the proposed scheme by automatically adapting normalization parameters.

Fuzzy-based propagation of prior knowledge to improve large-scale image analysis pipelines

“…An uncertainty formulation based on fuzzy set theory has been employed to perform pixel-or object-based classification tasks [11,39,52]. A further possibility to exploit the uncertainty information is to optimize parameter values of a respective operator in a feedback fashion such that the outcome minimizes a previously defined optimization criterion as demonstrated in [22,23]. Another example is the improvement of a graph-based watershed implementation, where uncertainties are used to assess the influence of individual edges on the final segmentation outcome [50].…”

The MATLAB Toolbox SciXMiner: User's Manual and Programmer's Guide

“…• validation strategies [69], and Images and single features Biology [52] Ceramic actuator optimization Time series and single features Engineering [17] Circadian parameter estimation Time series Medical technology [99] Climate change events on streets Time series Engineering [51] Control of prosthesis with muscle signals…”

A New Feedback-Based Method for Parameter Adaptation in Image Processing Routines

“…The pre-processing aims at unifying all images in a sense that noise is suppressed, shading is minimized etc. [ 1 , 2 ]. The segmentation discriminates between useful objects as foreground and a trivial or unwanted background region.…”