Threshold Selection Process Research Articles

Error minimization algorithm for comparative quantitative PCR analysis: Q-Anal

Current methods for comparative quantitative polymerase chain reaction (qPCR) analysis, the threshold and extrapolation methods, either make assumptions about PCR efficiency that require an arbitrary threshold selection process or extrapolate to estimate relative levels of messenger RNA (mRNA) transcripts. Here we describe an algorithm, Q-Anal, that blends elements from current methods to by-pass assumptions regarding PCR efficiency and improve the threshold selection process to minimize error in comparative qPCR analysis. This algorithm uses iterative linear regression to identify the exponential phase for both target and reference amplicons and then selects, by minimizing linear regression error, a fluorescence threshold where efficiencies for both amplicons have been defined. From this defined fluorescence threshold, cycle time (Ct) and the error for both amplicons are calculated and used to determine the expression ratio. Ratios in complementary DNA (cDNA) dilution assays from qPCR data were analyzed by the Q-Anal method and compared with the threshold method and an extrapolation method. Dilution ratios determined by the Q-Anal and threshold methods were 86 to 118% of the expected cDNA ratios, but relative errors for the Q-Anal method were 4 to 10% in comparison with 4 to 34% for the threshold method. In contrast, ratios determined by an extrapolation method were 32 to 242% of the expected cDNA ratios, with relative errors of 67 to 193%. Q-Anal will be a valuable and quick method for minimizing error in comparative qPCR analysis.

AUC: a misleading measure of the performance of predictive distribution models

ABSTRACTThe area under the receiver operating characteristic (ROC) curve, known as the AUC, is currently considered to be the standard method to assess the accuracy of predictive distribution models. It avoids the supposed subjectivity in the threshold selection process, when continuous probability derived scores are converted to a binary presence–absence variable, by summarizing overall model performance over all possible thresholds. In this manuscript we review some of the features of this measure and bring into question its reliability as a comparative measure of accuracy between model results. We do not recommend using AUC for five reasons: (1) it ignores the predicted probability values and the goodness‐of‐fit of the model; (2) it summarises the test performance over regions of the ROC space in which one would rarely operate; (3) it weights omission and commission errors equally; (4) it does not give information about the spatial distribution of model errors; and, most importantly, (5) the total extent to which models are carried out highly influences the rate of well‐predicted absences and the AUC scores.

Image Thresholding Computation Using Atanassov’s Intuitionistic Fuzzy Sets

In this paper, a new thresholding technique using Atanassov’s intuitionistic fuzzy sets (A-IFSs) and restricted dissimilarity functions is introduced. In recent years, various thresholding techniques ([18, 24]) based on fuzzy set theory have been introduced to overcome the problem of non-uniform illumination and inherent image vagueness. In this paper we analyze this task and propose a new method for handling the grayness ambiguity and vagueness during the process of threshold selection.

Extracting Objects with Adaptive Segmentation Techniques: Going Beyond Intensity Thresholding

Abstract In designing automated systems for interpretation of micrographs, it is often the goal to separate and discern various objects within the image data. After segmentation, measurements of the objects, called features, can then be calculated and used for process or statistical analysis. Simple segmentation schemes based on single threshold operations, often lack the sophistication to deal with intricate or subtle details of the image data, or require user intervention in the threshold selection process. This talk will present and discuss advanced techniques which adapt to the data, and which can operate autonomously without human supervision. Segmentation is a process which transforms pixel based image data into symbolic descriptors representing groups of pixel elements. These descriptors are called objects and take the form of lines, regions, polygons, points, windows of interests, or other unique representations. Most laboratory image analysis software packages utilize segmentation schemes based on the principle of intensity analysis.