Premium
Background determination‐based detection of scattered peaks
Author(s) -
Stroebel Armin,
Welzel Oliver,
Kornhuber Johannes,
Groemer Teja W.
Publication year - 2010
Publication title -
microscopy research and technique
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.536
H-Index - 118
eISSN - 1097-0029
pISSN - 1059-910X
DOI - 10.1002/jemt.20858
Subject(s) - histogram , gaussian , artificial intelligence , a priori and a posteriori , pattern recognition (psychology) , signal (programming language) , signal to noise ratio (imaging) , computer science , resolution (logic) , detection theory , standard deviation , mathematics , physics , detector , image (mathematics) , optics , statistics , philosophy , epistemology , quantum mechanics , programming language
In many instances of signal and image processing, it is indispensable to precisely distinguish scattered peaks from a background, e.g., camera signals in microscopy. Here we addressed the detection of Gaussian signals in simulated line profiles (LP) comparable with e.g., fluorescence microscopy data. In a first step, we measured the applicability of histogram‐based global background estimation. We find that the method is valid for typical scattered Gaussian signals if they are averagely separated by interpeak distances of 5.5 standard deviations. This enabled us to design global background determination‐based peak detection (GBPD). GBPD was compared with two local background determination‐based signal detection methods that had been designed for analysis of electrophysiological data andmicroscopy images, respectively. We were able to prove via receiver–operator characteristic (ROC) comparisons of signal‐to‐noise ratio (SNR), interpeak distance, and filtering behavior that, when applicable, GBPD brings advantages in knowledge needed a priori, performance at any SNR, controllability and spatial resolution. Microsc. Res. Tech. 73:1115–1122, 2010. © 2010 Wiley‐Liss, Inc.