A Laplacian of Gaussian-Based Approach for Spot Detection in Two-Dimensional Gel Electrophoresis Images

Abstract

Abstract. Two-dimension gel electrophoresis (2-DE) is a proteomic technique that allows the analysis of protein profiles expressed in a given cell, tissue or biological system at a given time. The 2-DE images depict protein as spots of various intensities and sizes. Due to the presence of noise, the inhomogeneous background, and the overlap between the spots in 2-DE image, the protein spot detection is not a straightforward process. In this paper, we present an improved protein spot detection approach, which is based on Laplacian of Gaussian algorithm, and we extract the regional maxima by morphological grayscale reconstruction algorithm, which can reduce the impact of noisy and background in spot detection. Experiments on real 2-DE images show that the proposed approach is more reliable, precise and less sensitive to noise than the traditional Laplacian of Gaussian algorithm and it offers a good performance in our gel image analysis software.

EmiNet: Annotation Free Moving Bacteria Detection on Optical Endomicroscopy Images

“…Pearson Correlation Coefficient DoG [38] 0.7389 GSOTH [39] 0.7765 LoG [40] 0.7929 Bayesian Approach [21] 0.8242 EmiNet 0.8922 information within the image across spatial and temporal dimensions for sequence segmentation. Additionally, EmiNet outperforms TransBTS which indicates that the use of motion features alongside the appearance features helps the model to capture the moving bacteria better.…”

“…We compared the detection count correlation with four existing bacteria detection algorithms. These are Laplacian of Gaussian (LoG) [40], difference of Gaussians (DoG) [38], greyscale opening top-hat filter (GSOTH) [39], [41] and the Bayesian approach [21]. The results are summarized in Table III.…”

Bayesian Statistical Analysis for Bacterial Detection in Pulmonary Endomicroscopic Fluorescence Lifetime Imaging

“…Despite being simple to implement, these methods have been utilised in literature [39]- [41] for detecting spots and blobs in different applications (including medical applications). These algorithms are Laplacian of Gaussian (LoG) [42], difference of Gaussians (DoG) [43], and greyscale opening top-hat filter (GSOTH) [44] [45]. In this paper to implement the LoG filter, a 5 × 5 kernel with a 0.8 standard deviation has been used, following preliminary runs aimed at performance optimization.…”

Patch-Based Sparse Representation For Bacterial Detection

“…In this subsection, we compare the proposed approach with popular spot-detection methods from the literature, namely the Laplacian of Gaussian (LoG) and its approximation; the difference of Gaussians (DoG) filters [9,29], and the grey scale opening top-hat filter (GSOTH) [7,8]. These methods, although simple, have been considered in the literature of spot and blob detection in various applications.…”

“…Outlier/anomaly detection problems can usually be addressed using unsupervised or supervised methods [1]. In unsupervised approaches, the objects/anomalies to be detected are learned from the data by fitting them with suitable distributions without using explicitly-provided labels [2][3][4][5][6][7][8][9][10]. On the other hand, considering supervised approaches, the dataset is usually divided into training and testing sets.…”