Automatic DNA microarray gridding based on Support Vector Machines

Abstract

Abstract-This paper presents a novel method for DNA microarray gridding based on Support Vector Machine (SVM) classifiers. It employs a set of soft-margin SVMs to estimate the lines of the DNA microarray grid by maximizing the margin between the lines and the spots. This process comprises an efficient and effective approach of separating the spots into distinct rows and columns. The classifiers are trained using the spot locations as training vectors. The results obtained from the application of the proposed method on reference microarray images illustrate its robustness in the presence of artifacts, noise and weakly expressed spots. The comparative evaluation presented reveals its advantageous performance over a state of the art gridding approach. The gridding quality achieved exceeds 95% in terms of the total number of perfectly gridded spots.

Reduction of capsule endoscopy reading times by unsupervised image mining

“…Improper choice of T may result even in a total loss of frames that are representative of some abnormalities [22]. To cope with this problem, in the next subsection we introduce a criterion for automatic tuning of parameter T.…”

“…The extracted frames can be regarded as a summary or as set of bookmarks indicating the frames of interest of the original video. The proposed methodology provides generality and independence from any training set, and it has two additional features over its preliminary instances presented in [20][21][22]: (a) an automatic tuning mechanism for the parameter controlling the number of frames to be extracted, and (b) applicability on full length WCE videos. This was still an open issue in [20][21][22], since a straightforward application of that methodology would yield a very large data matrix containing thousands of video frames, and its size would be prohibiting for processing by conventional computer systems.…”

Unsupervised SVM-based gridding for DNA microarray images

“…In comparison to the preliminary version presented in [12] which displayed promising results, the achieved accuracy is increased as several changes have been included in the proposed method, such as the automatic determination of valid spot sizes based on the distance between rows d r and columns d c , as well as the inclusion of the valid spots from the whole image into the training set of each SVM classifier. [12] 95.1% 4.5% 0.4% Zacharia et. al.…”

“…The microarray gridding is completed after the application of the above procedure for the determination of the grid lines that separate each pair of consecutive columns of spots. In order to enhance the reliability of the results, the data set used for evaluation is a superset of the one used in [10] and [12], as it includes all 54 images instead of only 25 used in the previous studies. The statistical analysis is performed correspondingly, in order to produce directly comparable results.…”

“…Genetic algorithms [10] have the potential to achieve high accuracy, but are very time-consuming as they have to evaluate a large number of possible solutions in order to converge. Our preliminary version [12] of the proposed method is not entirely unsupervised, as the gridding accuracy depends on the successful selection of a few parameters that have to be experimentally determined.…”