Automatic microarray image segmentation with clustering-based algorithms

Abstract

Image segmentation, as a key step of microarray image processing, is crucial for obtaining the spot expressions simultaneously. However, state-of-art clustering-based segmentation algorithms are sensitive to noises. To solve this problem and improve the segmentation accuracy, in this article, several improvements are introduced into the fast and simple clustering methods (K-means and Fuzzy C means). Firstly, a contrast enhancement algorithm is implemented in image preprocessing to improve the gridding precision. Secondly, the data-driven means are proposed for cluster center initialization, instead of usual random setting. The third improvement is that the multi features, including intensity features, spatial features, and shape features, are implemented in feature selection to replace the sole pixel intensity feature used in the traditional clustering-based methods to avoid taking noises as spot pixels. Moreover, the principal component analysis is adopted for various feature extraction. Finally, an adaptive adjustment algorithm is proposed based on data mining and learning for further dealing with the missing spots or low contrast spots. Experiments on real and simulation data sets indicate that the proposed improvements made our proposed method obtains higher segmented precision than the traditional K-means and Fuzzy C means clustering methods.

An image-to-answer algorithm for fully automated digital PCR image processing

“…A passive reference dye is added to the dPCR master mix to provide the background fluorescence signal, help the software recognize PW and NW in the chip, and improve statistical precision 15 . The pattern recognition method identifies the partitions with fluorescence using a machine-learning algorithm 16 , a region-growing algorithm 17 , and a clustering algorithm [18][19][20] . However, chip defects, such as dust and scratches, often result in bright image areas incorrectly identified as PW, while the partitions not filling the sample may be identified as NW, affecting the result's accuracy.…”

“…15 The pattern recognition method identifies the partitions with fluorescence using a machinelearning algorithm, 16 a region-growing algorithm, 17 and a clustering algorithm. [18][19][20] However, chip defects, such as dust and scratches, often result in bright image areas incorrectly identified as PW, while the partitions not filling the sample may be identified as NW, affecting the result's accuracy.…”

Automated lung tumor delineation on positron emission tomography/computed tomography via a hybrid regional network

“…7 Even with these limitations, the importance of PET/CT in diagnosis, staging, and therapy has prompted research on several automatic tumor segmentation methods in recent years. 8 These automatic segmentation methods range from simple techniques (e.g., fixedhard thresholding, 9 soft-probabilistic thresholding, 10 and clustering methods 11 ) to more complex techniques, such as deep learning methods using fully convolutional networks (FCNs). 12 Simple methods evolve from the idea that there is differential radiotracer uptake between a tumor and nearby normal tissues; therefore, by setting the threshold to a certain hard or soft level, we can segment or contour these different tissues.…”

Microarray Image Analysis: From Image Processing Methods to Gene Expression Levels Estimation

“…Unsupervised grid alignment methods for microarray images have been also proposed [6], [7] based on the use of optimal multilevel thresholding followed by a refinement procedure to find the positions of the sub-grids in the image and the positions of the spots in each detected sub-grid. For the segmentation of microarray spots, adaptive pixel clustering techniques were used in [8], [9], [10] . Alternate spatial methods, such as the snake fisher model or 3D spot modeling were used for spot segmentation in [11] and [12], respectively.…”