A semi-automatic optimization model has been proposed by Law et al. Then, breast cancer images were quantized through these thresholds to segment the nuclei. The objective function of Otsu’s multilevel thresholding has been maximized based on a particle swarm optimization to achieve the optimum thresholds 18. The seeds were extracted from the centroids of the objects via Otsu’s thresholding method and some basic morphological operators. A level set method guided by seed points has been proposed by Husham et al. Recently, optimization methods have been employed to introduce new segmentation techniques.
iterative-based 13, information-theoretic-based 14, 15, 16, histogram-based 17, 18, 19. active contours 2, 3, level set 4, 5, global minimizers 6, graph-based 7 2- machine-learning-based in which a machine/network is trained to recognize features, in particular deep convolutional neural networks have drawn many attention recently 8, 9, 10, 11, 12 3- threshold-based in which a set of thresholds are found, e.g. These methods can be categorized into three general groups: 1- optimization-based in which an energy/cost function is maximized/minimized, e.g. The main role of every segmentation method is to separate an image foreground from background. Under/over-segmentations happen in the presence of heavily clustered nuclei, due to the variability and complexity of data caused by noise, uneven absorption of stains, different cell types, etc. This leads to a valuable insight into the cell features and functionality which result in early diagnosis of diseases such as breast cancer and brain tumour.ĭespite considerable progress in automated segmentation, it remains a challenging task to separate a large clump of nuclei and delineate their boundaries with a high accuracy and speed. Automated nucleus segmentation is fundamental to cell counting, movement tracking, and morphological study, such as feature extraction and classification. Segmentation of cell nucleus from histopathological image, has been a focus of clinical practice and scientific research for more than half a century 1. Evaluated results show our method outperforms the state-of-the-art methods in terms of accuracy, precision, F1-measure, and computational time. We have applied our method to three image datasets: breast cancer stained for hematoxylin and eosin (H&E), Drosophila Kc167 cells stained for DNA to label nuclei, and mature neurons stained for NeuN. Finalized by a model-based merging step to eliminate oversegmentation and a model-based correction step to improve segmentation results and eliminate small objects. Followed by a new splitting technique based on multilevel thresholding and the watershed algorithm to separate clustered nuclei. Our algorithm is initiated by applying a new locally adaptive thresholding method on watershed regions. To address this problem, we propose a novel method applicable to variety of histopathological images stained for different proteins, with high speed, accuracy and level of automation. Despite considerable advances in automated segmentation, it still remains a challenging task to split heavily clustered nuclei due to intensity variations caused by noise and uneven absorption of stains. Automated cell nucleus segmentation is the key to gain further insight into cell features and functionality which support computer-aided pathology in early diagnosis of diseases such as breast cancer and brain tumour.
0 kommentar(er)
