Unsupervised segmentation and classification of cervical cell images

Abstract

a b s t r a c tThe Pap smear test is a manual screening procedure that is used to detect precancerous changes in cervical cells based on color and shape properties of their nuclei and cytoplasms. Automating this procedure is still an open problem due to the complexities of cell structures. In this paper, we propose an unsupervised approach for the segmentation and classification of cervical cells. The segmentation process involves automatic thresholding to separate the cell regions from the background, a multi-scale hierarchical segmentation algorithm to partition these regions based on homogeneity and circularity, and a binary classifier to finalize the separation of nuclei from cytoplasm within the cell regions. Classification is posed as a grouping problem by ranking the cells based on their feature characteristics modeling abnormality degrees. The proposed procedure constructs a tree using hierarchical clustering, and then arranges the cells in a linear order by using an optimal leaf ordering algorithm that maximizes the similarity of adjacent leaves without any requirement for training examples or parameter adjustment. Performance evaluation using two data sets show the effectiveness of the proposed approach in images having inconsistent staining, poor contrast, and overlapping cells.

Novel Benchmark Database of Digitized and Calibrated Cervical Cells for Artificial Intelligence Based Screening of Cervical Cancer

“…They used adaptable threshold decision method to distinguish the cell from the cervical smear image, followed by using the maximal gray-level-gradient-difference method, proposed by them, for extraction of the nucleus from the cell. The comparative analysis of NCC detector with the earlier available methods reveled that NCC detector performs better than the edge enhancement nucleus and cytoplast contour detector model and the then gradient vector flow-active contour model [24] proposed an unsupervised approach for segmentation and classification of cervical cells obtained from Pap smear slides. The segmentation process involves providing an automatic threshold for separating the cell regions from the background, a multi-scale hierarchical segmentation algorithm to partition these regions based on homogeneity and circularity, and a binary classifier to finalize the separation of nuclei from cytoplasm within the cell regions.…”

A Colour Space Based Detection for Cervical Cancer Using Fuzzy C-Means Clustering

“…This calculation used precision and recall, where True Positive (TP), False Positive (FP), and False Negative (FN) will be entered into (6) and (7)as follows [13].…”

Automatic counting method for complex overlapping erythrocytes based on seed prediction in microscopic imaging

“…An adaptive marker-controlled watershed approach, aimed at improving the automatic extraction of markers was proposed by Tonti et al 15 It takes advantage of domain-speci¯c knowledge about the textural and geometrical characteristics of cells to reduce the sensitivity to uneven illumination and over-segmentation errors. Genctav et al 16 segmented overlapping cells by ranking the cells based on their feature characteristics computed from the nuclei and cytoplasm regions. The ranking was generated via linearization of the leaves of a binary tree that was constructed using hierarchical clustering.…”

“…9 Various automatic, computer-aided blood cell counting techniques have been proposed in recent decades. [10][11][12][13][14][15][16] A large number of experiments show that most of the existed methods are untenable with a low precision in the case of cells overlapping together (also called clustering, some cells clustering together and forming into a big area) which often appears in actual blood smear images. To optimize this shortcoming, an automatic segmentation of complex overlapping RBCs based on seed prediction is presented.…”