Evaluation of Three Algorithms for the Segmentation of Overlapping Cervical Cells

Abstract

In this paper, we introduce and evaluate the systems submitted to the first Overlapping Cervical Cytology Image Segmentation Challenge, held in conjunction with the IEEE International Symposium on Biomedical Imaging 2014. This challenge was organized to encourage the development and benchmarking of techniques capable of segmenting individual cells from overlapping cellular clumps in cervical cytology images, which is a prerequisite for the development of the next generation of computer-aided diagnosis systems for cervical cancer. In particular, these automated systems must detect and accurately segment both the nucleus and cytoplasm of each cell, even when they are clumped together and, hence, partially occluded. However, this is an unsolved problem due to the poor contrast of cytoplasm boundaries, the large variation in size and shape of cells, and the presence of debris and the large degree of cellular overlap. The challenge initially utilized a database of 16 high-resolution ( ×40 magnification) images of complex cellular fields of view, in which the isolated real cells were used to construct a database of 945 cervical cytology images synthesized with a varying number of cells and degree of overlap, in order to provide full access of the segmentation ground truth. These synthetic images were used to provide a reliable and comprehensive framework for quantitative evaluation on this segmentation problem. Results from the submitted methods demonstrate that all the methods are effective in the segmentation of clumps containing at most three cells, with overlap coefficients up to 0.3. This highlights the intrinsic difficulty of this challenge and provides motivation for significant future improvement.

DeepPap: Deep Convolutional Networks for Cervical Cell Classification

“…On the Herlev benchmark dataset [16], [17], the attained nucleus segmentation accuracy ranging between 0.85 [11] and 0.92 [15]. On an overlapping cervical cell dataset [13], the cytoplasm segmentation accuracy ranges from 0.87 to 0.89 [13]. On the other hand, most cell classification studies assume that accurate segmentations of individual cytoplasms and nuclei are already available [17]- [19].…”

Automatic segmentation of overlapping cervical smear cells based on local distinctive features and guided shape deformation

“…Moreover, the obtained T P p was 0.89, which was still higher than the T P p value obtained by [54] ISBI approach. For more evaluation, a qualitative comparison of the nuclei and cytoplasm segmentation of our approach and the ISBI approaches is provided in Figure 10 using the visual results provided by the ISBI organizers [56]. As seen in the figure, the proposed approach provided the best, and semi-optimal, segmentation for the overlapping cells in the first image.…”

“…Qualitative evaluation of the proposed approach and the ISBI challenge methods[56]. mation process, whereas the refinement and filtering time is the execution time for DRLSE-based refinement step and the false candidates filtering step.The proposed approach took 18 seconds on average to classify the Pap smear image to its cellular components, and approximately 10 seconds to segment each cell in the image.…”

Radial feature descriptors for cell classification and recommendation

“…Alternative unsupervised machine learning schemes in [21,26] proposed graph-based algorithms to segment the nucleus and cytoplasm of cervical cells with different degrees of overlap in seconds. Despite those methods continuing to be among the best algorithms, as discussed in [24], accuracies drop drastically when applied to conventional Pap smears. Section 4 presents unsupervised machine learning schemes that evolved from this previous work and addresses digitized Pap tests from real-world scenarios.…”

“…Improvements on real cells segmentation [5,24] have shown new potential to analyze cervical cell using computer-aided systems for Pap tests.…”