Abstract
Nasopharyngeal carcinoma is one of the malignant neoplasm with high incidence in China and south-east Asia. Ki-67 protein is strictly associated with cell proliferation and malignant degree. Cells with higher Ki-67 expression are always sensitive to chemotherapy and radiotherapy, the assessment of which is beneficial to NPC treatment. It is still challenging to automatically analyze immunohistochemical Ki-67 staining nasopharyngeal carcinoma images due to the uneven color distributions in different cell types. In order to solve the problem, an automated image processing pipeline based on clustering of local correlation features is proposed in this paper. Unlike traditional morphology-based methods, our algorithm segments cells by classifying image pixels on the basis of local pixel correlations from particularly selected color spaces, then characterizes cells with a set of grading criteria for the reference of pathological analysis. Experimental results showed high accuracy and robustness in nucleus segmentation despite image data variance. Quantitative indicators obtained in this essay provide a reliable evidence for the analysis of Ki-67 staining nasopharyngeal carcinoma microscopic images, which would be helpful in relevant histopathological researches.
Highlights
Nasopharyngeal carcinoma (NPC) is one of the common cancers that occupies highest incidence rates in China and south-east Asia
We propose an automated image processing pipeline based on kmeans clustering without training samples needed, which is helpful for in-depth analysis of NPC tissue microstructures in clinic
All mouse procedures were approved by the Institutional Animal Care and Use Committees (IACUC) of Fujian Provincial Cancer Hospital and performed in accordance with institutional policies
Summary
Peng Shi1,*, Jing Zhong2,*, Jinsheng Hong[3], Rongfang Huang[4], Kaijun Wang1 & Yunbin Chen[2,5]. It is important to diagnose the malignant degree of tumor based on Ki-67 expressions, which makes IHC staining of Ki-67 an efficient tool for NPC cell characterization. The color distributions in Ki-67 staining images are always extremely uneven, which makes nuclei have irregular and unclear boundaries It is difficult for traditional image segmentation methods based on thresholding or morphological models to detect nucleus boundaries and quantify cells precisely. Taking single pixels as study objects, machine learning based methods classify pixels sharing similar characteristics as the same group, which perform the segmentation of Ki-67 staining image more efficiently, and measurements derived from adjusted nucleus boundaries are provided for further pathological analysis after post-processing. We propose an automated image processing pipeline based on kmeans clustering without training samples needed, which is helpful for in-depth analysis of NPC tissue microstructures in clinic. With morphological post-processing, precise boundaries of all segmented nuclei were obtained for quantitative analysis in pathological researches
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
