Level: Beginner
Presentation title: JMP as a tool for spatial data and morphological analysis - Grading upper airway intraepithelial neoplasia and its precursor lesions using shortest distance cluster analysis
Most malignant tumors of the upper respiratory tract mucosa, such as the oral cavity, pharynx, and larynx, are squamous cell carcinomas that occur in a layer of cells called stratified squamous epithelium that covers the surface of the mucosa. Furthermore, it is known clinically that lesions thought to be in the pre- or early stages of this are found as white or red patches on the mucosa, and based on the observation of microscopic specimens of tissue taken from patients, they are named epithelial dysplasia and carcinoma in situ, respectively. Dysplasia is further divided into three grades, from mild to severe, depending on the abnormality in cell morphology and the proportion of these in the epithelial layer. This differentiation is performed intuitively by pathologists and is thought to have a certain degree of reproducibility, but there have been few objective studies. In this study, we attempted to quantify the arrangement of cells (nuclei) in the epithelial layer and examine the differences between non-neoplastic (normal), dysplasia, and carcinoma in situ. Digital image analysis was used to extract the coordinates of the center of gravity of cell nuclei on micrographs, and minimum spanning trees (MST) connecting each center of gravity were generated using hierarchical cluster analysis using the shortest distance method in JMP ver. 15. Histograms of the lengths of each branch were compared, and differences were found between each group.
Ryoji Senba
Former lecturer at Tohoku University (Institute of Development, Aging and Cancer, Department of Pathological Organ Structure Research). Doctor of Medicine. Former Ministry of Education overseas research fellow (Medicine) (Aarhus University, Kingdom of Denmark).
In the field of human pathology, he has studied the pathogenesis of diseases using quantitative morphology that applies geometric probability theory and integral geometry, digital image analysis, and multivariate statistical analysis. He has published research papers on early cancers occurring in liver cirrhosis, alveolar epithelium, pancreatic ductal epithelium, and endometrium, as well as their precursor lesions and liver metastasis.
(https://pubmed.ncbi.nlm.nih.gov/7804428/ , https://pubmed.ncbi.nlm.nih.gov/7804429/, https://pubmed.ncbi.nlm.nih.gov/8402446/ , https://pubmed.ncbi.nlm.nih.gov/8135625/, https://pubmed.ncbi.nlm.nih.gov/7840839/ , https://pubmed.ncbi.nlm.nih.gov/10560494/)
He is interested in mathematical methods that can be applied to the analysis of cancer development processes and their tissue diagnosis, with a particular interest in numerical classification methods such as cluster analysis and discriminant analysis. As a statistical analysis platform, he has used Fortran statistical subroutines on a mainframe, SPSS and SYSTAT on a PC, and others, but has been using JMP since version 8, focusing on its excellent data table functions and flexible analytical environment.
Chiba Rui
Graduated from the Department of Complex Systems Intelligence, School of Systems Information Science, Future University Hakodate. During his time at university, he was interested in complex phenomena in physicochemical reactions, and conducted experiments and research on the capsule formation mechanism of thermally polymerized amino acids in an alcohol liquid phase. He is currently interested in morphological and image recognition and classification using digital image analysis, data science, and neural networks.
