Quantitative Histomorphometrical Analysis in Distinguishing Non-Hodgkin Malignant Lymphoma, Large Cell Type with Nasopharyngeal Carcinoma

Authors

  • Okky Husain PPDS1 PA FK UNPAD
  • Hermin Aminah Usman Patologi Anatomi, Fakultas Kedokteran, Universitas Padjadjaran
  • Anglita Yantisetiasti Patologi Anatomi, Fakultas Kedokteran, Universitas Padjadjaran
  • Astri Handayani Sekolah Teknik Elektronik dan Informatika, Institut Teknologi Bandung
  • Bethy S Hernowo Patologi Anatomi, Fakultas Kedokteran, Universitas Padjadjaran

DOI:

https://doi.org/10.55816/mpi.v33i1.526

Keywords:

Computational pathology, digital pathology, Histomorphometry, lymphoma, malignant, nasopharyngeal carcinoma

Abstract

Background

Non-Hodgkin’s Malignant Lymphomas (NHMLs) with large cell morphology have similar histopathological appearance with Nasopharyngeal Carcinoma, undifferentiated type therefore ancillary examination such as Immunohistochemistry (IHC) are required. Quantitative histomorphometrical analysis able to identify subtle detail. The aim of this study is to measure the accuracy of quantitative histomorphometrical analysis to distinguishing NHML, large cell type with Nasopharyngeal Carcinoma.

 

Methods

Up to 33 Hematoxylin and Eosin (HE) slides from NHML, large cell type, and 27 cases of Nasopharyngeal Carcinoma were photographed under 400 times magnification with a total of 7131 images. Background, nuclei, and extra-nuclear components between cells are segmented as mathematical objects. Objects’ properties were measured and represented as distribution factors (minimum, quartile 1, median, quartile 3, maximum, range, and interquartile range). Significant factors that can differentiate the study groups are measured with a fisher-exact statistical test. Models to predict NHML, large cell type from nasopharyngeal carcinoma were formulated with different numbers of factors. The HE diagnosis accuracy compared to IHC was measured

 

Results

A sum of 7131 with 3893 from NHMLs, large cell type, and 3238 images of nasopharyngeal carcinoma were collected. After features selection, up to 76 properties factors are selected. Models' accuracy are ranging from 72.67% with a single factor and up to 89.9% with the whole selected factors.

 

Summary

Models’ accuracy are higher (72.6-89.9%) compared to HE primary diagnosis (60%) in distinguishing NHML, large cell againts nasopharyngeal carcinoma. Quantitative histomorphometry can assist pathologist for early screening to distinguishing NHML, large cell type with Nasopharyngeal Carcinoma.

Downloads

Download data is not yet available.

Downloads

Published

2024-01-31

Issue

Section

Articles