Automatic Detection of Granuloma Necrosis in Pulmonary Tuberculosis Using a Two-Phase Algorithm: 2D-TB.

Pelin Kus,Metin N Gurcan,Gillian Beamer

doi:10.3390/microorganisms7120661

Pelin Kus, Metin N Gurcan + Show 1 more

Open Access

https://doi.org/10.3390/microorganisms7120661

Copy DOI

Journal: Microorganisms	Publication Date: Dec 7, 2019
Citations: 7	License type: CC BY 4.0

Affiliation: Wake Forest University, Tufts University

Abstract

Granuloma necrosis occurs in hosts susceptible to pathogenic mycobacteria and is a diagnostic visual feature of pulmonary tuberculosis (TB) in humans and in super-susceptible Diversity Outbred (DO) mice infected with Mycobacterium tuberculosis. Currently, no published automated algorithms can detect granuloma necrosis in pulmonary TB. However, such a method could reduce variability, and transform visual patterns into quantitative data for statistical and machine learning analyses. Here, we used histopathological images from super-susceptible DO mice to train, validate, and performance test an algorithm to detect regions of cell-poor necrosis. The algorithm, named 2D-TB, works on 2-dimensional histopathological images in 2 phases. In phase 1, granulomas are detected following background elimination. In phase 2, 2D-TB searches within granulomas for regions of cell-poor necrosis. We used 8 lung sections from 8 different super-susceptible DO mice for training and 10-fold cross validation. We used 13 new lung sections from 10 different super-susceptible DO mice for performance testing. 2D-TB reached 100.0% sensitivity and 91.8% positive prediction value. Compared to an expert pathologist, agreement was 95.5% and there was a statistically significant positive correlation for area detected by 2D-TB and the pathologist. These results show the development, validation, and accurate performance of 2D-TB to detect granuloma necrosis.

Highlights

Tuberculosis (TB) is diagnosed in 9–10 million human patients and causes 1–2 million deaths each year, surpassing mortality due to HIV/AIDS [1]
Granuloma, and lung tissue necrosis cause symptoms in pulmonary TB patients and contribute to transmission of M. tuberculosis bacilli [10,30], we developed and validated a method that allows investigators to automatically detect necrosis
We focused on the pattern of cell-poor necrosis within M. tuberculosis granulomas

Summary

Introduction

Tuberculosis (TB) is diagnosed in 9–10 million human patients and causes 1–2 million deaths each year, surpassing mortality due to HIV/AIDS [1]. Much research has identified pathways that trigger necrosis, for example: Intracellular bacterial replication and macrophage lysis; hypersensitivity to M. tuberculosis cell wall components and induction of necrotizing lipid pneumonia; vascular thrombosis resulting in infarction and necrosis due to hypoxia [11,16,20,21,22,23,24,25,26,27] These events converge on two visual patterns of necrosis in histological tissue sections, as interpreted by pathologists using light microscopes. A board-certified veterinary pathologist (GB) manually annotated 29 granulomas, 40 regions of cellular and nuclear debris, and 136 regions of cell-poor necrosis in lungs from 8 different super-susceptible DO mice. Annotations on the remaining 10 slides, containing 1–2 lung lobes from 10 different super-susceptible DO mice (which were not used in training) were used to test accuracy as compared to a board-certified veterinary pathologist (GB)

Measures of Accuracy to Assess 2D-TB Performance

Automatic Granuloma Detection

Detection of Cell-Poor Necrosis in Granulomas Using Machine Learning

Features for Detection of Cell-Poor Necrosis

Findings

Discussion

Conclusions