Tumor cell identification and classification in esophageal adenocarcinoma specimens by hyperspectral imaging

Marianne Maktabi,Susanne Braun,René Thieme,Claire Chalopin,Dietmar Lorenz,Yannis Wichmann,Michael Bange,Hannes Köhler,Ines Gockel,Henning Ahle

doi:10.1038/s41598-022-07524-6

Abstract

Esophageal cancer is the sixth leading cause of cancer-related death worldwide. Histopathological confirmation is a key step in tumor diagnosis. Therefore, simplification in decision-making by discrimination between malignant and non-malignant cells of histological specimens can be provided by combination of new imaging technology and artificial intelligence (AI). In this work, hyperspectral imaging (HSI) data from 95 patients were used to classify three different histopathological features (squamous epithelium cells, esophageal adenocarcinoma (EAC) cells, and tumor stroma cells), based on a multi-layer perceptron with two hidden layers. We achieved an accuracy of 78% for EAC and stroma cells, and 80% for squamous epithelium. HSI combined with machine learning algorithms is a promising and innovative technique, which allows image acquisition beyond Red–Green–Blue (RGB) images. Further method validation and standardization will be necessary, before automated tumor cell identification algorithms can be used in daily clinical practice.

Highlights

Esophageal cancer is the sixth leading cause of cancer-related death worldwide
esophageal adenocarcinoma (EAC) specimens with tumor cell rich areas, which were identified with conventional light microscopy, were selected to be analyzed by hyperspectral imaging (HSI)
Thereby, a differentiation of background (Bg), squamous epithelium (SE), EAC and tumor stroma cells can be done based on the synthesized RGB image from (Fig. 1)

Summary

Introduction

Esophageal cancer is the sixth leading cause of cancer-related death worldwide. Histopathological confirmation is a key step in tumor diagnosis. Hyperspectral imaging (HSI) data from 95 patients were used to classify three different histopathological features (squamous epithelium cells, esophageal adenocarcinoma (EAC) cells, and tumor stroma cells), based on a multi-layer perceptron with two hidden layers. Machine learning algorithms are powerful tools for cancer cell identification and classification using hyperspectral data[5] These algorithms combined with HSI technology have been applied for head and neck cancer[6], gastric c ancer[7], breast c ancer[8], and prostate c ancer[9]. HSI data with their complex and comprehensive hypercube structure are a predestined source for machine learning algorithms Several of these have been shown to support the identification and classification of cancer cells in HSI d ata[7,9–11]. We classified pixel-wise three different histopathological features (cells from squamous epithelium, EAC, and tumor stroma) based on machine learning methods. A multi-layer perceptron (MLP) with two hidden layers was used to separate the three classes in HSI images of histopathological specimens from 95 patients, who had undergone oncologic esophagectomy for EAC

Methods

Results

Conclusion