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.
Discrimination of malignant and non-malignant cells of histopathologic specimens is a key step in cancer diagnostics. Hyperspectral Imaging (HSI) allows the acquisition of spectra in the visual and near-infrared range (500-1000nm). HSI can support the identification and classification of cancer cells using machine learning algorithms. In this work, we tested four classification methods on histopathological slides of esophageal adenocarcinoma. The best results were achieved with a Multi-Layer Perceptron. Sensitivity and F1-Score values of 90% were obtained.
Hyperspectral imaging (HSI), as recently applied in medicine, is a novel technology combining imaging with spectroscopy. It might be used to identify, classify and discriminate malignant and non-malignant cells of histopathologic specimens. HSI allows the determination of a spectrum between the visual and near-infrared light (500-1000 nm).
After surgical resection, specimens (n = 96) of Barrett’s cancer were fixed in 4% formaldehyde and slices were conducted (3 μm), which were stained with hematoxylin and eosin (HE). Differences in the absorbance of squamous epithelium and esophageal adenocarcinoma (EAC) cells were determined at eosin’s and hematoxylin’s maximal absorption of 530 nm and 590 nm. A classification algorithm, which combines a multilayer perceptron (MLP) with a Gaussian filter and data reduction based on principal component analysis (PCA) was used for discrimination.
For the first time in literature, we were able to analyze esophageal adenocarcinoma, tumor stroma and squamous epithelium cells by HSI. For both, the squamous epithelium and the esophageal adenocarcinoma cells, the intragroup variances were quite low. A set of 20 specimens was used in a testing phase, which showed MLP by using reflectance data to provide the best results. Leave-one-patient-out-cross validation for all 96 specimens showed an accuracy of 77% with a sensitivity of 87% and a specificity of 72% for EAC determination and visualization.
Squamous epithelium and EAC cells show specific spectral alterations due to their HE-staining, when measured by HSI. However, the training algorithms need further validation to foster a semi-automatic decision-making process in histopathological tumor cell identification.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.