Infrared spectral histopathology (SHP): a novel diagnostic tool for the accurate classification of lung cancer

Abstract

We report results of a study utilizing a recently developed tissue diagnostic method, based on label-free spectral techniques, for the classification of lung cancer histopathological samples from a tissue microarray. The spectral diagnostic method allows reproducible and objective diagnosis of unstained tissue sections. This is accomplished by acquiring infrared hyperspectral data sets containing thousands of spectra, each collected from tissue pixels about 6 mm on edge; these pixel spectra contain an encoded snapshot of the entire biochemical composition of the pixel area. The hyperspectral data sets are subsequently decoded by methods of multivariate analysis, which reveal changes in the biochemical composition between tissue types, and between various stages and states of disease. In this study, a detailed comparison between classical and spectral histopathology (SHP) is presented, which suggests SHP can achieve levels of diagnostic accuracy that is comparable to that of multi-panel immunohistochemistry. KEYWORDS: artificial neural network analysis; histopathology; immunohistochemistry; lung cancer; spectral histopathology This paper reports a large-scale study of a new technology to classify four common forms of lung cancers, and distinguish them from normal tissues. The new methodology introduced here utilizes optical measurements on unstained tissue 1 for spectral data acquisition, and does not utilize any immunohistochemical or other stains or labels for classification. As the diagnostic procedure is instrument based and utilizes trained computer algorithms for classification, this method offers reproducibility, complete objectivity and improved accuracy over present methodology.Optical methods have been used in histology and pathology ever since these methods were first described. After all, staining tissues or cells by hematoxylin/eosin (H&E), followed by (visual) microscopic examination is a form of spectral analysis: different compartments of the cell respond differently to basophilic and eosinophilic stains and thus, allow a 'spectral analysis' using the eye as a detector. This method can reveal an amazing amount of information but is inherently subjective. More recent optical methods have used image capture at a few selected wavelengths, and computer analysis of the image planes, for tissue analysis. 2 Immunohistochemistry (IHC), to date the most advanced optical method to detect the presence of certain cancer signatures or markers 3,4 uses detection of specific antibodies labeled with easily observable stains.The new approach reported here is based on the observation of inherent spectral signatures (as opposed to any external stains or labels used to treat the sample) of cellular components to aid classical cytopathology and histopathology. 5 The paradigm for the spectral approach is that the transition from normal tissue or cells to diseased states is accompanied by changes in the overall biochemical composition of the tissue, along with well-known changes in cellular morphology and tissue archite...

Molecular pathology via IR and Raman spectral imaging

“…The overall accuracy is comparable to the results obtained in spectral histopathology (SHP) for the distinction of certain lung cancers [72]. In particular, we wish to emphasize that these results here were obtained from cells that were morphologically normal; that is, from cells that make up the vast majority in exfoliated cell samples.…”

“…This study follows several other references in the literature that have established that spectral diagnostic methods can identify the sites of primary tumors in metastatic cancers in the brain [71], differentiate metastatic cancers from different primary tumors in lymph nodes differentiate between different tumors in the same organ [72] detect spectral changes due to overexpression of individual cancer markers [73]. Thus, we wanted to establish whether SHP is able to differentiate the activation of lymphocytes in lymph nodes due to different primary cancers.…”

“…A number of such predictive algorithms exist, some very similar to the principles of PCA (for example, Soft Independent Modeling of Class Analogy, SIMCA), linear discriminant analysis, LDA, to more complicated classifiers such as Support Vector Machines, SVMs, which are based on separating spectral classes by complicated, multidimensional separation planes. At the LSpD, artificial neural networks have been used [72] for the trained prediction of class memberships.…”

“…Yet, the correct diagnosis is highly important, since different treatment regimens are generally called for. In 2012, we reported an 80-patient pilot study [72] to determine whether these three cancer types could be differentiated and diagnosed by SHP, based on different chemical composition and without the use of morphological criteria or advanced methods such as immunohistochemistry (IHC) or gene array technology. Since SHP is a non-destructive method that works on unstained histopathological specimens, it would offer a much reduced workflow of analysis over other methods.…”

“…The aforementioned pilot study for lung cancer diagnostics [72] also included tissue spots with small cell carcinomas, as well as normal tissue. Discrimination of all tissue types was accomplished by training hierarchical neural networks consecutively; the order of discrimination was based on the variance in the dataset.…”