Breast histopathology using random decision forests-based classification of infrared spectroscopic imaging data

Abstract

Current methods for cancer detection rely on clinical stains, often using immunohistochemistry techniques. Pathologists then evaluate the stained tissue in order to determine cancer stage treatment options. These methods are commonly used, however they are non-quantitative and it is difficult to control for staining quality. In this paper, we propose the use of mid-infrared spectroscopic imaging to classify tissue types in tumor biopsy samples. Our goal is to augment the data available to pathologists by providing them with quantitative chemical information to aid diagnostic activities in clinical and research activities related to breast cancer.

A four class model for digital breast histopathology using high-definition Fourier transform infrared (FT-IR) spectroscopic imaging

“…RF is a parallel training algorithm making it suitable and efficient for large datasets. [11] Initially, different combinations of the chosen metrics were looked at in terms of their ability to separate the different classes. They were also used in the RF classifier for different models and different sections of the same dataset.…”

“…[4] FT-IR spectroscopy in combination with a multitude of multivariate techniques has been employed for more than a decade for disease diagnoses in various applications. [5]- [11] Since the spectrum at each pixel is used as molecular fingerprint, it is necessary to have a sufficiently high spatial resolution as well as acceptable signal to noise ratio (SNR) for this approach to be sensitive for complex histological studies. Breast tissue is heterogeneous both from the clinical and the pathological point of view [12], making it especially imperative to capture the subtle biochemical and morphological alterations of the tissue.…”

“…[10], [11], [14], [15] Unsupervised classification methods fail to separate between inter and intra class variance so although they can be good at partitioning the data, supervised algorithms are more useful for accurate and robust tissue analysis. [16] This study reports a chemistry-based approach to augment digital methods for quantitative evaluation of disease.…”

High-definition Fourier transform infrared spectroscopic imaging of prostate tissue

“…It does not make any assumptions about the data and is very effective in exploring its heterogeneity. On the other side of analysis approaches lie supervised classification methods, which require a defined model of the groups present in the dataset 13 . Among many supervised classification methods, the Random Forest (RF) scheme is gaining interest, mostly due to its robustness and simplicity of optimization, since the main tuning parameter is the number of trees grown during the training stage 14 .…”

Automated osteosclerosis grading of clinical biopsies using infrared spectroscopic imaging

“…Infrared (IR) absorbance is often used to identify molecular signatures in organic materials. Fourier transform infrared (FTIR) has been applied for label free characterization and classification in histopathological studies [10][11][12][13][14][15][16] and has the potential to automate examinations to both save time and reduce diagnostic errors. This paper focuses on evaluating the potential for absorbance spectroscopy to automate osteosclerosis and collagen grading, particularly trabecular bone area (TBA), by defining a clinically-viable protocol for image acquisition and analysis.…”