20 December 2018 in The Analyst
Current methods for cancer detection rely on tissue biopsy, chemical labeling/staining, and examination of the tissue by a pathologist. Though these methods continue to remain the gold standard, they are non-quantitative and susceptible to human error. Fourier transform infrared (FTIR) spectroscopic imaging has shown potential as a quantitative alternative to traditional histology. However, identification of histological components requires reliable classification based on molecular spectra, which are susceptible to artifacts introduced by noise and scattering. Several tissue types, particularly in heterogeneous tissue regions, tend to confound traditional classification methods. Convolutional neural networks (CNNs) are the current state-of-the-art in image classification, providing the ability to learn spatial characteristics of images. In this paper, we demonstrate that CNNs with architectures designed to process both spectral and spatial information can significantly improve classifier performance over per-pixel spectral classification. We report classification results after applying CNNs to data from tissue microarrays (TMAs) to identify six major cellular and acellular constituents of tissue, namely adipocytes, blood, collagen, epithelium, necrosis, and myofibroblasts. Experimental results show that the use of spatial information in addition to the spectral information brings significant improvements in the classifier performance and allows classification of cellular subtypes, such as adipocytes, that exhibit minimal chemical information but have distinct spatial characteristics. This work demonstrates the application and efficiency of deep learning algorithms in improving the diagnostic techniques in clinical and research activities related to cancer.
1 May 2016 in Kidney International
27 May 2016 in Biomedical Optics Express
FGFrancesca V. Gambacorta
The importance of stroma as a rich diagnostic region in tissue biopsies is growing as there is an increasing understanding that disease processes in multiple organs can affect the composition of adjacent connective tissue regions. This may be especially true in the liver, since this organ’s central metabolic role exposes it to multiple disease processes. We use quantum cascade laser infrared spectroscopic imaging to study changes in the chemical status of hepatocytes and fibrotic regions of liver tissue that result from the progression of liver cirrhosis to hepatocellular carcinoma and the potentially confounding effects of diabetes mellitus.
22 April 2014 in The Analyst
Infrared spectroscopic imaging provides micron-scale spatial resolution with molecular contrast. While recent work demonstrates that sample morphology affects the recorded spectrum, considerably less attention has been focused on the effects of the optics, including the condenser and objective. This analysis is extremely important, since it will be possible to understand effects on recorded data and provides insight for reducing optical effects through rigorous microscope design. Here, we present a theoretical description and experimental results that demonstrate the effects of commonly-employed cassegranian optics on recorded spectra. We first combine an explicit model of image formation and a method for quantifying and visualizing the deviations in recorded spectra as a function of microscope optics. We then verify these simulations with measurements obtained from spatially heterogeneous samples. The deviation of the computed spectrum from the ideal case is quantified via a map which we call a deviation map. The deviation map is obtained as a function of optical elements by systematic simulations. Examination of deviation maps demonstrates that the optimal optical configuration for minimal deviation is contrary to prevailing practice in which throughput is maximized for an instrument without a sample. This report should be helpful for understanding recorded spectra as a function of the optics, the analytical limits of recorded data determined by the optical design, and potential routes for optimization of imaging systems.
1 November 2017 in The International Journal of Biochemistry & Cell Biology
16 January 2018 in The Analyst
Tissue histology utilizing chemical and immunohistochemical labels plays an important role in biomedicine and disease diagnosis. Recent research suggests that mid-infrared (IR) spectroscopic imaging may augment histology by providing quantitative molecular information. One of the major barriers to this approach is long acquisition time using Fourier-transform infrared (FTIR) spectroscopy. Recent advances in discrete frequency sources, particularly quantum cascade lasers (QCLs), may mitigate this problem by allowing selective sampling of the absorption spectrum. However, DFIR imaging only provides a significant advantage when the number of spectral samples is minimized, requiring a priori knowledge of important spectral features. In this paper, we demonstrate the use of a GPU-based genetic algorithm (GA) using linear discriminant analysis (LDA) for DFIR feature selection. Our proposed method relies on pre-acquired broadband FTIR images for feature selection. Based on user-selected criteria for classification accuracy, our algorithm provides a minimal set of features that can be used with DFIR in a time-frame more practical for clinical diagnosis.