KEYWORDS: Data modeling, Statistical analysis, Signal processing, Statistical modeling, Spectral models, Reliability, Probability theory, Overfitting, Modal decomposition, Information fusion
Spectral decomposition, a pivotal process in hyperspectral imaging, involves separating mixed signals into their constituent parts, known as endmembers, to extract meaningful information. The Bayesian Information Criterion, a statistical metric derived from Bayesian probability theory, serves as a valuable tool for model selection in spectral decomposition reducing the risk of overfitting and enhancing the robustness of the unmixing analysis.
In this work we utilise BIC in spectral decomposition through fitting models with varying numbers of endmembers and assessing the trade-off between model complexity and data fidelity, allowing the selection of the most parsimonious representation that best captures the underlying structure of the spectral data. This methodology results is a more refined and interpretable spectral decomposition, aiding in molecular interpretation of data science models in chemical imaging.
The identification of spectral markers differentiating disease states when using spectral data is challenging in the context of modelling with deep neural networks, particularly in scenarios where classification models are developed with multiple classes.
While a number of approaches do exist which can provide an insight into the features which are learnt by deep learning models, in biophotonics and chemical imaging these have received relatively little attention. In the present work we pilot the use of Fourier Transform chemical imaging with two deep-learning interpretation approaches within the context of a multi-class classification problem. Fully connected neural networks are developed on unfolded chemical imaging data captured on patient-derived xenografts developed from a colorectal cancer model. Separately, Shapley additive explanations and saliency approaches are used to derive feature sets which are discriminatory for class within this experimental model of colorectal cancer.
Preliminary results suggest that Shapley additive explanations provide differentiating spectral sets which may not be derived with saliency, although the feature sets which are identified are dependent upon spectral pretreatment methodology. A dual approach which employs both strategies may be an effective strategy for the identification of feature sets in this context.
Clinical pathological diagnosis and prognosis for cancer is often confounded by spatial tissue heterogeneity. This study investigates the utility of entropy as a robust quantitative metric of spatial disorder within Fourier Transform Infrared (FTIR) chemical images of breast cancer tissue. The use of entropy is grounded in its capacity to encapsulate the complexities of pixel-wise spectral intensity distributions, thus providing a detailed assessment of the spatial variations in biochemistry within tissue samples.
Here we explore the use of Shannon’s entropy as a single image-based metric of spectral biochemical heterogeneity within FTIR chemical images of breast cancer tissue. This metric was then analyzed statistically with respect to hormone receptor status. Our results suggest that while entropy effectively captures the heterogeneity of tissue samples, its role as a standalone predictor for diagnostic subtyping may be limited without considering additional variables or interaction effects. This work emphasizes the need for a multifaceted approach in leveraging entropy with chemical imaging for diagnostic subtyping in cancer.
The rapid advancement of imaging technologies in pathology has ushered in an era of data-intensive diagnostic workflows, generating large volumes of data that demand sophisticated segmentation and compression techniques. Chemical imaging approaches offer an all-digital objective approach to pathological analysis, though image segmentation is required for efficient computation.
Convolutional autoencoders are highly connected deep learning networks which can learn salient features within imaging data for the purposes of compression, data recovery, development of classifiers and/or segmentation.
In this study an objective analysis of a U-Net convolutional autoencoders for unsupervised image segmentation is conducted with respect to haematoxylin-eosin based ground-truth diagnostic pathology. We find that a light-weight network architecture may provide a suitable segmentation approach for chemical imaging.
In the realm of ex-vivo diagnosis, the integration of optical and chemical imaging data has emerged as a transformative approach, offering a comprehensive understanding of biological specimens at a molecular level. Chemical imaging of human tissue specimens provides an all-digital label-free approach to imaging in objective histopathology, though it requires reference to gold standard pathological (e.g. haematoxylin and eosin (H+E) stained) images for pathological interpretation.
Optical imaging techniques, such as microscopy and spectroscopy, provide detailed spatial information, capturing morphological features with high resolution. Concurrently, chemical imaging methods, including mass spectrometry and Raman spectroscopy, offer insights into molecular composition. The challenge lies in harnessing the complementary strengths of these disparate modalities to extract a holistic understanding of the sample.
In this work we present the results of several image alignment approaches for fusion and integration of chemical and pathological imaging data, demonstrating that the process of corner detection is crucial towards precise image alignment.
MicroRNAs are small ~22 nucleotide RNA sequences that are guided to the 3’ untranslated region (UTR) of protein-coding target mRNA sequences. One particular microRNA, miR155, plays a remarkable role in the immune system, where it is essential for mounting appropriate immune responses. However, its dysregulation has been identified in multiple inflammatory disorders such as Multiple Sclerosis (MS), arthritis, psoriasis and colitis. More specifically, miR-155 has been found to be elevated in the serum and brain lesions of MS patients. Importantly, therapeutic inhibition of miR-155 can inhibit progression of the MS disease model. One of us has identified that macrophages are major contributor to miR-155 elevation in the MS disease model, whilst its inhibition specifically in macrophages can limit the disease. Here macrophages were isolated from the femur and tibia of wild-type (WT) mice and mice with a knock-out (KO) of the gene regulating miR-155 production, and were cultured in-vitro and stimulated with lipopolysaccharide (LPS) to simulate an immune response. Cells were then prepared for spectral analysis by FTIR imaging with a Perkin-Elmer Spotlight 400 imaging microscope. After pre-processing the dimensionality of spectra were reduced using principal components analysis, kernel-PCA and universal manifold application and projection (UMAP) and classified using a support vector machine algorithm, delivering a classification performance approaching F1~0.89. Spectral features differentiating WT and KO classes were observed across the fingerprint region with no single spectral marker being the sole source of differentiation of the downstream molecular events. This study exemplifies the challenge in spectral discrimination of the complexity of molecular events in ex-vivo models of immune dysregulation.
Multi-modal spectroscopic analysis of biological systems may offer an improved overall non-invasive biophotonic metric of the status of the system, further enhancing the diagnostic and prognostic capabilities of these technologies. In the present study macrophages were extracted from wild-type mice and mice with a knock-out of the gene regulating miR-155, which has been observed to occur in patients with various autoimmune disorders, including multiple sclerosis (MS). Macrophages were stimulated in-vitro to produce an immune response and were then screened spectroscopically with FTIR and Raman spectroscopy (at 532nm and 660nm). Low, medium and high level data fusion strategies for classification of response to stimulation and miRNA regulation were piloted, using downstream principal components analysis-support vector machine classifiers to test the impact of these strategies on classification performance. These techniques allowed the development of a combined highlevel data-fusion, classification pipeline with a high level of classification accuracy (F1<0.9), with reduced variability in performance. Our proposed spectroscopic assay-data fusion strategy may provide an adjunct to clinical screening and diagnosis of various autoimmune disorders whose aetiology is associated with genetic dysregulation.
Severe radiation toxicity can continue years after the completion of radiotherapy for prostate cancer patients. Currently, it is impossible to predict before treatment which patients will experience these long-term side effects. New approaches based on vibrational spectroscopy have advantages over lymphocyte and genomic assays in terms of minimal sample preparation, speed and cost. A high throughput method has been developed to measure Raman spectra from liquid plasma in a cover glass bottomed 96 well plate. However, the Raman spectra can show contributions from glass and water. The current study aims to optimise pre-processing steps to improve classification performance.
Due to its high lateral resolution, Raman microspectrsocopy is rapidly becoming an accepted technique for
the subcellular imaging of single cells. Although the potential of the technique has frequently been
demonstrated, many improvements have still to be realised to enhance the relevancy of the data collected.
Although often employed, chemical fixation of cells can cause modifications to the molecular composition
and therefore influence the observations made. However, the weak contribution of water to Raman spectra
offers the potential to study live cells cultured in vitro using an immersion lens, giving the possibility to
record highly specific spectra from cells in their original state. Unfortunately, in common 2-D culture
models, the contribution of the substrates to the spectra recorded requires significant data pre-processing
causing difficulties in developing automated methods for the correction of the spectra. Moreover, the 2-D
in vitro cell model is not ideal and dissimilarities between different optical substrates within in vitro cell
cultures results in morphological and functional changes to the cells. The interaction between the cells and
their microenvironment is crucial to their behavior but also their response to different external agents such
as radiation or anticancer drugs. In order to create an experimental model closer to the real conditions
encountered by the cell in vivo, 3-D collagen gels have been evaluated as a substrate for the spectroscopic
study of live cells. It is demonstrated that neither the medium used for cell culture nor the collagen gels
themselves contribute to the spectra collected. Thus the background contributions are reduced to that of the
water. Spectral measurements can be made in full cell culture medium, allowing prolonged measurement
times. Optimizations made in the use of collagen gels for live cells analysis by Raman spectroscopy are
encouraging and studying live cells within a collagenous microenvironment seems perfectly accessible.
Raman spectroscopy, as an evaluation of the products of ionising radiation exposure in biological systems, has been utilised mainly in the evaluation of the impact of exposure in tissue, cellular constituents and live animals. It has also been recently demonstrated that Raman spectroscopy can demonstrate key spectroscopic changes in the live cell associated with significant apoptotic and necrotic chemical damage. The present preliminary work utilises Raman spectroscopy at 514.5 nm to evaluate the results of exposure to γ-rays in HaCaT cells from a Co-60 therapy source, in tandem with other biological assays. The results demonstrate that Raman spectral changes may be correlated with changes in the cell also identified in parallel biochemical assays.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.