Sample size planning (SSP) is crucial for experimental planning but is not well-established for spectroscopic and image data, especially in combination with deep learning. The existing approaches are typically quite complex for routine use in experimental planning. To make the existing approaches more accessible, we developed web-based tools for the existing approaches. Besides, we extended the approach to imaging data and deep learning by introducing transfer learning in the SSP pipeline.
ACKNOWLEDGMENT:
Financial support from the EU, the TMWWDG, the TAB, the BMBF, the DFG, the Carl-Zeiss Foundation, and the Leibniz Association is greatly acknowledged. This work is supported by the BMBF, funding program Photonics Research Germany (LPI-BT3-IPHT, FKZ: 13N15708) and is integrated into the Leibniz Center for Photonics in Infection Research (LPI). The LPI initiated by Leibniz-IPHT, Leibniz-HKI, Friedrich Schiller University Jena, and Jena University Hospital is part of the BMBF national roadmap for research infrastructures.
Deep learning models are widely used because of their high accuracy in solving classification problems in spectroscopy, but they lack interpretability. The challenge lies in the balance between interpretability and accuracy. Current interpretive methods, such as SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), can sometimes provide mathematical meaning but physically implausible interpretations by perturbing individual feature values. To address this gap, our research proposes a group-focused methodology that targets 'spectral zones' to estimate the impact of collective spectral features directly. This approach enhances the interpretability of deep learning models, diminishes noisy data, and provides a more comprehensive understanding of model behaviors. By applying group perturbations, the resultant interpretations are not only more intuitive but also offer results that are easily comparable with domain expertise, thus leading to an enriched analysis of the model's decision-making processes.
Optical microscopy is widely applied to the investigation of biomedical samples, and a variety of image processing approaches have been established to reduce artifacts generated by the measurement process. However, a standardized and reliable method for assessing image quality is still lacking. Our study contributes to the investigation of image evaluation methods for fluorescence microscopy. We present a set of no-reference metrics that can be used for the characterization of experimental artifacts. In addition, our method is incorporated into a machine learning approach for automatic classification of single artifacts. The metrics identify reliable markers for single artifacts in fluorescence microscopy measurements, can be easily interpreted, and allow the selection of the best image based on specific quality requirements. Our study provides a simple evaluation tool for optical microscopy that can also be extended to the different stages of the processing pipeline.
Transfer learning is an important technique to improve the model generalizability, which is crucial to apply Raman spectroscopy in biological applications considering the substantial spectral variations between batches. We systematically investigated the limit of existent model transfer methods and explored the possibilities of deep learning-based approaches in cases of limited sample size.
Photonic techniques are optimal tools to characterise samples in various research disciplines such as remote sensing, materials characterisation, life sciences and medicine. To exploit the full potential of these techniques, the entire data life cycle of photonic data needs to be investigated and optimised. The photonic data lifecycle starts with data generation and planning of the corresponding study/experiment, followed by data modelling using artificial intelligence (AI) techniques such as chemometrics, machine learning (ML) and deep learning (DL), and it ends by data storage and archiving.
In this contribution, we will present our studies aimed at the generation of correction procedures and inverse modelling tools for photonic data and heir measurement processes using data science methods. We will also present our research activities towards a repository for sharing vibrational spectroscopic data (VibSpecDB), which is embedded in the National Research Data Infrastructure Initiative in Germany (NFDI) and its chemistry consortium (NFDI4Chem).
Acknowledgements
This work is supported by the BMBF, funding program Photonics Research Germany (13N15466 (LPI-BT1-FSU), 13N15710 (LPI-BT3-FSU), 13N15708 (LPI-BT3-IPHT)) and is integrated into the Leibniz Center for Photonics in Infection Research (LPI). The LPI initiated by Leibniz-IPHT, Leibniz-HKI, Friedrich Schiller University Jena and Jena University Hospital is part of the BMBF national roadmap for research infrastructures. Parts are funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - 441958208 (NFDI4Chem).
We developed a simple and convenient magnetic bead-based sample preparation scheme for enabling a Raman spectroscopic differentiation of severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) positive and negative samples. By utilizing the angiotensin-converting enzyme 2 (ACE2) receptor protein as selective recognition element, we avoid having to identify the virus species based on its specific Raman signature. Instead we only need to verify the presence of the virus, which is significantly less difficult. For quantitative evaluation of the spectra, we calculated the Pearson coefficient and the Normalized Cross Correlation coefficient.
Raman spectroscopy is a label-free, non-invasive spectroscopic technique, which can be utilized for many biomedical and diagnostic investigations. To do so, chemometric modelling strategies are used, but they lead to a low generalizability of the models. To tackle this issue we investigated transfer learning (TL) approaches for deep learning (DL) based modelling of Raman spectra for classification of three bacterial spore species. In initial test we found that TL can facilitate the usage of DL for time-consuming measurement modalities, because it can help to deal with low dataset sizes.
Photonic data can be used to characterize the biochemical composition of samples and often in a non-destructive and label-free manner. To utilize these label-free measurements for applications like diagnostics or analytics, data driven modeling is utilized to translate photonic data into higher-level information. In this contribution, two scenarios of data driven modeling will be presented. We will present the translation of nonlinear multi-contrast images into diagnostic information like tissue types, disease types, and histopathological stainings. Additionally, we will demonstrate deep learning as tool for the extraction of the imaginary part of the third-order susceptibility of spectral CARS measurements.
Based on CARS-SHG spectroscopy biomolecular fingerprints of lipids/proteins were distinguished in isolated adult cardiomyocytes of α-Gal-A-Knockout and wild-type mice opening new prospects for diagnostic of cardiac manifestations of Morbus Fabry.
Significance: The potential of fluorescence lifetime imaging microscopy (FLIM) is recently being recognized, especially in biological studies. However, FLIM does not directly measure the lifetimes, rather it records the fluorescence decay traces. The lifetimes and/or abundances have to be estimated from these traces during the phase of data processing. To precisely estimate these parameters is challenging and requires a well-designed computer program. Conventionally employed methods, which are based on curve fitting, are computationally expensive and limited in performance especially for highly noisy FLIM data. The graphical analysis, while free of fit, requires calibration samples for a quantitative analysis.
Aim: We propose to extract the lifetimes and abundances directly from the decay traces through machine learning (ML).
Approach: The ML-based approach was verified with simulated testing data in which the lifetimes and abundances were known exactly. Thereafter, we compared its performance with the commercial software SPCImage based on datasets measured from biological samples on a time-correlated single photon counting system. We reconstructed the decay traces using the lifetime and abundance values estimated by ML and SPCImage methods and utilized the root-mean-squared-error (RMSE) as marker.
Results: The RMSE, which represents the difference between the reconstructed and measured decay traces, was observed to be lower for ML than for SPCImage. In addition, we could demonstrate with a three-component analysis the high potential and flexibility of the ML method to deal with more than two lifetime components.
Conclusions: The ML-based approach shows great performance in FLIM data analysis.
Raman spectroscopy has been applied to investigate the suitability of the drop coating deposition technique to study plasma samples from healthy donors and a patient with underlying cardiac condition. When blood plasma is deposited on a solid substrate, a droplet with coffee-ring is formed, and the plasma proteins will distribute inhomogeneously depending on the chemical and physical properties of the proteins. Changes in the fingerprint region of the Raman spectra were observed from the outer-ring and central zone of the droplet through a systematic investigation. For complete characterization of the sample, optimum measurement scheme has been proposed. To obtain clinically relevant information of the effects of immunoadsorption (IA) treatment of dilated cardiomyopathy (DCM) patient’s, Raman spectral information from outer-ring as well as from the central zone is required.
Although chemotherapeutics for cancer treatment are becoming increasingly efficient these days, they often cause severe dermal side effects. Systemically applied doxorubicin is known for inducing free radicals, which leads to the development of the hand-foot syndrome. This syndrome manifests itself through skin irritations, extending from blistering to open wounds. As doxorubicin exhibits a fluorescence signal in the 520-600 nm region if excited at 488 nm, the doxorubicin’s leakage onto the skin surface could be analyzed. It was found that part of the doxorubicin is ejected with the sweat onto the skin surface, where it spreads and penetrates into the skin like topically applied. By topical application of antioxidants, the doxorubicin could be prevented from inducing free radicals in the skin and consequently the hand-foot syndrome. Raman spectroscopy was used to show that the action mechanism of chemotherapeutics not showing fluorescence signals is similar to the action mechanism of doxorubicin.
Christoph Pohling, Thomas Bocklitz, Alex Duarte, Cinzia Emmanuello, Mariana Ishikawa, Benjamin Dietzeck, Tiago Buckup, Ortrud Uckermann, Gabriele Schackert, Matthias Kirsch, Michael Schmitt, Jürgen Popp, Marcus Motzkus
Multiplex coherent anti-Stokes Raman scattering (MCARS) microscopy was carried out to map a solid tumor in mouse brain tissue. The border between normal and tumor tissue was visualized using support vector machines (SVM) as a higher ranking type of data classification. Training data were collected separately in both tissue types, and the image contrast is based on class affiliation of the single spectra. Color coding in the image generated by SVM is then related to pathological information instead of single spectral intensities or spectral differences within the data set. The results show good agreement with the H&E stained reference and spontaneous Raman microscopy, proving the validity of the MCARS approach in combination with SVM.
Molecular imaging modalities, like multimodal imaging, e.g. the combination of coherent-anti-Stokes Raman scattering (CARS), second-harmonic generation (SHG) and two-photon-excited fluorescence (TPEF), feature a unique potential for disease diagnostics and treatment monitoring. In order to use the full potential of multi-modal imaging for diagnostic tasks powerful image analysis methods are necessary, which translate the physical measurements into medical meaningful information. This translation process requires standardization, correction procedures and statistical models like regression or classification models. With the help of these techniques medical relevant information can be extracted and used for the diagnostics of cancer or inflammation related diseases.
The determination of the concentration of xenobiotics in biological matrix followed by the change of the prescribing procedure plays a major role in the transition from general to personalized medicine. For this contribution, human urine samples collected from healthy volunteers and from patients having urinary tract infection were used as biological matrix to assess the potential and limitation of LoC-SERS to detected levofloxacin and nitroxoline. The determination of both antibiotics at clinically relevant concentrations, 1.38 mM ± 0.68 mM for levofloxacin and 10-40 µM for nitroxoline, will be presented. For quantification purposes the standard addition method is combined with LoC-SERS.
The realization of label-free molecule specific imaging of morphology and chemical composition of tissue at subcellular spatial resolution in real time is crucial for many envisioned applications in medicine, e.g., precise surgical guidance and non-invasive histopathologic examination of tissue. Thus, new approaches for a fast and reliable in vivo and near in vivo (ex corpore in vivo) tissue characterization to supplement routine pathological diagnostics is needed. Spectroscopic imaging approaches are particularly important since they have the potential to provide a pathologist with adequate support in the form of clinically-relevant information under both ex vivo and in vivo conditions. In this contribution it is demonstrated, that multimodal nonlinear microscopy combining coherent anti-Stokes Raman scattering (CARS), two photon excited fluorescence (TPEF) and second harmonic generation (SHG) enables the detection of characteristic structures and the accompanying molecular changes of widespread diseases, particularly of cancer and atherosclerosis. The detailed images enable an objective evaluation of the tissue samples for an early diagnosis of the disease status. Increasing the spectral resolution and analyzing CARS images at multiple Raman resonances improves the chemical specificity. To facilitate handling and interpretation of the image data characteristic properties can be automatically extracted by advanced image processing algorithms, e.g., for tissue classification. Overall, the presented examples show the great potential of multimodal imaging to augment standard intraoperative clinical assessment with functional multimodal CARS/SHG/TPEF images to highlight functional activity and tumor boundaries. It ensures fast, label-free and non-invasive intraoperative tissue classification paving the way towards in vivo optical pathology.
Raman spectroscopy has been proven to have tremendous potential as biomedical analytical tool for spectroscopic disease diagnostics. The use of fiberoptic coupled Raman spectroscopy systems can enable in-vivo characterization of suspicious lesions. However, Raman spectroscopy has the drawback of rather long acquisition times of several hundreds of milliseconds which makes scanning of larger regions quite challenging. By combining Raman spectroscopy with a fast imaging technique this problem can be alleviate in part. Fluorescence lifetime imaging (FLIm) offers a great potential for such a combination. FLIm can allow for fast tissue area pre-segmentation and location of the points for Raman spectra acquisition. Here, we introduce an optical fiber probe combining FLIm and Raman spectroscopy with an outer diameter of 2 mm. Fluorescence is generated via excitation with a fiber laser at 355 nm. The fluorescence emission is spectrally resolved using a custom-made wavelength-selection module (WSM). The Raman excitation power at 785 nm was set to 50 mW for the in-vivo measurements to prevent sample drying. The lateral probe resolution was determined to be <250 μm for both modalities. This value was taken as step size for several raster scans of different tissue types which were conducted to show the overlap of both modalities under realistic conditions. Finally the probe was used for in vivo raster scans of a rat’s brain and subsequently to acquire FLIm guided Raman spectra of several tissues in and around the craniotomy.
Medical diagnosis of biopsies performed by fine needle aspiration has to be very reliable. Therefore, pathologists/cytologists need additional biochemical information on single cancer cells for an accurate diagnosis. Accordingly, we applied three different classification models for discriminating various features of six breast cancer cell lines by analyzing Raman microspectroscopic data. The statistical evaluations are implemented by linear discriminant analysis (LDA) and support vector machines (SVM). For the first model, a total of 61,580 Raman spectra from 110 single cells are discriminated at the cell-line level with an accuracy of 99.52% using an SVM. The LDA classification based on Raman data achieved an accuracy of 94.04% by discriminating cell lines by their origin (solid tumor versus pleural effusion). In the third model, Raman cell spectra are classified by their cancer subtypes. LDA results show an accuracy of 97.45% and specificities of 97.78%, 99.11%, and 98.97% for the subtypes basal-like, HER2+/ER− , and luminal, respectively. These subtypes are confirmed by gene expression patterns, which are important prognostic features in diagnosis. This work shows the applicability of Raman spectroscopy and statistical data handling in analyzing cancer-relevant biochemical information for advanced medical diagnosis on the single-cell level.
We report on a Raman microspectroscopic characterization of the inflammatory bowel diseases (IBD) Crohn's disease (CD) and ulcerative colitis (UC). Therefore, Raman maps of human colon tissue sections were analyzed by utilizing innovative chemometric approaches. First, support vector machines were applied to highlight the tissue morphology ( = Raman spectroscopic histopathology). In a second step, the biochemical tissue composition has been studied by analyzing the epithelium Raman spectra of sections of healthy control subjects (n = 11), subjects with CD (n = 14), and subjects with UC (n = 13). These three groups exhibit significantly different molecular specific Raman signatures, allowing establishment of a classifier (support-vector-machine). By utilizing this classifier it was possible to separate between healthy control patients, patients with CD, and patients with UC with an accuracy of 98.90%. The automatic design of both classification steps (visualization of the tissue morphology and molecular classification of IBD) paves the way for an objective clinical diagnosis of IBD by means of Raman spectroscopy in combination with chemometric approaches.
Circulating epithelial tumor cells are of increasing importance for tumor diagnosis and therapy monitoring of cancer
patients. The definite identification of the rare tumor cells within numerous blood cells is challenging. Therefore, within
the research initiative "Jenaer Zell-Identifizierungs-Gruppe" (JenZIG) we develop new methods for cell identification,
micromanipulation and sorting based on spectroscopic methods and microfluidic systems. In this contribution we show,
that classification models based on Raman spectroscopic analysis allow a precise discrimination of tumor cells from
non-tumor cells with high prediction accuracies, up to more than 99% for dried cells. That holds true for unknown cell
mixtures (tumor cells and leukocytes/erythrocytes) under dried conditions as well as in solution using the Raman laser
as an optical tweezers to keep the cells in focus. We extended our studies by using a capillary system consisting of a
quartz capillary, fiber optics and an adjustable fitting to trap cells. This system allows a prediction accuracy of 92.2%
on the single cell level, and is a prerequisite for the development of a cell sorting and identification device based on a
microfluidic chip. Initial experiments show that tumor cell lines can be differentiated from healthy leukocyte cells with
an accuracy of more than 98%.
This contribution will present a variety of applications of lab-on-a-chip surface enhanced Raman spectroscopy in the
field of bioanalytic. Beside the quantification and online monitoring of drugs and pharmaceuticals, determination of
enzyme activity and discrimination of bacteria are successfully carried out utilizing LOC-SERS. The online-monitoring
of drugs using SERS in a microfluidic device is demonstrated for nicotine. The enzyme activity of thiopurine
methyltransferase (TPMT) in lysed red blood cells is determined by SERS in a lab-on-a-chip device. To analyse the
activity of TPMT the metabolism of 6-mercaptopurine to 6-methylmercaptopurine is investigated. The discrimination of
bacteria on strain level is carried out with different E. coli strains. For the investigations, the bacteria are busted by ultra
sonic to achieve a high information output. This sample preparation provides the possibility to detect SERS spectra
containing information of the bacterial cell walls as well as of the cytoplasm. This contribution demonstrates the great
potential of LOC-SERS in the field of bioanalytics.
Here we present our latest results concerning the application of Raman microspectroscopy in combination with
innovative chemometrics to characterize biological cells. The first part of this manuscript deals with the application of
micro-Raman spectroscopy to identify microbial contaminations while the main focus within the second part of this
presentation is concerned with Raman studies on eukaryotic cells where we will report about the development of an
algorithm to differentiate between breast cancer cells and normal epithelial cells.
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.