Photonic Data Science

Research group Prof. Dr. Thomas Bocklitz

Fig 3

Graphic: IPHT
Fig 3

Head of the group

Thomas Bocklitz, University Professor Dr

Head
Professorship of Photonic Data Science
Prof. Dr. Thomas Bocklitz
Image: Prof. Dr. Thomas Bocklitz
Link to download vCard
vCard
JenTower
Leutragraben 1
07743 Jena Google Maps site planExternal link

Scientific Profile

We explore the entire data life cycle of photonic data from generation to the data analysis and to data archiving. Following a holistic approach, we investigate procedures for experiment and sample size planning as well as data pretreatment and combine these procedures with chemometric procedures, model transfer methods and artificial intelligence methods in a data pipeline. In this way, data from various photonic processes can be used for analysis, diagnostics and therapy in medicine, life science, environmental sciences and pharmacy. The data pipeline are implemented in software components and are tested directly in the applicative environment, e.g. in clinical studies. Further focal points are data fusion of different heterogeneous data sources, the simulation of different measurement procedures in order to optimize correction procedures, methods for the interpretation of analysis models and the construction of data infrastructures for different photonic measurement data, which ensure the FAIR principles.

Research Topics

  • Machine learning for photonic image data
    Fig 1

    Fig 1

    Graphic: IPHT
  • Chemometrics / machine learning for spectral data
  • Correlation of different measurement methods and data fusion

Areas of application

  • Bio-medical diagnostics using spectral measurement methods and imaging techniques
  • Extraction of higher information from photonic measurement data
  • Simulation- und data-driven correction of photonic data
  • Guarantee of FAIR principles for photonic data

Staff

  14. Vulchi, Ravi Teja

    PhD Student Professorship of Photonic Data Science
    Link to download vCard
    vCard

    JenTower
    Leutragraben 1
    07743 Jena

  15. Yogita, Yogita

    PhD student Professorship of Photonic Data Science
    Link to download vCard
    vCard

    JenTower, Room 15S01
    Leutragraben 1
    07743 Jena

Filter 115 publications

Filter publications

Highlighted authors are members of the research group.

  1. Automatic Coregistration of High-Resolution MALDI-MSI and Raman Imaging Applied to Cardiac Tissue of Fabry Disease Mouse Models

    Authors
    J. Dierks, E. Brockmann, A. Arias-Loza, T. Bocklitz, P. Nordbeck, K. Lorenz, E. Tolstik, S. Heiles
    Year of publication
    Published in:
    Analytical chemistry
    Understanding early molecular changes in biological tissues is crucial for diagnosing pathological and genetic diseases and for elucidating their underlying mechanisms. However, localized molecular alterations of low-molecular-weight compounds (<2000 Da) are not inferred from conventional staining or genetic methods. Here, we established a multimodal imaging approach that integrates Raman spectroscopy and atmospheric pressure matrix-assisted laser desorption/ionization mass spectrometry imaging (AP-MALDI-MSI): two complementary, label-free techniques enabling molecular profiling of a broad spectrum of biomolecules from one single tissue section. This method was applied to detect globotriaosylceramide (Gb3) accumulation in the heart tissue of murine models of Fabry disease, including mice deficient in α-galactosidase A (GLA) activity (GLA knockout) and transgenic mice with a GLA knockout and an upregulation of Gb3 synthase. With AP-MALDI-MSI, we were able to discern the heterogeneous expression of Gb3 lipoforms with down to 5 μm pixel size and reveal the significantly increased Gb3 content in mice containing a GLA knockout combined with human Gb3 synthase overexpression compared to GLA knockout and wild-type samples. By employing Raman microscopy with a pixel size of 2 μm, we were able to contextualize the physiological alterations in cardiac tissue by identifying components associated with nuclei, tissue, collagen, and lipids for the same three genotypes. An automated coregistration algorithm aligned Raman and AP-MALDI-MSI data from the same tissue section with a (5.1 ± 1.6) μm precision, enabling overlay at 5 and 2 μm resolutions. The method resolved heterogeneous Gb3 distributions and distinct lipid species in cardiac mouse tissue.
    University Bibliography Jena:
    fsu_mods_00038296External link

  2. Siamese networks in Raman spectroscopy: Towards a better performance against replicate variability

    Authors
    S. Guo, T. Bocklitz
    Year of publication
    Published in:
    Talanta: the international journal of pure and applied analytical chemistry
    The power of Raman spectroscopy is largely enhanced by machine learning and chemometrics, which extract and translate the spectral features into high-level biological or clinical knowledge by constructing classical or deep learning models. The generalizability of such models, however, is often degraded due to the large variations between the training data and the data to be predicted. Model transfer showed great potential in this regard, which improved the prediction on the test data without re-building a new model from scratch. We developed a method based on Siamese neural network (SNet) and compared it with two basis models as well as two model transfer methods score movement (MS) and extensive multiplicative scattering correction (EMSC). The performance was systematically verified with a Raman spectral dataset measured from four bacterial species, each consisting of nine biological replicates. Its generalizability was further tested on a second Raman dataset from mice tissue samples. Siamese network was demonstrated to outperform the MS and EMSC, especially given large training datasets. The load on training data, however, is substantially lower than conventional networks and can be slightly reduced when variability between training and test data is properly incorporated into the loss function. Unlike MS and EMSC, more importantly, Siamese network does not require information of test data for model adjustment or data space adaptation, which makes it more advantageous in practice.
    University Bibliography Jena:
    fsu_mods_00030099External link

  3. Harnessing Machine Learning and Deep Learning Approaches for Laser-Induced Breakdown Spectroscopy Data Analysis: A Comprehensive Review

    Authors
    P. Dehbozorgi, L. Duponchel, V. Motto-Ros, T. Bocklitz
    Year of publication
    Published in:
    Analysis & Sensing
    Laser-induced breakdown spectroscopy (LIBS) is a rapid, accurate technique for material analysis, offering real-time, minimally destructive, and in situ detection capabilities with broad application potential. LIBS extends its applications across various fields, from geology to biomedicine. However, barriers like matrix effects, reproducibility, self-absorption, and spectral noise often restrict the proper interpretation of the spectra. This review paper examines literature from 2015 to 2025, focusing on the evolution of machine learning (ML) and deep learning (DL) techniques, in LIBS analysis. It evaluates the advancement of these techniques, assessing both the qualitative and quantitative performance of LIBS analysis. These observations support the complementary roles of ML and DL methodologies. ML captures general patterns, while DL, through convolutional neural networks (CNNs), excels at identifying high-level features. This literature review reveals that no single ML or DL tool consistently provides optimal solutions for LIBS applications. The analysis pipeline needs to be tailored based on the LIBS data and the goal of the study. Designing such a framework requires the incorporation of preprocessing techniques to enhance the quality of raw signals. This step should then be followed by integrating the data into predictive models, whether ML or DL, to accomplish tasks like classification or concentration prediction.
    University Bibliography Jena:
    fsu_mods_00027407External link

  4. Toward cancer theranostics via multimodal nonlinear endomicroscopy, microspectroscopy and femtosecond laser ablation

    Author
    M. Calvarese
    Year of publication
    This thesis advances nonlinear multimodal imaging techniques, including Coherent Anti-Stokes Raman Scattering (CARS), Two-Photon Excited Fluorescence (TPEF), Second Harmonic Generation (SHG), and broadband Coherent Raman Scattering (BCARS), toward clinical applications to cancer diagnosis and treatment. A multimodal endomicroscopic platform integrating CARS, TPEF, SHG, indocyanine green (ICG) fluorescence, and femtosecond laser ablation was developed and validated for real-time cancer detection and treatment. The system enables a “seek-and-treat” approach combining imaging, AI-based tissue analysis, and selective laser ablation for image-guided surgery. Preclinical studies on head and neck cancer tissue demonstrated high-quality imaging and robust performance comparable to commercial systems. AI-based analysis achieved 88% sensitivity in distinguishing tissue requiring resection from healthy tissue, while automated image-guided laser ablation demonstrated precise selective tissue removal. The thesis also investigates broadband CARS for enhanced chemical specificity and spectral histopathology. BCARS systems were developed and applied to biological samples, showing strong potential for molecular characterization, tissue classification, and disease detection. Preliminary results further demonstrate the feasibility of integrating BCARS into endomicroscopy, supporting future real-time intraoperative spectral imaging. Overall, this work highlights the potential of advanced nonlinear imaging for image-guided cancer diagnosis and treatment.
    University Bibliography Jena:
    fsu_mods_00036802External link

  5. Robust Spectroscopic Analysis Through Image-Based Spectral Representation and Deep Learning Techniques

    Authors
    A. Mokari, O. Ryabchykov, T. Bocklitz
    Year of publication
    Published in:
    Journal of Raman spectroscopy
    Variability in instrument calibration, both between different devices and over time, remains a significant obstacle to Raman spectroscopy. Even minor shifts in wavenumber can substantially reduce classification accuracy. Traditional spectral models struggle to remain robust under such conditions, particularly due to the one-dimensional (1D) nature of spectral inputs and the absence of pretrained deep learning architectures for Raman spectra. In this study, we systematically investigate the robustness of a previously introduced spider plot–based spectral representation specifically with respect to wavenumber calibration drift in test data. In this approach, each Raman spectrum is converted into a radial spider plot in which the angle of each sector represents the wavenumber, and the color indicates the normalized intensity, enabling the use of powerful image-based deep learning models. We employed EfficientNetB7, a pretrained convolutional neural network (CNN), to classify transformed spectra. Using a dataset of 5420 spectra from six bacterial species, we introduced synthetic wavenumber shifts of ±9 cm −¹ to simulate realistic calibration drift in the test data. The traditional method, which combined principal component analysis with linear discriminant analysis (PCA-LDA), showed a significant drop in balanced accuracy from 0.90 to 0.77 when shifts were present. By contrast, our spider plot–based approach retained strong performance, achieving balanced accuracy scores of 0.88 without shifts and 0.83 with shifts. Although no spectral augmentation was done in the training of either classification workflow, we attribute this robustness to the rotational consistency of spider plots under spectral shift and the tolerance to image rotation in CNNs, which are pretrained on large, augmented image datasets. Our findings demonstrate the advantage of using image-based methods to mitigate the effects of calibration variability in Raman spectroscopy.
    University Bibliography Jena:
    fsu_mods_00037697External link

  6. Blood cancer differentiation based on IR spectroscopy and chemometrics

    Authors
    L. Xie, S. Guo, T. Liu, X. Tang, R. Ji, X. Shen, Y. Xu, L. Chen, S. Wang, T. Bocklitz
    Year of publication
    Published in:
    Computer methods and programs in biomedicine
    Background and Objective White blood cells (WBCs) and their subpopulations play critical roles in detecting blood cancers due to their distinct biological and biochemical characteristics. Infrared (IR) spectroscopy offers a rapid, label-free, and non-destructive approach to probe molecular composition, making it a promising tool for biomedical diagnostics. The objective of this proof-of-principle study is to investigate the possibility of IR spectroscopy combined with chemometrics to differentiate leukemia from lymphoma, and to assess the capability of whole WBCs and their subpopulations in distinguishing the two diseases. Methods We based our study on 21 pediatric patients including 11 leukemia and 10 lymphoma cases, with in total 86,016 IR spectra measured from whole WBCs and the subpopulations. Data pipeline was established, including steps of spectral preprocessing, classification, and data fusion. Particularly, data fusion was implemented via low-, middle-, and high-level strategies, with the aim of combining spectra from different cell types and investigating their capability of differentiating the two blood cancers. Results The classification, both with and without data fusion, was benchmarked via the patient-wise cross-validation. A balanced accuracy of 80.0% was achieved based on IR spectra of whole WBCs. Further improvement was observed when combining whole WBCs and its subpopulations, with the best performance of 90.0% from combining whole WBCs and granulocytes with high-level data fusion strategy. The performance was observed consistent for both linear and nonlinear classifications based on linear discriminant analysis (LDA) and support vector machine (SVM), respectively. Conclusions The results indicate the promising potential of IR spectroscopy of blood samples to distinguish leukemia and lymphoma with the help of chemometric approaches. Further, WBC subpopulations, particularly granulocytes, were proven to contain complementary information to whole WBCs for differentiating leukemia from lymphoma. This provides critical insights for biomedical practice in blood cancer diagnostics.
    University Bibliography Jena:
    fsu_mods_00036197External link

  7. An optical microscopy simulator for automated generation of synthetic samples and image degradations

    Authors
    E. Corbetta, Y. Yogita, T. Bocklitz
    Year of publication
    Published in:
    Discover Applied Sciences
    Computational methods and data science increasingly support optical microscopy experiments for biological investigations and medical applications. Increasingly complex methods and critical decision-making connected to biomedical studies require large and varied datasets for training, validation, and benchmarking. However, due to ethical constraints, sharing policies, and data variability, new studies struggle to access exhaustively large image collections. To address the limited dataset availability, we present the Optical Microscopy Simulator, an open-source repository for generating optical microscopy measurements by simulating samples and experimental degradations. To be adaptable to many experimental scenarios, samples are simulated as tuneable geometric shapes, such as round blobs, filaments, and connected structures, while experimental artifacts include simulations of blurring, different noise sources, uneven illumination, and other signal degradation effects due to sample bleaching or camera acquisition process. The repository enables users to create images from scratch or build on existing high-quality measurements. The software tool enables user-friendly, tuneable generation of samples and experimental artifacts, thus supporting the development of computational methods with simulations of full experimental processes, which can be generated by combining simple building blocks to create combinations of image degradation processes. The open-source Optical Microscopy Simulator includes functions, notebooks, and templates for immediate testing and straightforward integration in existing image data pipelines.
    University Bibliography Jena:
    fsu_mods_00038734External link

  8. Spectral artifacts of oblique incidence on diffraction gratings in Raman spectroscopy

    Authors
    V. Morgunov, R. Vulchi, T. Bocklitz
    Year of publication
    Published in:
    Optics Continuum
    We analyze the spectral artifacts arising from oblique light incidence on a diffraction grating. Our analytical model reveals three distinct effects: (1) a complex, non-linear shift in peak positions; (2) nearly constant peak amplitudes; and (3) a subtle, asymmetric change in the peak shape. The dominant artifact is the spectral shift. We show that even after correcting for the main offset, a significant residual error of up to 2° remains for a common 600 lines/mm grating. In contrast, the spectral resolution is largely unaffected. The full width at half maximum changes by less than 0.3% for misalignments up to 5°, as a change in the peak’s angular profile is almost perfectly compensated by an opposing change in angular dispersion. This implies that grating misalignment is predominantly a complex spectral calibration problem (peak shift), not a resolution problem (peak broadening).
    University Bibliography Jena:
    fsu_mods_00038740External link

  9. Systematic investigation of pre-processing and feature extraction techniques in medical image analysis

    Authors
    P. Dehbozorgi, O. Ryabchykov, T. Bocklitz
    Year of publication
    Published in:
    Discover Artificial Intelligence
    As medical imagery remains the cornerstone of diagnosis, the success of complex classification tasks depends on high-quality data and diagnostic features. This study evaluates image pre-processing and feature extraction to optimize model performance. Our investigation assesses how variations in pre-processing and feature extraction impact classification efficiency across three imaging modalities: radiology (chest X-ray images), pathology (H&E (Hematoxylin and Eosin)-stained patches), and ophthalmology (OCT (Optical Coherence Tomography) scans). The experimental framework incorporates nine pre-processing techniques, adjustment (brightness, contrast, and histogram equalization), filtering (mean, median, and Gaussian), and three normalization schemes, systematically combined for each modality. Features were extracted using five pre-trained architectures (VGG16, ResNet50, DenseNet121, MobileNetV2, and InceptionV3) and classified via a PCA-LDA (Principal Component Analysis-Linear Discriminant Analysis) pipeline. Performance was evaluated using the mean sensitivity score. Mean sensitivity scores improved significantly: H&E-stained images increased from 74·9% to 96·95%, chest X-rays from 89·9% to 96·65%, and OCT scans from 82·4% to 98·8%. No single pre-processing configuration consistently dominated; optimal settings varied by modality, confirming the absence of a universal solution. VGG16 and DenseNet121 exhibited the greatest robustness. Implementing at least one pre-processing step combined with a robust DL feature extractor can substantially improve model efficacy in diagnostic detection tasks.
    University Bibliography Jena:
    fsu_mods_00038286External link

  11. Denoising and Baseline Correction of Low-Scan FTIR Spectra: a Benchmark of Deep Learning Models Against Traditional Signal Processing

    Authors
    A. Mokari, S. Raghunathan, A. Shydliukh, O. Ryabchykov, C. Krafft, T. Bocklitz
    Year of publication
    Published in:
    Bioengineering
    High-quality Fourier Transform Infrared (FTIR) imaging usually needs extensive signal averaging to reduce noise and drift, which severely limits clinical speed. Deep learning can accelerate imaging by reconstructing spectra from rapid, single-scan inputs. However, separating noise and baseline drift simultaneously without ground truth is an ill-posed inverse problem. Standard black-box architectures often rely on statistical approximations that introduce spectral hallucinations or fail to generalize to unstable atmospheric conditions. To solve these issues, we propose a physics-informed cascade Unet that separates denoising and baseline correction tasks using a new, deterministic Physics Bridge. This architecture forces the network to separate random noise from chemical signals using an embedded SNIP layer to enforce spectroscopic constraints instead of learning statistical approximations. We benchmarked this approach against a standard single Unet and a traditional Savitzky–Golay smoothing followed by SNIP baseline correction workflow. We used a dataset of human hypopharyngeal carcinoma cells (FaDu). The cascade model outperformed all other methods, achieving a 51.3% reduction in RMSE compared to raw single-scan inputs, surpassing both the single Unet (40.2%) and the traditional workflow (33.7%). Peak-aware metrics show that the cascade architecture eliminates spectral hallucinations found in standard deep learning. It also preserves peak intensity with much higher fidelity than traditional smoothing. These results show that the cascade Unet is a robust solution for diagnostic-grade FTIR imaging. It enables imaging speeds 32 times faster than current methods.
    University Bibliography Jena:
    fsu_mods_00035379External link

  12. Systematic investigation of preprocessing pipeline for MALDI data

    Authors
    M. Adhikari, O. Ryabchykov, S. Guo, T. Bocklitz
    Year of publication
    Published in:
    Results in Chemistry
    Background Matrix-assisted laser desorption/ ionization Mass Spectrometry (MALDI-MS) is a powerful tool to detect and characterize biomolecules, making it particularly useful in different fields and applications such as proteomics, clinical diagnostics, and biomarker discovery. MALDI data is commonly contaminated by the artefacts originated from both chemical and electrical noise. Data preprocessing is hence important to remove these artefacts and improve the accuracy and reliability of the subsequent (quantitative and qualitative) analysis. A systematic investigation of different preprocessing steps is necessary to establish an effective preprocessing pipeline. Results In this study, we systematically investigated the different steps including interpolation, smoothing, baseline correction, peak alignment, and peak binning, along with normalization to establish a preprocessing pipeline of MALDI spectral data. The performance of the preprocessing steps and pipeline was benchmarked by the balanced accuracy of differentiating hepatocellular carcinoma (HCC) and healthy (normal) based on MALDI spectral data of liver tissue samples. The established preprocessing pipeline improved the balanced accuracy from 61.3% to 77.6% under the patient-level cross-validation, and from 92.9% to 94.7% under spectral-level cross-validation. Significance Our findings demonstrated that the classification performance can be greatly affected by the quality of MALDI data, which can be improved by preprocessing steps. The large improvement from the patient-level validations after preprocessing demonstrated well a satisfying performance of the classification against patient-to-patient variability with the help of our preprocessing pipeline. This study will potentially benefit the MS community.
    University Bibliography Jena:
    fsu_mods_00034772External link
Pagination Page 1