The DOI matches its Crossref record; 118 of the 118 references it cites were found in a registry or an index; 11 of the 21 sources its claims cite were checked for support.
Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders
Open accessAbstract
Raman spectroscopy is widely used across scientific domains to characterize the chemical composition of samples in a nondestructive, label-free manner. Many applications entail the unmixing of signals from mixtures of molecular species to identify the individual components present and their proportions, yet conventional methods for chemometrics often struggle with complex mixture scenarios encountered in practice. Here, we develop hyperspectral unmixing algorithms based on autoencoder neural networks, and we systematically validate them using both synthetic and experimental benchmark datasets created in-house. Our results demonstrate that unmixing autoencoders provide improved accuracy, robustness, and efficiency compared to standard unmixing methods. We also showcase the applicability of autoencoders to complex biological settings by showing improved biochemical characterization of volumetric Raman imaging data from a monocytic cell.
Overview
CoChat Summary
TL;DR
The paper introduces physics-constrained autoencoder neural networks for blind hyperspectral unmixing of Raman spectroscopy data. Across synthetic mixtures, experimental sugar mixtures, and volumetric Raman imaging of a THP-1 cell, the autoencoders generally outperform conventional N-FINDR/VCA + NNLS/FCLS methods, especially under noise, artifacts, and no pure-pixel conditions.
Key findings
- Autoencoders outperformed conventional methods on synthetic Raman mixtures across 11 dataset variants, with improvement increasing for noisier and more artifact-contaminated scenarios; they matched conventional methods on the simplest ideal Chessboard datasets. Experiments used 6 models × 11 dataset variants × 25 replicates = 1,650 experiments.
- On experimental sugar mixtures, autoencoders performed comparably to conventional methods on clean high-SNR data when pure-solution reference spectra were included, but substantially outperformed them in low-SNR settings and when pure pixels were absent. The sugar data comprised 240 mixtures and high-SNR (1,920 spectra, 5 s integration) and low-SNR (7,680 spectra, 0.5 s integration) measurements.
- Autoencoders were computationally faster than N-FINDR+FCLS and VCA+FCLS on synthetic datasets up to 250,000 spectra, even using CPU-only computation and including training time.
- In a low-SNR volumetric Raman scan of a THP-1 cell, autoencoder unmixing produced more distinct endmember signatures and well-defined abundance features than VCA+NNLS, and detected cholesterol, which the original VCA+NNLS approach did not.
Methods
- The framework treats the autoencoder latent representation as fractional abundances and the decoder as a mixing function; physics-inspired constraints include nonnegative endmembers, abundance nonnegativity, and sum-to-one abundances. Four encoders were tested (Dense, Convolutional, Transformer, Convolutional Transformer) with linear or bilinear Fan-model decoders, trained self-supervised using spectral angle distance loss.
- Synthetic data were generated with known ground truth: 5 endmembers, 1,000 spectral bands, and 100×100 abundance scenes (10,000 spectra per dataset), spanning Chessboard, Gaussian, and Dirichlet scenes plus linear/bilinear mixtures and artifacts.
- Experimental validation used 240 in-house sugar mixtures of glucose, sucrose, fructose, and maltose, plus a volumetric THP-1 cell Raman scan; performance was measured by MSE for abundances and SAD for endmembers, with Hungarian matching to ground truth when available.
Limitations
- The paper states limitations including potentially higher data requirements for stable performance, challenging interpretability of learned internal representations, and sensitivity to hyperparameters requiring careful tuning.
- The text also notes that synthetic data are inherently idealized and that findings from synthetic data alone should be extrapolated with caution.
- For the THP-1 cell application, the paper states that direct comparisons are challenging due to the lack of ground truth.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures





Verification
Claims the paper makes
8 Valid · 0 Not valid · 3 Not checked · 36 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Finally, we remark that our unsupervised autoencoder models do not utilize any ground truth information during training, in contrast with previous supervised machine learning approaches for Raman cell imaging and characterization where specific target concentrations are used as labels to train the model (e.g., using separately acquired fluorescence images) (90–93).”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Results > Application of Unmixing Autoencoders to Biological Data: Volumetric Raman Imaging of a THP-1 Cell.
What the sources say
- Partly supportedcites [90] B. Manifold, S. Men, R. Hu, D. Fu, A versatile deep learning architecture for classification and label-free prediction of hyperspectral images. Nat. Mach. Intell. 3, 306– 315 (2021). 34676358 PMC8528004 sourceExperimental — accuracy not yet measured
The Methods section (“Training Parameters, Data Preparation, and Hardware”; page not available) says SRS images were used as inputs and fluorescence images as the truth for training, supporting the contrast with training that uses ground-truth information. It does not say that specific target concentrations were used as labels for its Raman cell-imaging models; that detail is not supported by this source. (LLM, conf=0.87)
- Partly supportedcites [91] J. Zhang, J. Zhao, H. Lin, Y. Tan, J. X. Cheng, High-speed chemical imaging by dense-net learning of femtosecond stimulated Raman scattering. J. Phys. Chem. Lett. 11, 8573– 8578 (2020). 32914982 PMC9154735 sourceExperimental — accuracy not yet measured
The indexed text (section labeled Abstract; no page given) says DeepChem was trained on paired Raman images and segmented subcellular-organelle maps, supporting the contrast with an unsupervised model. It does not say that specific target concentrations were training labels; it says separately acquired fluorescence images confirmed the segmentation results, not that they supplied the labels. (LLM, conf=0.87)
- Partly supportedcites [92] Z. Wei, Pixel-level multimodal fusion deep networks for predicting subcellular organelle localization from label-free live-cell imaging. Front. Genet. 13, 1002327 (2022). 36386823 PMC9644055 sourceExperimental — accuracy not yet measured
Methods §2.1 describes training a Raman-imaging model with a cell-image dataset, and Results §3.4 identifies fluorescence images as ground truth for predicting organelle locations. This supports the contrast with supervised Raman imaging, but the paper does not say that specific target concentrations were used as training labels. The citing authors’ statement about their own unsupervised autoencoders is not attributed to this source. (LLM, conf=0.88)
- Not checked — no full text of the sourcecites [93] Y. Zhang, From genotype to phenotype: Raman spectroscopy and machine learning for label-free single-cell analysis. ACS Nano 18, 18101– 18117 (2024). 38950145 source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Yet, despite initial explorations (67, 68), the utility of unmixing AEs for Raman spectroscopy data remains largely unexplored.”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Abstract
What the sources say
- Not checked — no full text of the sourcecites [67] N. Burzynski et al., “Deep learning techniques for unmixing of hyperspectral stimulated raman scattering images” in 2021 IEEE International Conference on Big Data (Big Data) (IEEE, 2021), pp. 5862–5864. source
- Supportedcites [68] D. Boildieu, D. Helbert, A. Magnaudeix, P. Leproux, P. Carré, “Multivariate curve resolution with autoencoders for CARS microspectroscopy” in Computational Imaging Conference, IS&T Electronic Imaging (2023). source
The Introduction and Related Works says that, to the authors’ knowledge, autoencoders had not been tested for unmixing in microspectroscopy. The Abstract and Results describe this paper’s application to coherent anti-Stokes Raman scattering (CARS) data, making it an initial exploration of the approach cited in the sentence. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Despite the wealth of information RS affords, the analysis and interpretation of experimental RS data remains a major challenge (37–39).”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Abstract
What the sources say
- Not checked — no full text of the sourcecites [37] O. Ryabchykov, S. Guo, T. Bocklitz, Analyzing Raman spectroscopic data. Phys. Sci. Rev. 4, 20170043 (2018). source
- Not checked — no full text of the sourcecites [38] S. Guo, J. Popp, T. Bocklitz, Chemometric analysis in Raman spectroscopy from experimental design to machine learning-based modeling. Nat. Protoc. 16, 5426– 5459 (2021). 34741152 source
- Supportedcites [39] R. Gautam, S. Vanga, F. Ariese, S. Umapathy, Review of multidimensional data processing approaches for Raman and infrared spectroscopy. EPJ Tech. Instrum. 2, 1– 38 (2015).26146600 source
The Abstract says that interpreting Raman and IR results from heterogeneous biological samples is difficult. The Introduction explains that complex spectra and interfering substances make meaningful analysis challenging, and the Multivariate data analysis section gives a mixture example in which other components obscure the signal of interest. (LLM, conf=0.91)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Nonetheless, despite significant advances, most methods remain focused on constrained supervised tasks where a model is learned assuming knowledge of target species and their values (55).”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Abstract
What the sources say
- Partly supportedcites [55] L. Pan, P. Zhang, C. Daengngam, S. Peng, M. Chongcheawchamnan, A review of artificial intelligence methods combined with Raman spectroscopy to identify the composition of substances. J. Raman Spectrosc. 53, 6– 19 (2022). sourceExperimental — accuracy not yet measured
The Introduction (p. 2) calls supervised learning algorithms the most useful classification methods, and the SVM discussion (p. 6) describes training on marked data. The review does not establish that most existing methods are supervised or that they generally require known target concentrations or values. (LLM, conf=0.82)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“However, accounting for nonlinear mixing interactions increases the complexity and computational cost of unmixing (96, 97), an issue of especial relevance in Raman spectroscopy where datasets (e.g., imaging/volumetric scans) are typically larger than in remote sensing.”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Materials and Methods > Hyperspectral Unmixing.
What the sources say
- Supportedcites [96] N. Dobigeon, Nonlinear unmixing of hyperspectral images: Models and algorithms. IEEE Signal Process. Mag. 31, 82– 94 (2013). source
Section IV (p. 16) states that nonlinear models generally require higher computational complexity than linear-mixing approaches. The citation is attached to that claim; the subsequent comparison of Raman and remote-sensing dataset sizes is not attributed to this source. (LLM, conf=0.93)
- Not checked — no full text of the sourcecites [97] R. Heylen, M. Parente, P. Gader, A review of nonlinear hyperspectral unmixing methods. IEEE J. Sel. Top. Appl. Earth Obs. Remote. Sens. 7, 1844– 1868 (2014). source
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
18 more claims
Create a free account to see all 47 claims and every source verdict.
- Probably not a substantial claim. 13
- It may state two separate claims as one. 12
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Results > Benchmarking Unmixing Autoencoders on Synthetic Raman Mixtures.
“The computational cost (measured as wall time) of autoencoders and conventional methods on synthetic datasets (Chessboard scene, +artifacts scenario) as the number of spectra is increased.”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Materials and Methods > Measuring Computational Cost.
“We profile the computational cost of unmixing methods on synthetic datasets (ideal scenario, Chessboard scene) of increasing sizes, from 2,500 to 250,000 spectra.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Materials and Methods > Unmixing Autoencoders.
“we enforce relevant constraints, such as the nonnegativity of W and the nonnegativity (ANC) and sum-to-one constraint (ANC) of z”
§ Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders > Materials and Methods > Hyperspectral Unmixing.
“the abundance sum-to-one constraint (ASC), ||α||1=1”
Citations
Cites
(2014) · Cited by 33,419
Smoothing and Differentiation of Data by Simplified Least Squares Procedures.
Savitzky, A., & Golay, M. J. E. (1964) · Analytical Chemistry · doi:10.1021/ac60214a047 · Cited by 21,904
Solving Least Squares Problems
Lawson, C. L., & Hanson, R. J. (1995) · doi:10.1137/1.9781611971217 · Cited by 7,103
Kruse, F. A., Lefkoff, A. B., Boardman, J. W., Heidebrecht, K. B., Shapiro, A. T., Barloon, P. J., & Goetz, A. F. H. (1993) · Remote Sensing of Environment · doi:10.1016/0034-4257(93)90013-n · Cited by 3,164
Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches
Bioucas-Dias, J. M., Plaza, A., Dobigeon, N., Parente, M., Du, Q., Gader, P., & Chanussot, J. (2012) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2012.2194696 · Cited by 2,836 · cited 2×
“…cost of unmixing [82, 83], an issue of especial relevance in Raman spectroscopy where datasets (e.g., imaging/volumetric scans) are typically larger than in remote sensing; hence, despite its limitations, the LMM remains a cornerstone of hyperspectral unmixing in most practical settings [76, 41].”
BackgroundMethodOn the Opportunities and Risks of Foundation Models
Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., et al. (2021) · arXiv · doi:10.48550/arxiv.2108.07258 · Cited by 2,676
Vertex component analysis: a fast algorithm to unmix hyperspectral data
Nascimento, J. M. P., & Dias, J. M. B. (2005) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2005.844293 · Cited by 2,669 · cited 1×
“Popular approaches include N-FINDR [43] and Vertex Component Analysis (VCA) [44] for endmember identification, and Non-negative Least Squares (NNLS) [45] and Fully Constrained Least Squares (FCLS) [46] for abundance estimation [41, 47].”
MethodKeshava, N., & Mustard, J. F. (2002) · IEEE Signal Processing Magazine · doi:10.1109/79.974727 · Cited by 2,549
Raman Spectroscopy of Biological Tissues
Movasaghi, Z., Rehman, S., & Rehman, I. U. (2007) · Applied Spectroscopy Reviews · doi:10.1080/05704920701551530 · Cited by 2,518
Surface‐Enhanced Raman Spectroscopy: Concepts and Chemical Applications
Schlücker, S. (2014) · Angewandte Chemie International Edition · doi:10.1002/anie.201205748 · Cited by 2,417 · cited 1×
“As a result, RS has become an important analytical tool in a myriad of scientific domains, from chemistry [11, 12], biology [13, 14, 15, 16], and medicine [17, 18, 19, 20, 21, 22, 23], to materials science [24, 25], pharmacology [26, 27, 28], environmental science [29, 30, 31], food quality control…”
Background
Research works printed but not linked: 5
- M. E. Winter, N-FINDR, An algorithm for fast autonomous spectral end-member determination in hyperspectral data in Imaging Spectrometry V. SPIE 3753, 266–275 (1999).
- I. Goodfellow Y. Bengio A. Courville Autoencoders in Deep learning (MIT Press 2016) pp. 499–523.
- A. Vaswani . “Attention is all you need” in Advances in Neural Information Processing Systems 30 I. Guyon et al. Eds. (NIPS 2017).
- P. H. Eilers, H. F. Boelens, Baseline correction with asymmetric least squares smoothing. Leiden Univ. Med. Centre Rep. 1, 5 (2005).
- M. Abadi . “A system for large-scale machine learning” in OSDI (Savannah GA 2016) vol. 16 pp. 265–283.
Cited by
Lin, L. L., Alvarez-Puebla, R., Liz-Marzán, L. M., Trau, M., Wang, J., Fabris, L., Wang, X., Liu, G., Xu, S., Han, X. X., et al. (2025) · ACS Applied Materials & Interfaces · doi:10.1021/acsami.4c17502 · Cited by 247
RamanSPy: An Open-Source Python Package for Integrative Raman Spectroscopy Data Analysis
Georgiev, D., Pedersen, S. V., Xie, R., Fernández-Galiana, Á., Stevens, M. M., & Barahona, M. (2024) · Analytical Chemistry · doi:10.1021/acs.analchem.4c00383 · Cited by 108
Self-citationMonitoring kinetic processes of drugs and metabolites: Surface-enhanced Raman spectroscopy
Luo, Z., Chen, H., Bi, X., & Ye, J. (2024) · Advanced Drug Delivery Reviews · doi:10.1016/j.addr.2024.115483 · Cited by 37
Raman Spectroscopy for Nitrate Detection in Water: A Review of the Current State of Art
Luciani, L., Nocera, A., Raimondi, M., Ciattaglia, G., Spinsante, S., Gambi, E., & Galassi, R. (2025) · ACS Measurement Science Au · doi:10.1021/acsmeasuresciau.5c00016 · Cited by 21
Wu, S., Huang, L., & Huang, R. (2026) · Journal of Food Composition and Analysis · doi:10.1016/j.jfca.2026.108860 · Cited by 23
Metadata
- Authors
- Georgiev, D., Fernández-Galiana, Á., Vilms Pedersen, S., Papadopoulos, G., Xie, R., Stevens, M. M., & Barahona, M.
- Affiliations
- Imperial College London (GB)
- Department of Materials, Faculty of Engineering, Imperial College London, London SW7 2AZ, United Kingdom
- University of Oxford (GB)
- Corresponding author
- Fernández-Galiana, Álvaro
- Stevens, Molly M.
- Barahona, Mauricio
- Venue
- Proceedings of the National Academy of Sciences
- Publisher
- National Academy of Sciences
- Published
- 2024-10-29
- Volume
- 121
- Issue
- 45
- Pages
- e2407439121-e2407439121
- ISSN
- 0027-8424 (ISSN-L)
- 1091-6490
- PMC
- PMC11551349
- OpenAlex
- W4403863275
- Cited by
- 50
- References
- 111
- Citations by year
- 2026: 30 · 2025: 17 · 2024: 3
- FWCI
- 9.61
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Proceedings of the National Academy of Sciences (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- Spiral (Imperial College London) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Funders
- UK Research and Innovation (EP/S023283/1)
- Schmidt Science Fellowship (N/A)
- Rhodes Fellowship (N/A)
- Independent Research Fund Denmark (N/A)
- UKRI | Engineering and Physical Sciences Research Council (EP/P00114/1, EP/T020792/1, EP/N014529/1, EP/T027258/1)
- Royal Academy of Engineering (CiET2021\94)
- Bio Innovation Institute (N/A)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Spectroscopy Techniques in Biomedical and Chemical Research
- Physical Sciences › Chemistry › Analytical Chemistry › Spectroscopy and Chemometric Analyses
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
- Keywords
- hyperspectral unmixing, Raman spectroscopy, autoencoder, spectral unmixing, chemometrics, monocytic cells, biochemical characterization
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders. https://app.cochat.ai/papers/10.1073/pnas.2407439121
- The paper
- Georgiev, D., Fernández-Galiana, Á., Vilms Pedersen, S., Papadopoulos, G., Xie, R., Stevens, M. M., & Barahona, M. (2024). Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders. Proceedings of the National Academy of Sciences. https://doi.org/10.1073/pnas.2407439121
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.