The DOI matches its Crossref record; 77 of the 77 references it cites were found in a registry or an index; 7 of the 22 sources its claims cite were checked for support.
Signal Estimation and Uncertainties Extraction in Terahertz Time-Domain Spectroscopy
Open accessAbstract
Terahertz time-domain spectroscopy (THz-TDS) systems have emerged as mature technologies with significant potential across various research fields and industries. However, the lack of standardized signal and noise estimation and reduction methods hinders its full potential. This article introduces a methodology to significantly reduce noise in THz-TDS time traces, providing a reliable and less-biased signal estimation. The method results in an improved signal-to-noise ratio (SNR), enabling the utilization of the full dynamic range of such setups. Additionally, we investigate the estimation of the covariance matrix to quantify the uncertainties associated with the signal estimator. This matrix is essential for extracting accurate material parameters by normalizing the error function in the fitting process. Our approach addresses practical scenarios where the number of repeated measurements is limited compared to the number of sampling points on the time axis. We envision this work as the initial step toward standardizing THz-TDS data processing. We believe that it will foster collaboration between the THz and signal processing communities, leading to the development of more sophisticated methods to tackle challenges posed by novel setups, such as those dedicated to dual-comb spectroscopy, which demands significant optimization of signal-to-noise performance.
Overview
CoChat Summary
TL;DR
The paper proposes signal-estimation and noise-covariance methods for THz time-domain spectroscopy, correcting delay-line drift, laser amplitude fluctuations, and periodic sampling errors to reduce bias and improve SNR. It also evaluates covariance and precision matrix estimators for uncertainty propagation in material-parameter fitting, implementing the approach in the open-source Correct@TDS software.
Key findings
- Correcting delay-line drift increased SNR from 28 dB to 55 dB for 1,000 time traces, an approximately 27 dB gain; adding laser amplitude correction further increased SNR to 62 dB, an approximately 7 dB gain.
- The raw average is biased by delay drift: attenuation exceeded 20% at 3 THz and worsened for 50,000 traces, while the corrected average matched the predicted attenuation model up to 4 THz.
- Periodic sampling errors produced artifacts beginning around 7.5 THz; the proposed correction reduced them using a Tukey window.
- For covariance estimation with n ≤ p, Graphical Lasso performed better than Ledoit-Wolf and Oracle Approximating Shrinkage, giving good estimation with as few as 100 time traces; corrections reduced covariance values by three orders of magnitude, and precision-matrix partial correlations increased by one order of magnitude after correction. The conclusion reports over 30 dB SNR enhancement at 1 THz for 1,000 accumulated traces.
Methods
- Experiments used a MenloSystems TeraSmart THz-TDS setup with a 100 MHz repetition-rate laser, an 850 ps delay line, 33 fs time step, and typical bandwidth around 4 THz. Tests included dark noise, THz-dark perturbation, and 100,000 dry-nitrogen time traces over 100 ps.
- The signal estimator aligns traces to a reference via Fourier shift, corrects delay-line speed variation using a first-order Taylor expansion, corrects laser amplitude fluctuations, and reduces periodic sampling errors; these are implemented in Correct@TDS.
- Covariance and precision matrices were estimated from corrected repeated measurements, comparing empirical covariance, Ledoit-Wolf shrinkage, Oracle Approximating Shrinkage, and Graphical Lasso from scikit-learn. Target matrices were computed from 100,000 traces with p = 3000 sampling points, and performance was assessed by normalized Frobenius distances and matrix norms.
Limitations
- The paper states there is no standard protocol for THz-TDS data acquisition/processing or uncertainty extraction, and that covariance inversion requires more repeated measurements than time-sampling points, which is often impractical: a 400 ps trace has about 12,500 points, and 12,501 traces take about 2 hours.
- The paper notes that lasso-penalized methods can be inconsistent on some graph structures, precision-matrix estimation has no optimal solution, and the covariance-estimation scope was limited to reference data without samples.
- Evident from the text: tests used one specific commercial setup and selected experimental conditions; delay-line speed correction was negligible for a single-peak reference but important for echoes outside the paper’s scope; target matrices were assumed true based on 100,000 traces.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 5 Not valid · 7 Not checked · 72 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“These methods have the advantage to be able to recover the off diagonal structure of the covariance matrix.”
What the sources say
- Disputedcites [57] sourceExperimental — accuracy not yet measured
The Abstract identifies the structural zeros targeted by covariance selection as entries of the inverse covariance matrix, not the covariance matrix. Section 3 says Lasso neighborhood selection recovers the graph’s edge set, which represents conditional-independence structure; it does not claim recovery of the covariance matrix’s off-diagonal structure. Page numbers are unavailable in our copy. (LLM, conf=0.87)
Checked by a reasoning model — may be wrong.
- Not checked — no full text of the sourcecites [58] source
- Not checked — no full text of the sourcecites [59] 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.
- Disputed
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Its average power decreases when accumulating, it goes from 1 × 10 -1 µ V for an average on 30 time traces to 1 . 1 × 10 -2 µ V for an average of 3000 time traces.”
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.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The two shrinkage methods also produce here similar results.”
plain- 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.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The ratio ranges from approximately 0 . 6 to 1 depending on the estimation method.”
plainstates an association- 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.
- Rests on an assumption not shown here.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our signal correction significantly improves the signal-to-noise ratio(over 30 dB enhancement at 1 THz for 1000 accumulated traces).”
plainstates a cause- 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.
28 more claims
Create a free account to see all 88 claims and to verify the 2 references still open.
- Probably not a substantial claim. 21
- It may state two separate claims as one. 37
- The quote only partly states this claim. 2
Citations
Cites
Scikit-learn: Machine Learning in Python
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Müller, A., Nothman, J., Louppe, G., et al. (2012) · arXiv · doi:10.48550/arxiv.1201.0490 · Cited by 71,016
Scikit-learn: Machine Learning in Python
(2011) · Journal of Machine Learning Research · doi:10.5555/1953048.2078195 · Cited by 8,539 · cited 1×
“To achieve this, we utilize methods implemented in the widely used Python library, Scikit-learn [54].”
Sparse inverse covariance estimation with the graphical lasso
Friedman, J., Hastie, T., & Tibshirani, R. (2008) · Biostatistics · doi:10.1093/biostatistics/kxm045 · Cited by 6,355
A well-conditioned estimator for large-dimensional covariance matrices
Ledoit, O., & Wolf, M. (2004) · Journal of Multivariate Analysis · doi:10.1016/s0047-259x(03)00096-4 · Cited by 3,046 · cited 1×
“To avoid this inversion problem, the shrinkage method has been proposed by Ledoit and Wolf [55] (Ledoit–Wolf shrinkage).”
High-dimensional graphs and variable selection with the Lasso
Meinshausen, N., & Bühlmann, P. (2006) · The Annals of Statistics · doi:10.1214/009053606000000281 · Cited by 2,521 · cited 1×
“There are also methods that rely on solving the covariance selection problem to approximate sparse inverse covariance matrices with lasso regularization [57], [58], [59].”
Model selection and estimation in the Gaussian graphical model
Yuan, M., & Lin, Y. (2007) · Biometrika · doi:10.1093/biomet/asm018 · Cited by 1,627 · cited 1×
“There are also methods that rely on solving the covariance selection problem to approximate sparse inverse covariance matrices with lasso regularization [57], [58], [59].”
A reliable method for extraction of material parameters in terahertz time-domain spectroscopy
Duvillaret, L., Garet, F., & Coutaz, J.-L. (1996) · IEEE Journal of Selected Topics in Quantum Electronics · doi:10.1109/2944.571775 · Cited by 994 · cited 2×
“Historically [14], [15], [16], [17], the sequence of information extraction included a fast Fourier transform (FFT), phase unwrap, refractive index evaluation thanks to a basic and often valid photonic model, and finally, the parameter estimation.”
Material parameter estimation with terahertz time-domain spectroscopy
Dorney, T. D., Baraniuk, R. G., & Mittleman, D. M. (2001) · Journal of the Optical Society of America A · doi:10.1364/josaa.18.001562 · Cited by 807 · cited 2×
“Historically [14], [15], [16], [17], the sequence of information extraction included a fast Fourier transform (FFT), phase unwrap, refractive index evaluation thanks to a basic and often valid photonic model, and finally, the parameter estimation.”
Shrinkage Algorithms for MMSE Covariance Estimation
Chen, Y., Wiesel, A., Eldar, Y. C., & Hero, A. O. (2010) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2010.2053029 · Cited by 567
Duvillaret, L., Garet, F., & Coutaz, J.-L. (1999) · Applied Optics · doi:10.1364/ao.38.000409 · Cited by 505 · cited 1×
“Historically [14], [15], [16], [17], the sequence of information extraction included a fast Fourier transform (FFT), phase unwrap, refractive index evaluation thanks to a basic and often valid photonic model, and finally, the parameter estimation.”
Research works printed but not linked: 2
- Denakpo, 2024, Software: Correct@tds
- Peretti, 2024, Full Acquisition of 100k & 50k Terahertz Time Domain Spectroscopy Reference Time Traces
Cited by
Laser-driven few-cycle terahertz sources with high average power
Löscher, R., Vogel, T., Mansourzadeh, S., Khalili, M., Omar, A., Wang, Y., Hoffmann, M., & Saraceno, C. J. (2025) · APL Photonics · doi:10.1063/5.0288917 · Cited by 9
Chen, M., Xue, M., liu, Z., Chen, T., Li, Y., Li, M., Huang, W., Liu, L., Chen, Y., Chen, Y., et al. (2026) · Journal of Food Composition and Analysis · doi:10.1016/j.jfca.2026.108872 · Cited by 3
Hernandez, M., Hämäläinen, M., Mura, S., et al. (2026) · doi:10.1145/3818305.3830268
Uncertainty Analysis of a Surface Scattering-Aware Material Extraction Algorithm for THz-TDs Systems
Sari, A., Farahi, Y., Constantinou, C., et al. (2026) · doi:10.23919/eucap68105.2026.11612344
“Consequently—aside from modeling assumptions and errors, external or system noise, and other instrumentation uncertainties [4], [11], [12], [13]—the algorithm is directly influenced by the user-defined values of d 0 and σ h .”
Research on a method for analysing the degree of wood modification based on THz-TDS
Xing, L., Huang, R., Cheng, X., et al. (2026) · Nondestructive Testing And Evaluation · doi:10.1080/10589759.2026.2634229
Metadata
- Authors
- Denakpo, E., Hannotte, T., Osseiran, N., Orieux, F., & Peretti, R.
- Affiliations
- Centre National de la Recherche Scientifique (FR)
- Venue
- IEEE Transactions on Instrumentation and Measurement
- Published
- 2025
- Volume
- 74
- Pages
- 1-13
- ISSN
- 0018-9456 (ISSN-L)
- 1557-9662
- OpenAlex
- W4403241476
- Semantic Scholar
- f5799be752b7ee22dd8486be25333517cd6d7da8
- Cited by
- 6
- References
- 67
- Citations by year
- 2026: 5 · 2025: 1
- FWCI
- 1.35
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Instrumentation and Measurement (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- HAL (submittedVersion, other-oa)
- Figshare (submittedVersion, public-domain)
- LillOA (Université de Lille (University Of Lille)) (submittedVersion, other-oa)
- arXiv (Cornell University)
- Funders
- R?gion ?Hauts-de-France?
- Diagnotera Start-AIRR project
- DEUS MAROONER STIMulE project
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Terahertz technology and applications
- Physical Sciences › Chemistry › Spectroscopy › Spectroscopy and Laser Applications
- Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Laser-Matter Interactions and Applications
- Keywords
- terahertz time-domain spectroscopy, noise reduction, signal estimation, signal-to-noise ratio, covariance matrix estimation, uncertainty quantification, parameter fitting, dual-comb spectroscopy
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: Signal Estimation and Uncertainties Extraction in Terahertz Time-Domain Spectroscopy. https://app.cochat.ai/papers/10.1109/tim.2025.3554287
- The paper
- Denakpo, E., Hannotte, T., Osseiran, N., Orieux, F., & Peretti, R. (2025). Signal Estimation and Uncertainties Extraction in Terahertz Time-Domain Spectroscopy. IEEE Transactions on Instrumentation and Measurement. https://doi.org/10.1109/tim.2025.3554287
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.