pith. sign in

arxiv: 2512.01585 · v1 · submitted 2025-12-01 · ⚛️ physics.optics

Unsupervised and Supervised Algorithms for Identification of Sample Pixels in FTIR Images

Xiangyu Zhao , Yudong Tian , Jingzhu Shao , Chongzhao Wu This is my paper

Pith reviewed 2026-05-17 03:18 UTC · model grok-4.3

classification ⚛️ physics.optics
keywords FTIR imagessample pixel identificationunsupervised algorithmssupervised algorithmsmid-infrared spectroscopyimage processingbiological tissuesspectral data
0
0 comments X

The pith

Three unsupervised and supervised algorithms accurately identify sample pixels versus background in FTIR images of biological tissues.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces three algorithms for distinguishing sample pixels from background pixels in FTIR images, a key pre-processing step for effective feature extraction. Mid-InfraRed spectroscopy through FTIR spectrometers generates chemical maps by acquiring spectral data from multiple spatial points. The algorithms, realized in both unsupervised and supervised approaches, demonstrate accurate prediction of sample and background pixels. The supervised method additionally supports automatic detection, providing thorough solutions to the sample pixels detection problem in FTIR images.

Core claim

The authors present three algorithms using unsupervised and supervised approaches that accurately predict sample and background pixels in FTIR images, with the supervised method enabling automatic detection for improved FTIR signal processing.

What carries the argument

Unsupervised and supervised classification algorithms that separate pixels based on their spectral signatures in FTIR images.

If this is right

  • Accurate pixel identification enables more effective feature extraction from FTIR images.
  • The supervised approach allows for automatic detection without manual intervention.
  • These methods provide robust solutions for sample pixel detection in FTIR imaging.
  • Improved pre-processing contributes to better analysis of morphological and pathological alterations in biological tissues.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar algorithms might apply to other label-free spectroscopic imaging methods beyond FTIR.
  • Automation could facilitate larger-scale studies of tissue samples in pathology research.
  • Integration with real-time scanning devices could support on-the-fly analysis during experiments.

Load-bearing premise

FTIR images have sufficiently distinct spectral signatures between sample and background pixels that allow reliable separation by the algorithms without extra tuning.

What would settle it

Application of the algorithms to FTIR images where sample and background pixels have overlapping or similar spectral features, measuring if prediction accuracy significantly decreases.

Figures

Figures reproduced from arXiv: 2512.01585 by Chongzhao Wu, Jingzhu Shao, Xiangyu Zhao, Yudong Tian.

Figure 1
Figure 1. Figure 1: Workflow of the multivariate sample detection algorithms for FTIR images. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Details of the deep neural network (DNN) for automatic sample pixels detection in FTIR images. (a) The training procedure of the DNN. The spectrum and label of each pixel are paired as an ’input-ground truth’ issue. The DNN calculates the probability of the pixel belonging to the sample pixels based on the input spectrum, which is finally compared with the ground truth label for the error and its backward.… view at source ↗
Figure 3
Figure 3. Figure 3: Results of background detection with unsupervised linear regression approach on human thyroid tissue section [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Results of background detection with unsupervised linear regression approach on mouse kidney tissue section [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Variation of the Jaccard index between the unsupervised approach predictions and the ground truth as a function [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Results of background detection with supervised deep neural network approach on tissue section samples. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
read the original abstract

Mid-InfraRed spectroscopy is a promising label-free technique that can offer insights into morphological and pathological alterations in biological tissues at the molecular level. Owing to the development of the Fourier Transform InfraRed (FTIR) spectrometer, combined with scanning devices, FTIR images can be produced by simultaneously acquiring spectral data from multiple spatial points, generating comprehensive chemical maps. In the data pre-processing, the identification of the sample pixels, with the background pixels excluded, is important for further effective feature extraction in FTIR images. Here, we present three algorithms realized in unsupervised and supervised approaches for the identification of the sample pixels. The algorithms demonstrate accurate prediction results of the sample and background pixels, and the supervised method further enables the automatic detection. These findings highlight thorough and robust solutions to the sample pixels detection problem in FTIR images, contributing to the FTIR signal processing and future research with FTIR images.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript presents three algorithms realized in unsupervised and supervised approaches for identifying sample pixels versus background pixels in FTIR images of biological tissues. The central claim is that these algorithms achieve accurate prediction of sample and background pixels, with the supervised method additionally enabling automatic detection, thereby providing robust solutions for pre-processing FTIR data to support feature extraction.

Significance. If the performance claims hold under quantitative scrutiny, the work addresses a practical bottleneck in FTIR hyperspectral imaging for label-free biomedical analysis. Unsupervised methods could be especially useful where labeled training data are limited, while the supervised component offers automation potential. This could improve reproducibility in chemical mapping workflows, though the absence of metrics limits assessment of real-world impact.

major comments (1)
  1. [Abstract] Abstract: The assertion that 'the algorithms demonstrate accurate prediction results' is unsupported by any quantitative metrics, error analysis, validation procedures, or baseline comparisons (e.g., against standard clustering or classification methods). This is load-bearing for the central empirical claim and prevents evaluation of whether the methods reliably separate pixels based on distinct spectral signatures.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and constructive comments, which highlight important areas for strengthening the empirical support in our work. We address the major comment point by point below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The assertion that 'the algorithms demonstrate accurate prediction results' is unsupported by any quantitative metrics, error analysis, validation procedures, or baseline comparisons (e.g., against standard clustering or classification methods). This is load-bearing for the central empirical claim and prevents evaluation of whether the methods reliably separate pixels based on distinct spectral signatures.

    Authors: We agree that the abstract claim would be more robust with explicit quantitative support. The manuscript presents visual results and qualitative demonstrations of pixel identification in FTIR images of biological tissues, showing clear separation based on spectral features for both unsupervised and supervised variants. However, to directly address this concern and enable proper evaluation, we will revise the abstract to reference specific performance metrics from our experiments. We will also add a dedicated quantitative evaluation subsection in the results, including accuracy, precision, recall, and F1 scores, along with error analysis, cross-validation details, and direct comparisons to baselines such as k-means clustering (unsupervised) and support vector machines (supervised). These additions will substantiate the reliability of the spectral-signature-based separation. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents three algorithms (unsupervised and supervised) for identifying sample versus background pixels in FTIR images and reports that they achieve accurate empirical results on the task. No derivation chain, equations, fitted parameters, or self-citations are described that reduce a claimed prediction or result to the input data by construction. The central claim rests on the observed performance of the algorithms rather than any self-referential definition or load-bearing self-citation, making the work self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based on abstract only, the central claim rests on standard assumptions about spectral separability in FTIR data and the applicability of generic unsupervised/supervised learning methods. No explicit free parameters, axioms, or invented entities are stated.

axioms (1)
  • domain assumption FTIR images exhibit distinguishable spectral features between sample tissue and background regions
    Implicit in the problem setup for pixel identification

pith-pipeline@v0.9.0 · 5453 in / 1206 out tokens · 30768 ms · 2026-05-17T03:18:21.823358+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

  1. [1]

    Benex K Babu, Murali Manohar Yadav, Shantanu Singh, and Vijay Kumar Yadav. Fuel forensics: recent advance- ments in profiling of adulterated fuels by atr-ftir spectroscopy and chemometric approaches.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 312:124049, 2024

    work page 2024

  2. [2]

    Bo Deng, Lingfen Rao, and Lucia Rodriguez-Freire. Evaluation and optimization of ftir spectroscopy to quantify pha production by municipal wastewater sludge.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 312:124012, 2024

    work page 2024

  3. [3]

    Surface-enhanced infrared absorption sensor incorporating liquid galinstan with three-dimensional metagratings.Acs Photonics, 11(5):1857– 1865, 2024

    Jiang Zhu, Wei Wei, Bo Chen, Ping Tang, Xiangyu Zhao, and Chongzhao Wu. Surface-enhanced infrared absorption sensor incorporating liquid galinstan with three-dimensional metagratings.Acs Photonics, 11(5):1857– 1865, 2024

    work page 2024

  4. [4]

    Fundamental developments in infrared spectroscopic imaging for biomedical applications.Chemical Society reviews, 45(7):1935–1957, 2016

    Michael Pilling and Peter Gardner. Fundamental developments in infrared spectroscopic imaging for biomedical applications.Chemical Society reviews, 45(7):1935–1957, 2016

    work page 1935

  5. [5]

    Automatic identification of paraffin pixels on ftir images acquired on ffpe human samples.Analytical Chemistry, 93(8):3750–3761, 2021

    Warda Boutegrabet, Dominique Guenot, Olivier Bouché, Camille Boulagnon-Rombi, Aude Marchal Bressenot, Olivier Piot, and Cyril Gobinet. Automatic identification of paraffin pixels on ftir images acquired on ffpe human samples.Analytical Chemistry, 93(8):3750–3761, 2021

    work page 2021

  6. [6]

    Tumor margin identification and prediction of the primary tumor from brain metastases using ftir imaging and support vector machines.Analyst, 138(14):3983–3990, 2013

    Norbert Bergner, Bernd FM Romeike, Rupert Reichart, Rolf Kalff, Christoph Krafft, and Jürgen Popp. Tumor margin identification and prediction of the primary tumor from brain metastases using ftir imaging and support vector machines.Analyst, 138(14):3983–3990, 2013

    work page 2013

  7. [7]

    Characterization of human breast cancer tissues by infrared imaging.Analyst, 141(2):606–619, 2016

    Magali Verdonck, Amandine Denayer, Benjamin Delvaux, Soizic Garaud, Roland De Wind, Christine Desmedt, Christos Sotiriou, Karen Willard-Gallo, and Erik Goormaghtigh. Characterization of human breast cancer tissues by infrared imaging.Analyst, 141(2):606–619, 2016

    work page 2016

  8. [8]

    Investigation of the effect of pd-l1 blockade on triple negative breast cancer cells using fourier transform infrared spectroscopy.V accines, 7(3):109, 2019

    Mohamed HM Ali, Salman M Toor, Fazle Rakib, Raghvendra Mall, Ehsan Ullah, Kamal Mroue, Prasanna R Kolatkar, Khalid Al-Saad, and Eyad Elkord. Investigation of the effect of pd-l1 blockade on triple negative breast cancer cells using fourier transform infrared spectroscopy.V accines, 7(3):109, 2019

    work page 2019

  9. [9]

    A simple model for cell type recognition using 2d-correlation analysis of ftir images from breast cancer tissue.Journal of Molecular Structure, 1163:472–479, 2018

    Mohamed H Ali, Fazle Rakib, Khalid Al-Saad, Rafif Al-Saady, Fiona M Lyng, and Erik Goormaghtigh. A simple model for cell type recognition using 2d-correlation analysis of ftir images from breast cancer tissue.Journal of Molecular Structure, 1163:472–479, 2018. 10 APREPRINT

    work page 2018

  10. [10]

    Characterization of colorectal mucus using infrared spectroscopy: a potential target for bowel cancer screening and diagnosis.Laboratory Investigation, 100(8):1102–1110, 2020

    Jayakrupakar Nallala, Charles Jeynes, Sarah Saunders, Neil Smart, Gavin Lloyd, Leah Riley, Debbie Salmon, and Nick Stone. Characterization of colorectal mucus using infrared spectroscopy: a potential target for bowel cancer screening and diagnosis.Laboratory Investigation, 100(8):1102–1110, 2020

    work page 2020

  11. [11]

    Micro atr-ftir spectroscopic imaging of colon biopsies with a large area ge crystal.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 228:117695, 2020

    Cai Li Song and Sergei G Kazarian. Micro atr-ftir spectroscopic imaging of colon biopsies with a large area ge crystal.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 228:117695, 2020

    work page 2020

  12. [12]

    Jingzhu Shao, Xiangyu Zhao, Ping Tang, Bo Chen, Borui Xu, Han Lu, Zhen Qin, and Chongzhao Wu. Label-free investigation of infected acute pyelonephritis tissue by ftir microspectroscopy with unsupervised and supervised analytical methods.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 321:124753, 2024

    work page 2024

  13. [13]

    A label-free approach by infrared spectroscopic imaging for interrogating the biochemistry of diabetic nephropathy progression

    Vishal K Varma, Andre Kajdacsy-Balla, Sanjeev K Akkina, Suman Setty, and Michael J Walsh. A label-free approach by infrared spectroscopic imaging for interrogating the biochemistry of diabetic nephropathy progression. Kidney international, 89(5):1153–1159, 2016

    work page 2016

  14. [14]

    Fourier transform infrared applications to investigate induced bio- chemical changes in liver.Applied Spectroscopy Reviews, 55(9-10):840–872, 2020

    Sumedha Liyanage and Noureddine Abidi. Fourier transform infrared applications to investigate induced bio- chemical changes in liver.Applied Spectroscopy Reviews, 55(9-10):840–872, 2020

    work page 2020

  15. [15]

    Diagnosis of lung cancer by atr-ftir spectroscopy and chemometrics.Frontiers in Oncology, 11:753791, 2021

    Xien Yang, Quanhong Ou, Kai Qian, Jianru Yang, Zhixun Bai, Weiye Yang, Youming Shi, and Gang Liu. Diagnosis of lung cancer by atr-ftir spectroscopy and chemometrics.Frontiers in Oncology, 11:753791, 2021

    work page 2021

  16. [16]

    Huanyu Li, Jun Wang, Xiaoyun Li, Xianhong Zhu, Shaomei Guo, Hongluan Wang, Jie Yu, Xiaoqun Ye, and Fan He. Comparison of serum from lung cancer patients and from patients with benign lung nodule using ftir spectroscopy.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 306:123596, 2024

    work page 2024

  17. [17]

    Comparison of mid-infrared handheld and benchtop spectrometers to detect staphylococcus epidermidis in bone grafts.Bioengineering, 10(9):1018, 2023

    Richard Lindtner, Alexander Wurm, Katrin Kugel, Julia Kühn, David Putzer, Rohit Arora, Débora Cristina Coraça-Huber, Philipp Zelger, Michael Schirmer, Jovan Badzoka, et al. Comparison of mid-infrared handheld and benchtop spectrometers to detect staphylococcus epidermidis in bone grafts.Bioengineering, 10(9):1018, 2023

    work page 2023

  18. [18]

    Characterization of ovarian cancer cells and tissues by fourier transform infrared spectroscopy.Journal of ovarian research, 11(1):64, 2018

    Lei Li, Xiaoning Bi, Hengzi Sun, Simiao Liu, Mei Yu, Ying Zhang, Shifu Weng, Limin Yang, Yanan Bao, Jinguang Wu, et al. Characterization of ovarian cancer cells and tissues by fourier transform infrared spectroscopy.Journal of ovarian research, 11(1):64, 2018

    work page 2018

  19. [19]

    Investigation on the cancer invasion and metastasis of skin squamous cell carcinoma by raman spectroscopy.Molecules, 24(11):2059, 2019

    Xu Zhang, Fan Yu, Jie Li, Dongliang Song, Heping Li, Kaige Wang, Qingli He, and Shuang Wang. Investigation on the cancer invasion and metastasis of skin squamous cell carcinoma by raman spectroscopy.Molecules, 24(11):2059, 2019

    work page 2059

  20. [20]

    Expression data of fos and jun genes and ftir spectra provide diagnosis of thyroid carcinoma.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 304:123305, 2024

    João Paulo da Silva Queiroz, Breno Pupin, Tanmoy Tapobrata Bhattacharjee, Miyuki Uno, Roger Chammas, Marco Aurelio Vamondes Kulcsar, and Renata de Azevedo Canevari. Expression data of fos and jun genes and ftir spectra provide diagnosis of thyroid carcinoma.Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 304:123305, 2024

    work page 2024

  21. [21]

    Combination of ftir spectral imaging and chemometrics for tumour detection from paraffin-embedded biopsies.Analyst, 133(2):197–205, 2008

    Elodie Ly, Olivier Piot, Rolf Wolthuis, Anne Durlach, Philipe Bernard, and Michel Manfait. Combination of ftir spectral imaging and chemometrics for tumour detection from paraffin-embedded biopsies.Analyst, 133(2):197–205, 2008

    work page 2008

  22. [22]

    Extended multiplicative signal correction in vibrational spectroscopy, a tutorial.Chemometrics and Intelligent Laboratory Systems, 117:92–99, 2012

    Nils Kristian Afseth and Achim Kohler. Extended multiplicative signal correction in vibrational spectroscopy, a tutorial.Chemometrics and Intelligent Laboratory Systems, 117:92–99, 2012. 11

    work page 2012