An Impact Innovation Funded Project
Priority Area:
Beyond Lifecycle Analysis
FasTEX: FAST identification of TEXtile fibres by Fourier transform infrared spectroscopy (FTIR) and machine learning (ML)
Textile pollution, particularly from microfibres released during washing or wearing clothes, is a growing environmental concern. To understand and address this issue, researchers must be able to rapidly identify the different types of fibres present in environmental samples. This requires knowing what the fibres are made of. However, current methods for determining fibre composition are slow, and often rely on human judgment, which can lead to inconsistency.
This project aims to revolutionise how textile fibres are identified by using a combination of infrared spectroscopy and artificial intelligence. By training machine learning models to recognise different fibre types from their chemical signatures, the process can be made faster, more reliable, and fully automated.
The result will be a powerful new tool that speeds up fibre analysis. FasTEX will support more accurate and efficient studies of textile shedding and microfibre pollution, helping researchers and policymakers better understand where harmful fibres are coming from and how to reduce their environmental impact.
Dr. Matteo Gallidabino
Lecturer in Forensic Chemistry
Matteo is an Assistant Professor in Forensic Chemistry at King`s College London (UK). Having obtained his PhD from the University of Lausanne, his highly inter-disciplinary research focuses on the analysis of chemical and trace evidence.
Name of Lead Applicant:
Job Title:
Organisation:
Additional Team Members:
Dr Matteo Gallidabino
Lecturer in Forensic Chemistry
King’s College London
N/A
At present, the characterisation of textile fibres in environmental samples involves multiple manual and time-consuming steps. The process typically begins with fibre recovery, followed by morphological and chemical analysis using sequential techniques such as high-power light microscopy (HPM), polarised light microscopy (PLM), and Fourier transform infrared (FTIR) spectroscopy. While accurate, this workflow is labour-intensive and subject to interpretative bias due to human involvement.
FasTEX (FAST identification of TEXtile fibres using FTIR and machine learning) aims to address this by developing an AI-assisted workflow capable of automating the interpretation of FTIR spectra for textile fibre identification and characterisation. By integrating chemometric and machine learning (ML) approaches, the project seeks to accelerate fibre analysis, improve consistency, and ultimately lay the foundations for a standardised, high-throughput method suitable for environmental monitoring and regulatory purposes.
This is a novel and timely opportunity to apply AI tools in a practical, interdisciplinary context, bridging material science, environmental chemistry, and forensic analysis.
This short-term, high-impact project will build a solid foundation for future applications in regulatory science, environmental monitoring, sustainable textile development, and academic-industry partnerships focused on textile circularity and pollution prevention.
This project will focus on four interlinked objectives:
- Establish a curated spectral database of textile fibres: we will collect and catalogue FTIR spectra from a diverse selection of textile fibres using three reference collections. Spectral acquisition will be carried out and measurements will be performed on fibre tuffs. Spectra will be collected across the mid-infrared range and stored alongside structured metadata to ensure traceability and reusability.
- Develop and test an initial machine learning classification model: using this spectral dataset, we will explore a range of supervised machine learning algorithms. Models will be tuned using cross-validation with resampling strategies, aiming to maximise classification accuracy while maintaining generalisability.
- Demonstrate method performance on test samples: the trained model will be evaluated using a small set of textile fibres obtained from garments of known composition. We will also explore potential sources of misclassification (e.g. overlapping spectral features, mixed fibre content) to identify key areas for improvement.
- Define next steps for tool refinement and broader application: building on the results of this proof-of-concept study, we will critically assess the limitations of the current approach and propose concrete avenues for improvement.
This is a novel and timely opportunity to apply AI tools in a practical, interdisciplinary context, bridging materials science, environmental chemistry, and forensic analysis. FasTEX aligns strongly with the IMPACT+ vision by:
- Moving beyond traditional life cycle analysis, offering tools that can support environmental research and pollution mitigation post-consumer use.
- Contributing to transparent pathways by reducing subjectivity in fibre identification and supporting data reproducibility.
- Strengthening circular knowledge systems, as the models and spectral libraries developed could be shared and reused across research and policy contexts.
This short-term, high-impact project will build a solid foundation for future applications in regulatory science, environmental monitoring, sustainable textile development, and academic-industry partnerships focused on textile circularity and pollution prevention.
Insights from this project will also support the development of future research proposals aimed at scaling the method to FTIR microscopy and environmental monitoring applications, contributing to standardisation efforts in microfibre analysis and informing data-driven sustainability initiatives across the fashion and textile value chain.
Expected start date: 15 July 2025
Expected end date: 15 December 2025
Project duration: 5 months