Quick Guide
| Introduction | Features | Overview |
1
| Data Overview |
The PlastiTrax Finder is operated through a simple, intuitive web interface that guides users through the complete analysis workflow. The platform supports data upload from the most common file formats generated by leading imaging systems (incl. Bruker, Perkin Elmer, Thermo Scientific, Agilent) and is compatible with both transmission and reflection measurement techniques.
Once the spectral data have been uploaded, the data analysis can be started with a few clicks. All completed analyses are stored securely within the integrated workspace, allowing users to review previous results, continue interrupted projects, and access historical datasets at any time.
Before exporting, all identified particles can be reviewed manually, allowing users to verify the results and remove potential false-positive identifications with a single click. In addition, powerful filtering functions enable the selective review of particles based on user-defined criteria, such as correlation score, making it easy to identify and remove low-confidence matches. Once the review is complete, PlastiTrax Finder enables direct export of both the chemical map and a .csv file containing all identified particles, facilitating documentation, further data analysis, and integration into external reporting workflows.
After the analysis is completed, results can be exported directly from the platform. PlastiTrax Finder provides both the chemical map and a .csv file containing all identified particles, enabling straightforward documentation, further data processing, and integration into external workflows or reporting systems.
2
| Chemical and Optical Image |
The Chemical Map provides a visual overview of all identified microplastic particles on the detected area. For enhanced interpretation, the chemical map can be overlaid with the corresponding optical image, allowing direct comparison between the particle morphology and the spectroscopic identification. Moving the cursor over a particle displays its particle ID and assigned polymer type. By simply clicking on a particle, users can access detailed spectral information, including the measured spectrum (average spectrum of a single particle) and the corresponding reference spectrum, enabling rapid verification of each individual identification.
3
| Spectral Information |
The Spectral Information panel provides detailed insight into every identified particle. Instead of displaying a single measurement point, PlastiTrax Finder calculates and plots the average spectrum of all spectra acquired for an individual particle, resulting in a more representative and robust spectral profile. This average spectrum is displayed together with the corresponding reference spectrum from the spectral database, allowing users to visually verify the quality of the identification.
In addition, the software displays an Attention Map as an overlaid bar chart. This visualization highlights the relative importance of individual wavelength regions used by the AI model during the classification process. Peaks with a higher attention score indicate spectral features that contributed most strongly to the polymer identification, providing greater transparency into the AI's decision-making process and helping users understand which spectral regions were most influential for the final classification.
4
| Particle Table |
The Particle Table provides a comprehensive overview of all identified particles and enables efficient quality control prior to data export. Each entry includes a checkbox, allowing individual particles to be excluded from the final results with a single click, for example in the case of false-positive identifications. The table further displays the Particle ID (a unique sequential identifier), the assigned polymer type, the AI Confidence, and the Pearson Correlation.
The AI Confidence is expressed as a percentage and indicates how confidently the AI model assigns a particle to a specific polymer class. In parallel, the Pearson Correlation provides a percentage-based measure of the mathematical similarity between the particle spectrum and the corresponding reference spectrum from the database. By combining AI confidence with a classical correlation metric, users obtain two indicators to assess the reliability of each identification.
To further simplify data validation, the Particle Table includes a wide range of filtering options. Users can rapidly filter particles based on parameters such as polymer type, confidence score, correlation value, or other analytical criteria. This enables the fast identification of low-confidence or potentially false-positive results, which can then be removed from the final dataset with a single click, ensuring a transparent and efficient review process.