Tracking microfibres along the sludge line of a full-scale wastewater treatment plant using conventional microscopy and AI-based image analysis

Wastewater treatment plants (WWTPs) are key pathways for microfibres (MFs) entering the environment. This study evaluated the occurrence of MFs along the sludge treatment line of a WWTP by comparing a traditional stereomicroscopy-based method with an artificial intelligence (AI) image analysis tool, MicrofibreDetect (MFD). Results from the traditional method showed a high prevalence of MFs in all sludge streams. Transparent MFs, mainly associated with cellulose materials, were the most abundant, accounting for 78% of total MFs. The highest concentrations were found in primary sludge, likely due to the retention of cellulose fibres from toilet paper. Most fibres, both transparent and coloured, ranged between 100 and 500 μm. When comparing both approaches, MFD detected 80.5% of coloured MFs and 74.9% of transparent MFs, with better performance in liquid streams. Undetected fibres were mainly very light-coloured MFs that were difficult to distinguish from the filter background. Regarding length estimation, MFD correctly measured 65.0% of coloured MFs and 57.1% of transparent MFs, with errors mainly due to curved shapes and shadow effects. Overall, MFD significantly reduced analysis time by 97.5%, demonstrating its potential as a rapid and efficient alternative for MF detection in sludge treatment processes.