Simulation monitoring of tetracyclines in wastewater based on fluorescence image processing and machine learning classifier

Long Yu,Xin Li,Yunfei Lu,Pengchen Su,Mi Yuan,Qiang Zhang,Lingxiao Wang,Degui Wang,Mingtai Sun,Suhua Wang
DOI: https://doi.org/10.1016/j.snb.2023.133678
2023-03-18
Abstract:Antibiotics, as an emerging pollutant, have attracted extensive attention due to their trace residues and toxic side effects in the environment. Here, we constructed a new detection mechanism for tetracyclines in sewage by online trace sensing through machine learning (ML) algorithm and image fingerprint extraction method. The image fingerprint information was obtained from the blue to red specific fluorescence response generated by the interaction between the luminescent metal-organic framework (MOF) with tetracyclines, which greatly improved the ability against background interference. Fluorescence images of samples with different concentrations were collected, and the RGB mean value of sample images was extracted through the program as "fingerprint information". Then k-means clustering model (KCM) and hierarchical clustering model (HCM) were constructed to analyze the differences in fingerprint information of samples with different concentrations by clustering analysis method in machine learning. The clustering results showed good clustering characteristics of samples with similar concentrations, which provided reliable support for rapid detection while avoiding the use of expensive large instruments and time-consuming analysis methods.
chemistry, analytical,electrochemistry,instruments & instrumentation
What problem does this paper attempt to address?