Laser-induced Breakdown Spectroscopy (LIBS) is an increasingly popular method for elemental bioimaging, primarily due to its fast and multi-elemental analysis. However, achieving quantitative accuracy is highly challenging, mainly due to the strong matrix effect present in biological matrices and the lack of spatially adaptive calibration strategies. When using the conventional pixel-to-pixel calibration approach, it typically assumes a uniform sample matrix, which leads to high quantification errors when applied to biological and plant tissues. Therefore, there is a need for novel calibration methods based on a delocalized approach that take matrix variability into consideration when dealing with biological samples. This study introduces a novel delocalized approach for quantitative bioimaging of elements, particularly cadmium (Cd) and calcium (Ca), in plant tissue. A controlled sample set of Cannabis sativa plants contaminated with three different Cd concentrations was analysed at matching spatial resolution using micro-X-ray fluorescence (micro-XRF) and LIBS. A conventional pixel-to-pixel calibration was initially employed as a baseline strategy but yielded high mean absolute percentage errors (MAPE) exceeding 40 % for Cd. Therefore, a delocalised approach was developed to overcome these limitations, leveraging clustering algorithms to construct a matrix-based calibration model. This method significantly improved quantification accuracy, reducing MAPE for Cd to as low as 8.7 %, while Ca quantification achieved a score of 1.1 % MAPE. The model also exhibited minimal bias, with errors in the parts-per-million range. These results demonstrate that advanced feature selection and clustering-based calibration enable accurate quantification in highly heterogeneous plant matrices. The proposed delocalized approach demonstrates a significant advance in LIBS imaging methodology by addressing a key limitation in pixel-to-pixel calibration and may be applicable to a broader range of elements in other biological and heterogeneous systems.
扫码关注我们
求助内容:
应助结果提醒方式:
