Adaptive spectral-fluorescence image fusion for Ca²⁺ detection based on modification-free fiber-optic probe sensors

The precise determination of calcium ion (Ca²⁺) concentration is essential for water quality assessment. Nevertheless, current detection methods commonly face challenges related to operational complexity and have limitations in the effective integration of dual-mode heterogeneous data. In this paper, a dual-mode detection method based on spectral-fluorescence image fusion is proposed, which utilizes a modification-free multimode fiber-optic probe sensor. A range-aware adaptive fusion stacking ensemble (RAAFSE) framework is designed, which addresses spectral-fluorescence image dual-mode heterogeneous data, and achieves concentration prediction through deep feature fusion. Additionally, a modular portable detection system has been developed, which uses calcein as the fluorescent indicator. The specific chelation reaction between calcein and Ca²⁺ generates fluorescence signals, enabling the simultaneous acquisition of spectral curves and fluorescence spatial distribution images. Experimental results demonstrate that the proposed sensor exhibits excellent detection performance over a wide Ca²⁺ concentration range of 0.14-3000 µM. The maximum detection sensitivities in the low-, medium-, and high-concentration regions are 0.0032 µM^-1, 0.0015 µM^-1, and 0.0006 µM^-1, respectively. The prediction performance of the proposed model is evaluated using root mean square error (RMSE) and mean absolute error (MAE), yielding values of 27.12 µM and 22.17 µM, respectively. This study, taking Ca²⁺ detection as an example, achieved a wide-range, specific and rapid detection. This approach is characterized by high accuracy, simple structure, low cost, and user-friendly operation, indicating broad application prospects in the environmental detection field.