Image-Based Colorimetric Quantification of Boron Using a Harris Hawks Optimization-Optimized Artificial Neural Network
This study proposes an image-based colorimetric quantification framework that combines color features extracted from the red, green, blue (RGB), hue, saturation, value (HSV) and International Commission on Illumination L*a*b* (CIELab) color spaces with an artificial neural network optimized by Harris Hawks optimization (HHO-ANN). From each image, 27 color features were constructed from the reaction region, reference region, and the difference between the two regions. The HHO-ANN model achieved a coefficient of determination (R2) of 0.9968, a root mean square error of calibration (RMSEC) of 1.5 µg/mL and a root mean square error of prediction (RMSEP) of 1.2 µg/mL, with recoveries of 99.1–103.7%. To illustrate the applicability of the modeling framework, boron was selected as a case study, and the model was integrated into the Quantivis application to process images of reaction paper strips automatically and estimate the boron concentration. For artificially prepared samples, the results obtained with the HHO-ANN model were not significantly different from those obtained by UV-Vis spectrophotometry according to the paired Student’s t-test.
Authors
- Nguyễn Duy Lưu
- Nguyen Xuan Huy (ORCID: https://orcid.org/0009-0005-0964-7294)
- Nguyen Hung Nhat Duy (ORCID: https://orcid.org/0009-0002-3367-8611)
- Trinh Hoang Nhan (ORCID: https://orcid.org/0009-0007-5799-4324)
Institutions
- Trường Đại học Kỹ thuật Y - Dược Đà Nẵng (VN)
Publication Details
- Journal
- Analytical Letters
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1080/00032719.2026.2767164
- Primary Topic
- Spectroscopy and Chemometric Analyses
- Type
- article
- Field-Weighted Citation Impact
- 0.00