NanoMetrix: An app for chemometric analysis from near infrared spectra

Abstract

Near-infrared spectroscopy (NIR) associated with chemometric methods has been widely used in the determination of various compounds, as it is a fast, nondestructive, and analytical method with minimal sample preparation and universal application. In addition, the trend of miniaturization has allowed the development of new portable equipment, increasing the possibilities of point-of-use analysis at low cost. In this context, the objective of this study was to develop a portable analytical methodology using a low-cost DLP NIRscan Nano spectrophotometer (Texas Instruments ®) connected to a smartphone for data storage. For this, an app called NanoMetrix was developed, created natively for Android from Android Studio IDE, according to a software design standard known as Model-View-ViewModel (MVVM). To demonstrate the applicability of the method, 36 samples of plant tissue were analyzed to determine total nitrogen content. The calibration models were compared to those obtained with the SOLO+MIA software (Eigenvector Research, Inc.), 8.6.1. The results presented root mean square error of calibration (RMSEC) 1.90 and 1.82 g kg −1 , root mean square error of prediction (RMSEP) 2.00 and 1.97 g kg −1 , and R 2 pred 0.973 and 0.975, for the NanoMetrix application and the SOLO+MIA software, respectively. Therefore, the results indicate that the NanoMetrix integrated with the portable spectrophotometer, besides the acquisition of spectra, allows the processing of data in the field and reducing the time for the analysis and obtaining the results. Also, the chemometric associated with the NIR spectroscopy is a viable alternative to replace, or complement, the methods used for nitrogen determination.

Fast and low‐cost method for direct and simultaneous determination of nitrogen and carbon in soybean leaves using benchtop and portable near‐infrared devices

“…For data acquisition and storage, the spectrophotometer was linked to a smartphone through Bluetooth®. 6 Development of multivariate calibration models The partial least-squares calibration models were developed with the aid of the SOLO+MIA software (Eigenvector Research Inc., Wenatchee, WA, USA), with 27 soybean leaf samples. These regression models were applied in a second set of soy leaf samples (13 samples) collected from different soybean crops (2021 and 2022).…”

“…Therefore, it is an efficient alternative to the traditional analytical methods used for leaf tissue analysis in agronomic laboratories, which are admittedly more expensive and time-consuming. 6,8,[10][11][12] In this context, the purpose of this study was to develop a green analytical methodology 13,14 using NIRS and chemometric methods for the direct and simultaneous determination of nitrogen and carbon content in soybean leaves and to compare the performance in benchtop and portable spectrophotometers. Furthermore, this study aims to replicate genuine conditions for the analysis of plant tissue samples within an Figure 2.…”

“…These techniques include near-infrared spectroscopy (NIRS) and chemometric methods for classifying or quantifying primary macronutrients in plants. [5][6][7][8] Another typical technique is hyperspectral remote sensing, which has been used to monitor changes in leaf traits and estimate soybean leaf carbon and nitrogen content. Additionally, digital images obtained using remotely piloted aircraft have been utilized to detect and monitor abiotic stresses in soybeans using digital aerial images.…”

“…To meet the need for plant nutritional diagnosis, some studies have demonstrated the use of non‐destructive techniques for estimating nitrogen content in plant tissue. These techniques include near‐infrared spectroscopy (NIRS) and chemometric methods for classifying or quantifying primary macronutrients in plants 5‐8 . Another typical technique is hyperspectral remote sensing, which has been used to monitor changes in leaf traits and estimate soybean leaf carbon and nitrogen content.…”

Multispectral Cameras and Machine Learning Integrated into Portable Devices as Clay Prediction Technology

“…In addition, machine learning tools as Partial Least Regression (PLSR) have been applied for multivariate calibration in soil spectroscopy [20], images [21] and sensor data [22]. These algorithms eliminate variables that do not correlate with the property of interest, such as those that add noise, non-linearities or irrelevant information [23].…”