Spectrum Image Preprocessing for AI Analyte Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing electrochemical measurement spectra, such as chronoamperometry and cyclic voltammetry, are inefficient as they ignore non-faradaic currents and information about interfering substances, leading to incomplete feature extraction and inaccurate analyte prediction.
Innovation Solution
An apparatus and method that preprocesses the measurement spectrum image by drawing an axis and filling or inverting partial areas, then uses a convolutional neural network (CNN) to learn and predict the features of the analyte from the test sample spectrum image, enhancing learning efficiency and feature extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods for analyzing electrochemical measurement spectra are used, then the analysis process is simple, but the feature extraction is incomplete and prediction accuracy is low
Solution Approach 1:
The patent applies preliminary action by performing spectrum image preprocessing before analysis. The system converts raw electrochemical spectra into standardized images, applies noise filtering, and enhances feature visibility in advance. This preprocessing step prepares the data optimally for subsequent AI analysis, improving prediction accuracy without adding excessive complexity during the main analysis phase.
Solution Approach 2:
The patent introduces an intermediary approach by using spectrum images as a mediator between raw electrochemical data and AI analysis. Instead of directly analyzing complex electrochemical signals, the system transforms them into visual spectrum images that preserve essential features while making them more suitable for image-based machine learning algorithms, thereby bridging the gap between chemical data and computational analysis.
2Productivity
If existing spectrum analysis methods are used, then the processing speed is fast, but the learning efficiency of spectrum images is low
Solution Approach 1:
The system performs preliminary transformation of electrochemical spectra into image format with optimized dimensions and contrast enhancement before feeding them to the neural network. This preprocessing prepares the data in an optimal format that accelerates subsequent learning operations, reducing the computational burden during training and inference phases.
Solution Approach 2:
The patent applies parameter changes by adjusting key parameters of the spectrum images during preprocessing, such as normalization ranges, contrast enhancement factors, and dimension scaling. These parameter optimizations make the input data more suitable for the neural network architecture, improving convergence speed and learning efficiency while maintaining reasonable preprocessing time.
Data Source
AI summary
One embodiment of the present invention can provide a method of analyzing a measured spectrum image including the steps of obtaining a measured spectrum image; creating an analysis model by learning the spectrum image with artificial intelligence; and predicting the feature of a test sample of an analyte by inputting the spectrum image of the test sample to the analysis model.