Artificial neural network model for photosynthetic pigments identification using multi wavelength chromatographic data

The development of rapid and automatic pigment characterization method become an important issue due to the fact that there are only less than 1% of plant pigments in the earth have been explored. In this research, a mathematical model based on artificial intelligence approach was developed to simpl...

Full description

Saved in:
Bibliographic Details
Main Authors: Prilianti, K. R., Hariyanto, S., Natali, F. D. D., Indriatmoko, Adhiwibawa, M. A. S., Limantara, L., Brotosudarmo, T. H. P.
Format: Conference Proceeding
Language:English
Subjects:
Artificial neural networks
Brain
Carotene
Chromatography
Fingerprints
High performance liquid chromatography
Lutein
Model accuracy
Neural networks
Photosynthesis
Pigments
Principal components analysis
Two dimensional models
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The development of rapid and automatic pigment characterization method become an important issue due to the fact that there are only less than 1% of plant pigments in the earth have been explored. In this research, a mathematical model based on artificial intelligence approach was developed to simplify and accelerate pigment characterization process from HPLC (high-performance liquid chromatography) procedure. HPLC is a widely used technique to separate and identify pigments in a mixture. Input of the model is chromatographic data from HPLC device and output of the model is a list of pigments which is the spectrum pattern is discovered in it. This model provides two dimensional (retention time and wavelength) fingerprints for pigment characterization which is proven to be more accurate than one dimensional fingerprint (fixed wavelength). Moreover, by mimicking interconnection of the neuron in the nervous systems of the human brain, the model have learning ability that could be replacing expert judgement on evaluating spectrum pattern. In the preprocessing step, principal component analysis (PCA) was used to reduce the huge dimension of the chromatographic data. The aim of this step is to simplify the model and accelerate the identification process. Six photosynthetic pigments i.e. zeaxantin, pheophytin a, α-carotene, β-carotene, lycopene and lutein could be well identified by the model with accuracy up to 85.33% and processing time less than 1 second.
ISSN:0094-243X
1551-7616
DOI:10.1063/1.4945074