Luciferase-based bioassay for rapid pollutants detection and classification by means of multilayer artificial neural networks

Тип публикации: статья из журнала

Год издания: 2017

Идентификатор DOI: 10.1016/j.snb.2016.11.071

Ключевые слова: Bioluminescence, Luciferase, Bioassay, Artificial neural networks, Perceptron, Machine learning

Аннотация: Biosensors for rapid environmental pollution detection can be designed with biomodule based on the bacterial bioluminescent system. Usually this method returns total value of toxicity and does not allow to distinguish pollutants types. Herein we demonstrate the classification of pollutants by the kinetic analysis utilizing artificial neural networks with multilayer perceptron architecture. The kinetics of light emission of NAD(P)H:FMN-oxidoreductase-luciferase bioluminescent reaction was measured for clean water and in the presence of three environment pollutants (1,4-benzoquinone, copper sulfate and 1,3dihydroxybenzene) separately with various concentrations. The efficiency of using multilayer perceptron with sigmoid activation function for processing of kinetics of light emission was estimated. It was shown that multilayer perceptrons allowing to distinguish pollutant class and concentration after sufficient training. The architecture consisted of 61 inputs neurons, 3 hidden layers and 3 output neurons was found optimum in sense of learning time for classification of three pollutants. Usage of simplest activation function sigmoid and backpropagation method for multilayer perceptron teaching providing the results been useful for smart signal processing in computational modules of biosensors. (C) 2016 Elsevier B.V. All rights reserved.

Ссылки на полный текст

Издание

Журнал: SENSORS AND ACTUATORS B-CHEMICAL

Выпуск журнала: Vol. 242

Номера страниц: 653-657

ISSN журнала: 09254005

Место издания: LAUSANNE

Издатель: ELSEVIER SCIENCE SA

Авторы

  • Denisov Ivan A. (Siberian Fed Univ, Lab Bioluminescent Biotechnol, Krasnoyarsk 660041, Russia)

Вхождение в базы данных

Информация о публикациях загружается с сайта службы поддержки публикационной активности СФУ. Сообщите, если заметили неточности.

Вы можете отметить интересные фрагменты текста, которые будут доступны по уникальной ссылке в адресной строке браузера.