Scholarlink

<p>To help detect cancer at an early stage and reduce the mortality rate, a Surface Plasmon Resonance (SPR) biosensor based on optical fiber has been proposed in this study. The SPR mechanism is created by a novel ZnO/Ag/graphene design, to accurate detection of six different cancer cells. In addition, in this article, we present new approaches for modeling SPR biosensors using artificial neural networks (ANN). Thus, a multilayer perceptron regression model (MLP) was used to predict the sensitivity values of the SPR biosensor and the effect of physical parameters on the SPR response especially the Detection Accuracy (DA) and the Quality Factor (QF). In results, the use of ZnO layer provide stronger electric field localization and enhanced plasmonic coupling. Subsequently the maximum wavelength sensitivity is achieved as 4200 nm/RIU, 5571.42 nm/RIU, 5208.33 nm/RIU, 6428.57 nm/RIU, 6928.57 nm/RIU, 7285.71 nm/RIU for the detection of cancer cells in skin (basal), cervical (HeLa), adrenal gland (PC12), blood (Jurkat), and breast (MCF-7 and MDA MB-231) respectively. Moreover, the proposed sensor achieves a high QF of 35.0483 RIU<sup>−1</sup> along with a DA of 0.0053 nm<sup>−1</sup>. The MLP-ANN model showed good performance metrics. In addition, the optimization of various physical and geometric parameters was successfully predicted using the same approach, showing strong consistency between predicted and actual data in each case. Furthermore, our results prove the potential of the MLP-ANN model to accurately approximate the performance of the biosensor and highlight its ability to significantly simplify and accelerate the design and optimization of this type of biosensor. Thanks to the advanced results and the application of machine learning, this sensor can be used as a fast, effective and inexpensive cancer detection device.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance Prediction Using a Machine Learning Model of a Highly Sensitive Surface Plasmon Resonance-based Fiber Optic Biosensor for Cancer Cells Detection

  • 期刊论文

  • 发表日期：

    2026年7月24日

查看全文
  • Yosra Saad,
  • Emna Ammar Elhadjamor,
  • Taoufik Ladhari,
  • Mohamed Hichem Gazzah

摘要

To help detect cancer at an early stage and reduce the mortality rate, a Surface Plasmon Resonance (SPR) biosensor based on optical fiber has been proposed in this study. The SPR mechanism is created by a novel ZnO/Ag/graphene design, to accurate detection of six different cancer cells. In addition, in this article, we present new approaches for modeling SPR biosensors using artificial neural networks (ANN). Thus, a multilayer perceptron regression model (MLP) was used to predict the sensitivity values of the SPR biosensor and the effect of physical parameters on the SPR response especially the Detection Accuracy (DA) and the Quality Factor (QF). In results, the use of ZnO layer provide stronger electric field localization and enhanced plasmonic coupling. Subsequently the maximum wavelength sensitivity is achieved as 4200 nm/RIU, 5571.42 nm/RIU, 5208.33 nm/RIU, 6428.57 nm/RIU, 6928.57 nm/RIU, 7285.71 nm/RIU for the detection of cancer cells in skin (basal), cervical (HeLa), adrenal gland (PC12), blood (Jurkat), and breast (MCF-7 and MDA MB-231) respectively. Moreover, the proposed sensor achieves a high QF of 35.0483 RIU−1 along with a DA of 0.0053 nm−1. The MLP-ANN model showed good performance metrics. In addition, the optimization of various physical and geometric parameters was successfully predicted using the same approach, showing strong consistency between predicted and actual data in each case. Furthermore, our results prove the potential of the MLP-ANN model to accurately approximate the performance of the biosensor and highlight its ability to significantly simplify and accelerate the design and optimization of this type of biosensor. Thanks to the advanced results and the application of machine learning, this sensor can be used as a fast, effective and inexpensive cancer detection device.