Year: 2026
Type: article
Abstract: A generative approach for the automatic generation of surface plasmon resonance (SPR) image-responses is presented. The modern development of SPR sensors is driven by machine learning (ML) interventions that require data related to the sensors’ operation, with the intention of anticipating behaviors, predicting responses, and accelerating the design, optimization, and utilization of this type of sensor. The cost to produce SPR data in terms of time, material consumption, and experimental complexity associated with the absence of a public dataset of SPR images, makes it difficult to develop effective intelligent solutions for SPR sensors. The methodological proposition is based on a generative adversarial network (GAN) model to generate synthetic SPR images, reducing experimental data dependence, accelerating research workflows, and supporting intelligent biosensor advancement through complementary synthetic data. Manual creation of Fresnel images based on Kretschmann-configuration experimental SPR images of aqueous solutions was employed as training data. The results of synthetic images were quantitatively and qualitatively evaluated, in particular for a deep-convolutional GAN (DCGAN), indicating the feasibility of the model to generate two-dimensional SPR images. The latent space generated by DCGAN was explored in different scenarios of image generation, and the utility of the generative approach is attested in the substance classification task, leading to a model improvement of the main performance metrics.
Source: Scientific Reports
Language: English
Subfield: Atomic and Molecular Physics, and Optics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7233608694
FWCI:
Cites:
Cited by:
Related to:
