A Fluorescent Biosensor for Sensitive Detection of Salmonella Typhimurium Using Low-Gradient Magnetic Field and Deep Learning via Faster Region-Based Convolutional Neural Network

Abstract

In this study, a fluorescent biosensor was developed for the sensitive detection of Salmonella typhimurium using a low-gradient magnetic field and deep learning via faster region-based convolutional neural networks (R-CNN) to recognize the fluorescent spots on the bacterial cells. First, magnetic nanobeads (MNBs) coated with capture antibodies were used to separate target bacteria from the sample background, resulting in the formation of magnetic bacteria. Then, fluorescein isothiocyanate fluorescent microspheres (FITC-FMs) modified with detection antibodies were used to label the magnetic bacteria, resulting in the formation of fluorescent bacteria. After the fluorescent bacteria were attracted against the bottom of an ELISA well using a low-gradient magnetic field, resulting in the conversion from a three-dimensional (spatial) distribution of the fluorescent bacteria to a two-dimensional (planar) distribution, the images of the fluorescent bacteria were finally collected using a high-resolution fluorescence microscope and processed using the faster R-CNN algorithm to calculate the number of the fluorescent spots for the determination of target bacteria. Under the optimal conditions, this biosensor was able to quantitatively detect Salmonella typhimurium from 6.9 × 101 to 1.1 × 103 CFU/mL within 2.5 h with the lower detection limit of 55 CFU/mL. The fluorescent biosensor has the potential to simultaneously detect multiple types of foodborne bacteria using MNBs coated with their capture antibodies and different fluorescent microspheres modified with their detection antibodies.

Development of a Dual Mode UCNPs-MB Biosensor in Combination with PCR for Sensitive Detection of Salmonella

“…Compared with the results of gel electrophoresis and qPCR methods (with LOD of 8.1 × 10 3 and 8.1 × 10 2 CFU/mL, respectively, in Figure S3 ), the proposed dual-mode PCR biosensor showed superior performance. Furthermore, it exhibited comparable or even higher sensitivity, a relatively wide detection range, and a short detection time compared with the analytical performance of other reported methods ( Table 1 ) [ 37 , 38 , 39 , 40 , 41 , 42 ], confirming its ideal performance.…”

A Framework for Biosensors Assisted by Multiphoton Effects and Machine Learning

“…NNs algorithms fulfill the function of processing the images obtained from fluorescent bacteria. NNs processing manages to calculate the amount of fluorescent points faster to determine the target bacteria [ 156 ].…”

A Systematic Literature Review of Deep Learning-Based Detection and Classification Methods for Bacterial Colonies

“…In [24], the authors developed a biosensor to detect foodborne bacteria, specifically 'Salmonella Typhimurium' responsible for numerous infectious diseases. They aimed to detect fluorescence spots in microscopic images using Regionbased R-CNN.…”

“…In [44], the authors used supervised learning and annotated microscopic images obtained from waste water. The authors in [24] used supervised learning for identification of food-borne bacteria, specifically 'Salmonella Typhimurium'. The authors in [53] make use of an annotated dataset of microscopic images to identify bacteria from images.…”

“…The image size is 10x30x30. The authors in [24] applied a number of pre-processing steps to their datasets. For instance, a magnetic field was used to convert 3D signals into 2D data.…”