It is known that in biosensing applications, specific analytes’ characterization is done through nano and micro-scale particles. These analytes can be viruses, proteins, and bacteria. The biosensors utilize surface chemistry coated particles so that they form clusters of target analyte by sticking to them. The size of the clusters is directly proportional to the concentration of target analyte. This method helps us in analyzing the presence and concentration of target analyte in the sample.
Presently, all the methods that perform this analysis are less as they either depend on a bulky and expensive microscope or have the capability of just rough readout. This drawback limits the application of these methods in biosensing.
The Solution: Artificial Intelligence Application
To eliminate the drawbacks posed by existing methods, an automated and quick biosensing way is derived by UCLA researchers. This method uses holography and machine learning technology for carrying out the process.
The micro-particles and clusters of particles present in the sample are 3D captured as holograms. This is done in one go as the area is 20 mm2 or larger, which is approximately 10-fold greater than the area captured in the optical microscope.
After this, a deep neural network is utilized to process the captured 3D holograms. These holograms are remade into cluster forms – almost similar to results attained with a scanning microscope. However, this process is a lot faster than the standard process.
Additionally, at the time this process is being carried out, the clusters of particles formed are counted with the accuracy and efficiency equivalent to the laboratory-grade microscope.
The Test Sample
To prove the efficiency of the method, UCLA researchers showed an application of this deep machine learning biosensing method by detecting HSV (Herpes Simplex Virus). This sample run acquired detection of 5 viruses every micro-liter, which level of sensitivity is clinically relevant as far as HSV is concerned.
The biosensing application of artificial intelligence and deep learning systems can make biosensors easily available. This is a convenient, quick, and highly effective alternative to the traditional system, which can observe wide-scale use in many nations across the globe.