Artificial Intelligence-Assisted Digital Immunoassay Based on a Programmable-Particle-Decoding Technique for Multitarget Ultrasensitive Detection.
Yang ZhouWeiqi ZhaoYao-Ze FengXiaohu NiuYongzhen DongYiping ChenPublished in: Analytical chemistry (2022)
The development of a multitarget ultrasensitive immunoassay is significant to fields such as medical research, clinical diagnosis, and food safety inspection. In this study, an artificial intelligence (AI)-assisted programmable-particle-decoding technique (APT)-based digital immunoassay system was developed to perform multitarget ultrasensitive detection. Multitarget was encoded by programmable polystyrene (PS) microspheres with different characteristics (particle size and number), and subsequent visible signals were recorded under an optical microscope after the immune reaction. The resultant images were further analyzed using a customized, AI-based computer vision technique to decode the intrinsic properties of polystyrene microspheres and to reveal the types and concentrations of targets. Our strategy has successfully detected multiple inflammatory markers in clinical serum and antibiotics with a broad detection range from pg/mL to μg/mL without extra signal amplification and conversion. An AI-based digital immunoassay system exhibits great potential to be used for the next generation of multitarget detection in disease screening for candidate patients.
Keyphrases
- artificial intelligence
- label free
- deep learning
- machine learning
- big data
- loop mediated isothermal amplification
- sensitive detection
- real time pcr
- gold nanoparticles
- convolutional neural network
- end stage renal disease
- molecularly imprinted
- quantum dots
- healthcare
- newly diagnosed
- chronic kidney disease
- peritoneal dialysis
- risk assessment
- single cell
- human health
- high resolution
- mass spectrometry
- patient reported outcomes
- climate change
- high speed
- optical coherence tomography