On the robustness of machine learning algorithms toward microfluidic distortions for cell classification via on-chip fluorescence microscopy

Abstract

Single-cell imaging and sorting are critical technologies in biology and clinical applications. The power of these technologies is increased when combined with microfluidics, fluorescence markers, and machine learning. However, this...

Real-time fluorescence imaging flow cytometry enabled by motion deblurring and deep learning algorithms

“…[36][37][38] Specifically, these algorithms have been applied in the analysis of low-quality cell images, such as acquiring high-resolution images from low-resolution images, 39 removing blur from out-of-focus images, 40 and identifying healthy cells from degraded images. 41 However, directly utilizing deep learning algorithms to identify cell types from motion-blur images presents challenges as it requires annotated datasets for model training. This means that the motion-blur images must be annotated before training, but direct annotation of blurred images by human experts is difficult, particularly for natural samples that include various cell types, such as blood cell samples 42 and cells at different cell cycle stages.…”

Boosting the Clinical Translation of Organ-on-a-Chip Technology

“…Indeed, significant efforts have been invested in improving the physiological relevance of OoC (e.g., new designs—modular or standardized architectures; integration of human-based materials; high throughput characteristics; simpler manipulation; improved biology; and others), obtaining key cell biology insights. The adaptation of OoC to future, or growing technologies, such as miniaturized analytical sensors, artificial intelligence and machine learning systems for automated biomarker detection, image processing and data analysis, would provide unprecedented advantages to the clinical team in decision-making [ 67 ]. Next, even though the future of OoC for clinical applications is optimistic, the challenges mentioned above, mainly related to OoC validation, standardization, throughput, and compatibility with existing analytic/imaging technologies, will need to be addressed.…”

AI-enhanced biomedical micro/nanorobots in microfluidics

“…62,63 By these approaches, the transparency, robustness, fairness and privacy of models can be enhanced. On-chip cell classification, 64,65 drug design 66 and delivery, 67 target recognition, 68 and nucleic acid amplification prediction 69 have been successfully assisted by DL models (Fig. 2).…”