Proceedings Article | 2 March 2022
KEYWORDS: Cell death, Microscopy, Cancer, Video, Phase contrast, Neural networks, Luminescence, Hydrogen, Breast cancer, Scattering
Chemotherapy is one of the most common anti-cancer treatments, that targets rapidly dividing cells by inducing DNA damage and inhibition of mitosis. Most chemotherapeutic drugs are known to induce apoptosis, which is programmed cell death. Therefore, monitoring cell death mechanism, in addition to its viability, is important for understanding the efficacy of treatment, and is particularly important during drug screening. Here we present an automated label-free method of testing the efficacy of chemotherapeutic drugs by identification of apoptotic cell death, based on the scattering signature of cancer cells, using dark field microscopy. Breast cancer cells (BT-20) were treated with different chemo-drugs and simultaneously imaged during the drug incubation step. A neural network was trained to identify the cells that remain alive, as well as distinguish between the cells undergoing apoptosis and necrosis, the two most common cell death mechanisms. The network identifies the cell death mechanism, based on the temporal changes in morphological properties of the cells, e.g. volume shrinkage, blebbing, membrane damage etc. Our results show that the network trained using a specific chemo-drug can then be used for identifying the cell death processes induced by other types of chemo-drugs, with over <95% accuracy, confirmed using western blot assay. This automated technique which can predict the cell death mechanism and viability in real time, during drug incubation, eliminates the additional steps, such as staining or adding conjugates, required for fluorescence imaging and western blot respectively, thereby making it user friendly, cost-effective and high throughput.
