JERUSALEM, Jan. 22 (Xinhua) -- Israeli researchers have improved the AI method that virtually stains living cells to help scientists better "peek" inside the cell and observe the movement in real-time, Ben-Gurion University of the Negev said in a statement on Thursday.
For the past century, scientists have been using fluorescent staining to "turn on flashlights" inside a living cell to see where each component is located. However, the method may damage cells or disrupt their natural behavior.
In recent years, virtual staining has emerged as artificial intelligence can learn to "translate" microscope images without external dyes into detailed, colorful fluorescent images. However, the AI method can struggle when they analyze isolated elements without a broader context.
In the new study published in Nature Methods, researchers from Ben-Gurion University trained the AI algorithm to incorporate contextual cues like cell shape, neighboring cells, and position within a colony, rather than learning only from the cell's image. Using the broader context allowed the AI to generalize better and accurately identify rare events, such as cell division, that typically confuse other systems.
In future studies, the researchers plan to broaden the notion of context to include factors like cell type, imaging technology, disease state, and drug exposure.
Their ultimate goal is to create a comprehensive foundation model of cells that can interpret complex biological information across microscopes and cell types, revealing detailed insights into living systems without altering them. ■
