Fluorescence Image Visualization using Multi-Channel Minimax Optimization (MCMO)

Fluorescence microscope data analysis for liquid biopsy needs to be evidenced by object visualization on the display. This visualization heavily depends on the operator's tuning of the display settings, which is typically achieved through conventional linear stretching or windowing/leveling techniques. However, manual tuning can introduce variability and bias, leading to reduced throughput of analysis verifications. We propose a novel method for enhancing multi-channel fluorescence images by optimizing the visualization of assay-related objects. This problem is formulated as a variational minimax optimization of the histogram-equalizing energy with data distortion regularization. Our method seamlessly merges the resolved, enhanced objects of interest with the background, resulting in high performance and overall visual coherence.