Framework for signal-to-noise ratio estimation in fluorescence imaging systems

Fluorescence imaging enables researchers and clinicians to observe molecular activity that would otherwise remain unseen, playing a key role in medical diagnostics and biological studies. One major challenge is detecting extremely faint signals against substantial background interference. To evaluate how well imaging systems manage this, the signal-to-noise ratio (SNR) is commonly used. However, a standardized method tailored to fluorescence systems has not yet been established.

This thesis presents a more dependable and systematic technique for assessing SNR in fluorescence imaging setups. The approach involves capturing a series of identical images and using statistical methods to more effectively isolate meaningful signal data from random noise. It adapts concepts from EMVA 1288, a widely recognized standard in machine vision, and tailors them for fluorescence applications—where such methodologies have been underutilized.

We applied this approach using the fluorescent dye indocyanine green (ICG), evaluating performance under varying lighting conditions, sample distances, and dye concentrations. Results demonstrate that this method successfully identifies optimal imaging conditions and improves image clarity and accuracy.

By refining how SNR is measured, this work supports the development of more reliable fluorescence imaging tools, which is especially beneficial in fields like cancer diagnostics, pharmaceutical testing, and biomedical research where accurate images can impact critical decisions.