Intravital Microscopy

AI Noise Canceling for High Quality In Vivo Imaging with Live Animals

AI Noise-Canceling Technology employs artificial intelligence algorithms to reduce or eliminate noise from real-time images and videos captured by IVM. This technology allows for a more precise understanding of the images by delivering smoother, noise-free backgrounds. For example, it enables differentiation of nanoparticle drugs even amidst noise, thereby enhancing comprehension of in vivo interactions involving the target compound or sample.

Raw

0.2625 s

0.2650 s

2.2300 s

2.2325 s

Principle of SUPPORT

Recent advancements in intravital and calcium imaging hasenabled recording of the population activity of neurons at an unprecedented throughput, which opens up the possibility of a system-level understanding of neuronal circuits. It is essential to record the activities with high temporal precision to investigate causality within neuronal activities.
Increasing the temporal resolution in functional imaging data inevitably results in a decrease in the SNR. The decrease in SNR not only hinders the accurate detection of the neurons’ location but also compromises the timing precision of the detected temporal events, which nullifies the increase in temporal resolution. Fortunately, all functional imaging data have high inherent redundancy in the sense that each frame in a dataset shares a high level of similarity with other frames apart from noise, which offers an opportunity to denoise or distinguish the signal from the noise in the data.
IVIM Technology adopted SUPPORT (statistically unbiased prediction using spatiotemporal information in imaging data), a self-supervised denoising method for functional imaging data that is robust to fast dynamics in the scene compared to the imaging speed. SUPPORT is based on the insight that a pixel value in functional imaging data is highly dependent on its spatiotemporal neighboring pixels, even when its temporally adjacent frames alone fail to provide useful information for statistical prediction. By learning and using the spatiotemporal dependence among the pixels, SUPPORT can accurately remove Poisson-Gaussian noise in intravital imaging data which cannot be inferred from information in other frames.