박인우 Email & Phone Number

My research field is related to optical imaging. Based on fundamental optics, I have studied how to build optical imaging, predict measurements of the imaging system, and optimize the imaging system based on numerical simulation. One of my research title is "Label-free Multi-contrast imaging system and deep learning technique". The most favored optical imaging methods in biology and clinic are Fluorescence and H&E staining. But, they have room to be enhanced in terms of time and labor cost during sample preparation and imaging process. Furthermore, fluorescence microscopy needs efforts to avoid photobleaching due to chemical alteration of the indicator dye and toxicity. As an alternative, label-free imaging combined with deep learning technique, Generative Adversarial Newtworks (GAN), is proposed. GAN generates fluorescence or H&E staining image by converting label-free image acquired in relatively very low consumption of time and labor coast for sample. Well-trained GAN was based on well-registered two images data sets, (fluorescence, label-free) or (H&E, label-free) acquired in two different systems. We demonstrate fast and virtual fluorescence imaging from a label free image based in multi-contrast label-free imaging and deep learning technique, Generative Adversarial Networks (GAN). Our imaging system offers fluorescence image and label-free image simultaneously. Fluorescence and label-free image set acquired in single body platform was used to train a GAN without image registration procedure. We confirmed that our trained GAN well performs virtual fluorescence labeling with highsensitivity and accuracy.