Project Description

Microscopy is an important tool for studying the structure and properties of materials, and for identifying and characterizing small objects or organisms. However, the resolution we can achieve from optical microscopes is limited by diffraction. However recent years, many new optical techniques have come up with higher resolutions that break this diffraction limit. These microscopy techniques are called super-resolution microscopy or nanoscopy. There are several super-resolution microscopy techniques such as Structured Illumination microscopy (SIM), Stimulated Emission Depletion (STED) microscopy, Single molecule localization microscopy (SMLM) etc.

Structured Illumination Microscopy (SIM) even though gives a modest resolution enhancement, it stands out from others in many ways due to its practical usages. The biggest limitation of SIM is that even though it is faster than other super resolution microscopy technique, it is still limited to two times resolution enhancement. SIM enables super-resolution imaging by encoding high-frequency spatial information through patterned light. While traditional Fourier based reconstruction methods are prone to artifacts under suboptimal conditions, recent deep learning and AI approaches have shown promise in obtaining fast and accurate super-resolved reconstruction, but often require large training datasets and hence lack adaptability across different imaging setups.

The objective of the present project will be to construct a transmission structured illumination microscopy setup and develop novel AI/ML techniques specifically applied to the SIM captured images for obtaining fast and accurate super-resolved reconstructions.