Adaptive optics techniques for multi-photon microscopy

Research context and potential impact

Optical microscopes have enabled researchers to explore the function of living tissue with unprecedented resolution and frame rates. However, significant dvancements can still be achieved. As research increasingly focuses on deep imaging in live specimens, the distorting effect of light traveling through tissue has become more critical (1).

When light passes through biological material, inhomogeneities in refractive index cause distortions in the optical wavefront. These aberrations have a major negative impact on imaging performance and are challenging to address as they vary from sample to sample.

One highly effective solution to this limitation is Adaptive Optics (AO), a set of techniques that employs a dynamic element to correct the wavefront aberrations in real-time (2). Our research will focus on developing novel control algorithms for AO, specifically the implementation of machine learning (ML) methods for the correction of aberrations in multiphoton imaging systems, a type of microscope that is very sensitive to wavefront distortions.

The impact of our work ould be substantial, particularly in fields like neuroscience and cancer research, where deep tissue imaging is essential for understanding in vivo biological rocesses. Aims and Objectives The following objectives have been outlined for our research:

1.Enhance the current machine learning (ML) methods employed in the lab by exploring new data encoding schemes and alternative ML architectures 2.Investigate alternative mathematical descriptions of optical aberrations, such as wavelet analysis (3)

3.Develop and implement a robust, general Adaptive Optics (AO) control scheme that can be seamlessly integrated into any existing system Novelty of the Research Methodology

Our group was the first to successfully apply ML control to AO in real imaging scenarios, including three-photon live imaging of mouse brain activity. Leveraging our in-depth understanding of the image generation process, we have been able to develop novel data processing and ML pipelines that significantly reduce data size and computational time compared to general architectures(4).

We aim to build on this success by further exploring these approaches, utilizing simulated microscope models and the extensive array of physical systems available in our lab. Alignment with EPSRC Strategy

This project aligns with the EPSRC's Optical Devices and Subsystems research area, with significant relevance to the themes of Engineering and Healthcare Technologies. Our development of a general adaptive optics (AO) system has the potential to become the leading solution for deep tissue imaging, capable of integration into almost any existing microscope.

This would facilitate more accurate and advanced research, benefiting both the UK and the international scientific community. We aim to transform this technology into a commercially viable product for healthcare and medical research, in full alignment with the EPSRC's mission to foster world-class innovation and deliver meaningful impact.

References 1. Yoon S, Cheon SY, Park S, Lee D, Lee Y, Han S, et al. Recent advances in optical imaging through deep tissue: imaging probes and techniques. Vol. 26, Biomaterials Research. BioMed Central Ltd; 2022. 2. Hampson KM, Turcotte R, Miller DT, Kurokawa K, Males JR, Ji N, et al. Adaptive optics for high-resolution imaging.

Vol. 1, Nature Reviews Methods Primers. Springer Nature; 2021. 3. Antonello J, Barbotin A, Chong EZ, Rittscher J, Booth MJ. Multi-scale sensorless adaptive optics: application to stimulated emission depletion microscopy. Opt Express. 2020 May 25;28(11):16749. 4. Hu Q, Hailstone M, Wang J, Wincott M, Stoychev D, Atilgan H, et al. Universal adaptive optics for microscopy

through embedded neural network control. Light Sci Appl. 2023 Dec 1;12(1).