Optomechatronics Design and Control for Confocal Laser Scanning Microscopy

Optomechatronics Design and Control for Confocal Laser Scanning Microscopy

Yoo, H.W.

Verhaegen, M.H.G. (promotor)
Schitter, G. (promotor)

Mechanical, Maritime and Materials Engineering

Delft Center for Systems and Control

2015-03-30

Confocal laser scanning microscopy (CLSM) is considered as one of the major advancements in microscopy in the last century and is widely accepted as a 3D fluorescence imaging tool for biological studies. For the emerging biological questions CLSM requires fast imaging to detect rapid biological processes and aberration-corrected imaging to localize the targeted biomolecule precisely through optical disturbances by specimen. In this thesis, optomechatronics design and control are discussed for improving this temporal and spatial resolution of CLSM to respond the needs in biological research. To improve temporal resolution of CLSM imaging, the scanning speed has to be improved. For galvanometer scanners as the most popular scanner of commercial CLSM, iterative learning control (ILC) is proposed to achieve a high speed, linear, and accurate bidirectional scanning control. Two stable inversion methods of zero phase shifts and phase fitting by input delays are used for designing stable ILCs enabling a wide control bandwidth. Experimental results verify the benefits of ILCs allowing a faster scanning over 2000 lines per second with high accuracy without a phase lag and a gain mismatch, achieving up to a 73 times smaller root mean square (RMS) error than a conventional feedback controller. Although the encoder measurements follow the reference signal by the developed ILC, actual beam trajectories can have errors at high scanning rates due to non-collocation sensing by the encoder. A transformation-based iterative learning control is proposed to improve the accuracy of fast beam scanning with the non-collocated galvanometer scanner. The proposed ILC is extended from the previous ILC design by adding a reference transformation filter, which is based on the transfer functions between the mirror and the encoder. An error analysis in theory shows that the proposed ILC can reduce the error of the actual mirror angle, especially for the image scanning applications. Experimental results with the proposed transformation based ILC show up to 7.5 times better beam accuracy as compared to the previous ILC. To improve spatial resolution in CLSM, the spherical aberrations induced by coverslip thickness mismatch have to be corrected. An automated adjustment of the coverslip correction collar is proposed to compensate for the spherical aberrations by means of motorization of the collar with a correction algorithms. An axial image model is derived to suppress noise of the measured axial image and to analyze of the influence of the spherical aberrations by the coverslip thickness mismatch. To search for the best correction collar adjustment, axial scans of the coverslip reflection are recorded, processed, and evaluated by correction quality measures such as the maximum intensity, sharpness, and entropy. The benefits of the proposed automated correction are demonstrated with various coverslips with biological specimens. The Imaging examples illustrate the improved resolution with sharp and accurate multicolor images of the confocal microscope. For the general aberration correction in the deep tissue imaging, an adaptive optics (AO) is developed for the commercial CLSM to verify its concept. The AO system consists of a piezoelectric deformable mirror and a Shack Hartmann wavefront sensor (SH-WFS), which measures the wavefront of the fluorescence from the specimen. The wavefront sensor is equipped with an adjustable pinhole for confocal wavefront sensing (CWFS) to confine the optical thickness of wavefront measurements. Using the adjustable pinhole, a referencing method of the SH-WFS and the evaluation of the AO correction quality, pinhole intensity ratio, are proposed. Experimental results with fluorescence beads on the coverslip and 40?m deep in a sphere cell cluster show that the developed AO system and proposed algorithms with adjustable pinhole can improve the measured full width at half maximum (FWHM). The proposed pinhole intensity ratio using the adjustable pinhole can also show the improvement of imaging quality by the proposed AO. For CWFS, a small pinhole is desirable for rejecting out-of-focus light while it can degrade the wavefront measurement qualities. A wavefront reconstruction technique is proposed to recover the degraded phase information by the finite size of pinhole. The aberration modification by the pinhole can be modeled as a 2D convolution of the pupil function in complex domain, i.e. phase and intensity of the beam. Based on the Fresnel approximation, the 2D deconvolution problem can be simplified to the 1D deconvolution, which also allows retrieval from multiple measurements by diversified pinhole sizes. With the verified model by experimental results, the simulation results of various pinhole sizes show that the distortion of the output pupil functions by the finite pinhole can be recovered by the proposed retrieval technique, reducing the RMS phase error up to 46 %. The proposed retrieval technique is evaluated for arbitrary aberrations generated from statistics of the wavefront measurements of a biological specimen. Simulation results show that about the wavefront errors level with 3 airy unit (AU) can be achieved by the recovery algorithm and the pupil measurement with 1.5 AU pinhole, allowing an accurate wavefront sensing with higher optical sectioning ability.

Confocal Laser Scanning Microscopy
Iterative Learning Control
Galvanometer Scanner
Coverslip Correction Collar
Adaptive Optics
Confocal Wavefront Sensing

https://doi.org/10.4233/uuid:07f3165a-573b-4cca-a7fc-38f5f03e37d0

2015-08-31

978-94-6203-812-7

Institutional Repository

doctoral thesis

(c) 2015 Yoo, H.W.