Cover Page
The handle http://hdl.handle.net/1887/106088 holds various files of this Leiden University dissertation.
Author: Tang, X.
Title: Computational optimisation of optical projection tomography for 3D image analysis
Computational optimisation of
Optical Projection Tomography for
3D image analysis
PROEFSCHRIFT
ter verkrijging van de graad van doctor
aan de Universiteit Leiden,
op gezag van de Rector Magnificus Prof. Mr. C.J.J.M. Stolker,
volgens besluit van het College voor Promoties,
op woensdag 10 juni 2020
klokke 16:15 uur
door
Xiaoqin Tang
Promotiecommissie
Promotor: Prof. Dr. Ir. F.J. Verbeek
Overige Leden: Prof. Dr. J.B.L. Bard University of Edinburgh University of Oxford
Prof. Dr. R.E. Poelmann Institute of Biology Leiden Prof. Dr. F.S. de Boer CWI, the Netherlands Dr. K.F.D. Rietveld
Prof. Dr. A. Plaat
ISBN: 978-94-6332-638-4
The studies described in this thesis were performed at the Leiden Institute of Advanced Computer Science (LIACS), Leiden University, Leiden, The Netherlands. The samples were mainly prepared and imaged at the Institute of Biology Leiden, The Netherlands.
The research was partially supported by the China Scholarship Council (CSC), Grant No. 201506990026
Copyright © 2020 by X. Tang
Contents
Chapter 1 ... 1
Introduction... 1
1.1 Importance of three-dimensional imaging in biomedical research ... 3
1.2 Introduction of OPT imaging system ... 5
1.2.1 Introduction of OPT imaging schema ... 5
1.2.2 Experimental OPT imaging setup ... 6
1.2.3 OPT imaging software ... 8
1.2.4 Experimental sample preparation ... 10
1.3 Computational approaches of OPT imaging ... 10
1.3.1 3D reconstruction ... 10
1) Fast reconstruction and optimisation ... 11
2) Iterative reconstruction and optimisation ... 12
1.3.2 3D segmentation of OPT reconstructions with applications to zebrafish ... 13
1.3.3 Quantification of volumetric fluorescence in zebrafish ... 15
1.4 Research questions and perspectives ... 16
1.5 Thesis structure ... 19
Chapter 2 ... 23
Fast Post-processing Pipeline for Optical Projection Tomography ... 23
2.1 Introduction ... 25
2.1.1 Research problem ... 25
2.1.2 Related work ... 25
2.2 Materials and methods ... 27
2.2.1 OPT imaging ... 27
2.2.2 OPT reconstruction software ... 28
2.2.3 Cluster computing: the LLSC ... 28
2.3 Implementation ... 29
1) Sinogram selection ... 29
2) Interest point detection... 30
3) CoR localization and alignment ... 34
2.3.2 Reconstruction and fusion ... 35
2.3.3 Parallel setting ... 36
2.4 Experiments ... 37
2.4.1 Experiments on the fast post-processing pipeline ... 37
2.4.2 Comparison of different CoR corrections on different data ... 40
2.5 Conclusions ... 42
2.6 Acknowledgement ... 43
Chapter 3 ... 45
Deblurring Images from 3D Optical Projection Tomography Using Point Spread Function Modelling ... 45
3.1 Introduction ... 47
3.1.1 Background: 3D image deconvolution ... 47
3.1.2 Related work ... 48
3.2 Materials and methods ... 49
3.2.1 Sample preparation of a single fluorescence sphere ... 49
3.2.2 PSF modelling concerning different magnifications ... 50
3.2.3 Deconvolution of 3D images in coronal plane ... 54
3.3 Experiments ... 55
3.3.1 Image comparison of deconvolution ... 55
3.3.2 Image blur measurement on slices ... 58
3.3.3 Quantitative 3D image quality improvement of deblur ... 60
3.4 Conclusions ... 61
Chapter 4 ... 63
Segmentation-driven Optimisation for Iterative Reconstruction in Optical Projection Tomography: An Exploration ... 63
4.1 Introduction ... 65
4.2 Iterative reconstruction for OPT ... 67
4.3 Parameter optimisation for iterative reconstruction ... 69
4.3.1 Framework of parameter optimisation for iterative reconstruction ... 70
4.3.2 Segmentation approach ... 71 1) Network structure ... 72 2) Network training ... 73 3) Segmentation ... 76 4.3.3 Evaluation criterion ... 76 4.4 Experiments ... 77
4.4.1 Streak artefacts and elimination ... 77
4.4.2 Parameter optimisation ... 77
1) Dataset and experimental settings ... 77
2) Iteration number and initial reconstruction ... 80
4.4.3 Comparison of segmentation performance between OSEM and FBP ... 86
4.4.4 Discussion ... 90
4.5 Conclusions ... 90
4.6 Acknowledgement ... 91
Chapter 5 ... 93
Automated Detection of Reference Structures for Fluorescent Signals in Zebrafish with a Case Study in Tumour Quantification ... 93
5.1 Introduction ... 95
5.1.1 Research questions ... 95
5.1.2 OPT as a solution for whole-mount imaging ... 96
5.1.3 Multi-channel analysis of whole-mount zebrafish ... 96
5.1.5 Structure of this chapter ... 99
5.2 Materials and methods ... 99
5.2.1 Zebrafish ... 99
5.2.2 OPT imaging and reconstruction ... 99
5.2.3 Relative quantification ... 100
5.3 Design and implementation ... 100
5.3.1 Segmentation of reference structures... 100
1) 2D Unet ... 101
2) 3D Unet ... 101
5.3.2 Learning scheme ... 101
1) Loss & Metrics ... 101
2) Optimizer & Learning rate... 103
5.4 Experiments and results ... 104
5.4.1 Detection of 3D reference structures ... 104
1) Evaluation metrics ... 104
2) Detection of 3D Body reference structure ... 105
3) Detection of 3D Eye reference structure ... 107
5.4.2 Case study in tumour ... 109
1) 2D relative quantification with manual labelling ... 110
2) 3D relative quantification with manual labelling ... 110
3) Comparisons of automated detection and manual labelling of RS for 3D quantification... 113
5.5 Conclusions and discussion ... 115
5.6 Acknowledgment ... 115
Chapter 6 ... 117
Exploration of 3D Structure Annotation and Visualization of Zebrafish Reconstructions from Optical Projection Tomography Imaging ... 117
6.1 Introduction ... 119
6.2.1 Zebrafish and OPT 3D imaging system ... 120
6.2.2 Annotation method ... 120
6.2.3 Visualization software ... 121
6.3 Experiments ... 121
6.3.1 Manual annotation and visualization ... 121
1) Amira ... 121
2) TDR-3Dbase and MeshLab ... 126
6.3.2 Automated 3D annotation of 5 dpf zebrafish ... 127
6.4 Conclusions and discussion ... 129
6.5 Acknowledgement ... 131
Chapter 7 ... 133
Conclusions & Discussion ... 133
7.1 Main contributions ... 135
7.2 Achievements of research presented in this thesis. ... 138
7.3 Limitations and possible solutions ... 138
