Enhancing quantification of PET-CT without MRI : development and validation of computational analysis = MRI 없이 PET-CT 를 이용한 PET 정량 분석의 개발과 검증

다국어 초록 (Multilingual Abstract)

Amyloid and tau positron emission tomography (PET) images are essential tools for diagnosing Alzheimer's disease. While CT offers anatomical information, it has lower spatial resolution than MRI. Traditional PET quantitative analysis requires both PET and MRI images. However, PET-CT scanners provide both PET and CT images in a single scan. This led us to develop quantitative analysis methods for amyloid PET and Tau PET using PET-CT scans, streamlining the process and reducing the need for multiple imaging modalities.
In Chapter 1, I aimed to develop a CT-based regional method for quantifying amyloid-beta (Aβ) uptake using FBB and FMM PET ligands. Both MRI-based and CT-based methods correlated well between FBB and FMM uptake, both globally and regionally. This suggests the feasibility of converting regional FBB or FMM PET values into a common scale without MRI scans.
Chapter 2 focused on exploring the relationship between CT-based regional dcCL scales and Aβ pathological findings. The results showed significant correlations between dcCL scales and neuritic plaque burdens, globally and regionally. Regional dcCL thresholds effectively differentiated memory and cognitive functions between groups, aiding clinicians in biomarker-guided diagnosis.
In Chapter 3, I introduced a novel CT image segmentation approach using deep learning techniques, which allowed for the quantification of Tau PET images into Braak staging system. My deep learning-based quantitative analysis demonstrated high accuracy, making it a promising tool for assessing Tau PET, particularly with varying regional uptake patterns.
In conclusion, I have developed and validated CT-based amyloid and tau PET quantitative analysis methods, and these methods might contribute to the clinically valuable diagnosis.

국문 초록 (Abstract)

아밀로이드와 타우 양전자 방출 단층촬영 (PET) 이미지는 알츠하이머병 진단에 필수적인 분자 진단 도구입니다. CT는 해부학 정보를 제공하지만 MRI에 비해 공간 해상도가 낮습니다. 전통적인 정량분석에는 PET 및 MRI 이미지 모두 필요하지만, PET-CT 스캐너는 PET 및 CT 이미지를 단일 스캔으로 제공하는 장점이 있습니다. 이로써 나는 PET-CT 스캔을 활용하여 아밀로이드PET와 타우PET의 정량분석 방법을 개발하여 복수의 이미징 모달리티 필요성을 줄이고 프로세스를 간소화하였습니다.
제1 장에서는 FBB 및 FMM PET 방사성의약품을 사용하여 아밀로이드-베타 (Aβ) 흡수를 정량화하는 CT 기반 영역별 정량 분석 방법을 개발하고 검증하는 것이 목표입니다. MRI 기반 및 CT 기반 정량 분석 방법은 FBB 및 FMM 흡수 사이에 전체적으로 및 영역별로 강한 상관 관계를 나타내어 MRI 스캔 없이 영역별 아밀로이드 PET 값을 공통 표준화 스케일로 변환 가능성을 시사합니다.
제2 장에서는 CT 기반 영역별 dcCL 스케일과 Aβ 병리적 결과 간의 상관 관계를 평가하고 병리학적 결과 임계값을 활용하여 임상 유용성을 검증하였습니다. 결과는 전체적으로 및 영역별로 dcCL 스케일과 신경플라크 부하 간에 유의한 상관 관계가 있음을 보여주었습니다. 영역별 dcCL 임계치의 적용은 그룹 간의 기억 및 인지 기능 차이를 명확히 하며 바이오마커 기반 진단에 도움을 줍니다.
제3 장에서는 딥러닝 기술을 활용하여 CT 이미지를 세분화하는 혁신적인 방법을 소개하고 타우PET 영상을 Braak 단계로 정량화하였습니다. 딥러닝 기반의 정량분석은 높은 정확도를 달성하였으며, 다양한 영역별 섭취 패턴을 가진 타우 PET을 평가하는 데 유용한 도구로 입증되었습니다.
마무리로, 나는 CT를 기반으로 한 아밀로이드 및 타우 PET 정량분석 방법을 개발하고 검증했으며, 이러한 방법들은 임상적으로 실용적인 진단에 기여할 수 있을 것입니다.

목차 (Table of Contents)

• Chapter 1. Development and clinical validation of CT-based regional modified Centiloid method for amyloid PET 1
• 1.1 Introduction 1
• 1.2 Materials and methods 4
• 1.2.1 Participants 4
• 1.2.2 MRI data acquisition 5
• 1.2.3 Aβ PET-CT data acquisition 5
• 1.2.4 Regional visual assessment and rdcCL scales 7
• 1.2.5 Development of MRI-based global and regional CTX VOIs 7
• 1.2.6 Development of CT-based global and regional CTX VOIs 9
• 1.2.7 Development of MRI and CT-based rdcCL 10
• 1.2.8 Validation of the clinical efficacy of CT-based rdcCLs in the independent cohort 14
• 1.2.9 Statistical analysis 15
• 1.3 Results 17
• 1.3.1 Demographics of the participants 17
• 1.3.2 Visual assessment and rdcCL scales 20
• 1.3.3 CT-based rdcCLs 22
• 1.3.4 Reliability and precision in the MRI-based and CT-based rdcCLs 23
• 1.3.5 Validation of CT-based rdcCLs in independent participants 27
• 1.4 Discussion 30
• 1.5 Conclusions 34
• 1.6 Supplementary information 35
• Chapter 2. Clinical and pathological validation of CTbased regional harmonization methods of amyloid PET 57
• 2.1 Introduction 57
• 2.2 Materials and methods 59
• 2.2.1 Participants 59
• 2.2.2 A PET-CT imaging 62
• 2.2.3 Regional CT-based dcCL scales 62
• 2.2.4 Neuropathological assessment measurements 63
• 2.2.5 Validation of the clinical utility of CT-based dcCLs in an independent cohort 65
• 2.2.6 Statistical analysis 66
• 2.3 Results 68
• 2.3.1 Demographics of the participants 68
• 2.3.2 Relationships between regional dcCLs and Aβ plaques in each region 71
• 2.3.3 Optimal threshold values of regional dcCLs through Aβ pathological scores 74
• 2.3.4 Clinical utility of CT-based regional dcCLs in the independent cohort 76
• 2.4 Discussion 82
• 2.5 Conclusions 87
• 2.6 Supplementary information 88
• Chapter 3. Deep learning-based quantitative analysis of Tau PET-CT imaging for Alzheimer's disease 90
• 3.1 Introduction 90
• 3.2 Materials and methods 93
• 3.2.1 Participants 93
• 3.2.2 MRI data acquisition 95
• 3.2.3 PET-CT data acquisition 95
• 3.2.4 Data preparation and image processing 96
• 3.2.5 Model architectures 97
• 3.2.6 Cascaded networks for Tau PET quantification using deep learning predictive models 100
• 3.2.7 Deep learning model evaluation for performance 104
• 3.2.8 Statistical analysis 104
• 3.3 Results 105
• 3.3.1 Demographics of the participants 105
• 3.3.2 Evaluation of segmentation performance 106
• 3.3.3 Evaluation of Tau PET quantification 110
• 3.4 Discussion 115
• 3.5 Conclusions 119
• References 121
• Korean Abstract 136

참고문헌 (Reference)

1. To stage, or not to stage, H. Braak, Del Tredici, K. and, 61: p. 1022, , 2020

2. Tau in Alzheimer's disease, E. Mandelkow, Mandelkow, E. M. and, Trends in cell biology 8(11): p. 425427, , 1998

3. Seoul neuropsychological screening, D. L. Na and, S. Hahn, Kang, Y., Incheon: Human brain research & consulting co, , 2003

4. Betaamyloid imaging with florbetaben, Sabri, O. et al., 3(1): p. 1326, , 2015

5. PETonly 18FAV1451 tau quantification, Bourgeat, P., et al, in 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), , 2017

6. PET staging of amyloidosis using striatum, Hanseeuw, B. J. et al., 14(10): p. 12811292, , 2018

7. Tau imaging in neurodegenerative diseases, Dani, M., D. Brooks and, P. Edison, 43: p. 11391150, , 2016

8. Identity mappings in deep residual networks, He, K., et al, in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands Proceedings Part IV 14. 2016. Springer, , 2016

9. Comparing images using the Hausdorff distance, W. J. Rucklidge, G. A. Klanderman and, Huttenlocher, D. P., 15(9): p. 850863, , 1993

10. In vivo staging of regional amyloid deposition, Grothe, M. J. et al., 89(20): p. 20312038, , 2017

11. Tau PET imaging: present and future directions, SaintAubert, L. and, L. Lemoine, 12, , 2017

12. 3D segmentation in the clinic: A grand challenge, M. Styner, Van Ginneken, B., T. Heimann and, in MICCAI workshop on 3D segmentation in the clinic: a grand challenge, , 2007

13. Tau imaging: early progress and future directions, Villemagne, V. L. et al., 14(1): p. 114124, , 2015

14. Neuropathological alterations in Alzheimer disease, SerranoPozo, A. et al., 1(1): p. a006189, , 2011

15. Automatic 3D modelbased neuroanatomical segmentation, Collins, D. L. et al., 3(3): p. 190208, , 1995

16. Neuropathological stageing of Alzheimerrelated changes, E. Braak, Braak, H. and, 82(4): p. 239259, , 1991

17. Agespecific CT and MRI templates for spatial normalization, Rorden, C. et al., 61(4): p. 95765, , 2012

18. FFlorbetaben PET betaamyloid binding expressed in Centiloids, Rowe, C. C. et al., 44(12): p. 20532059, , 2017

19. Hybrid cascaded neural network for liver lesion segmentation, Y. Hong, Dey, R. and, in 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI) IEEE, , 2020

20. The role of tau in Alzheimer's disease and related disorders, D. BagliettoVargas and, F. M. LaFerla, Medeiros, R., 17(5): p. 51424, , 2011

21. Measures of the amount of ecologic association between species, Dice, L. R., 26(3): p. 297302, , 1945

22. Monai: An opensource framework for deep learning in healthcare, Cardoso, M. J. et al., arXiv preprint arXiv:2211.02701, , 2022

23. Mild cognitive impairment: clinical characterization and outcome, Petersen, R. C. et al., 56(3): p. 3038, , 1999

24. 3D MRI brain tumor segmentation using autoencoder regularization in, Myronenko, A, Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries: 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Granada, Spain Revised Selected Papers, Part II 4 Springer, , 2018

25. Quantification of amyloid beta and tau PET without a structural MRI, Landau, S. M. et al., 2023 19(2): p. 444455, , 2023

26. 84 UNet: deep learning for cell counting, detection, and morphometry, Falk, T., et al., 16(1): p. 6770, , 2019

27. Tau protein and the neurofibrillary pathology of Alzheimer's disease, Goedert, M., Trends in neurosciences 16(11): p. 460465, , 1993

28. Amyloid involvement in subcortical regions predicts cognitive decline, Cho, S. H., et al., 45(13): p. 2368 2376, , 2018

29. Staging betaAmyloid Pathology With Amyloid Positron Emission Tomography, Mattsson, N. et al., 76(11): p. 1319 1329, , 2019

30. Fully Automated Segmentation of Head CT Neuroanatomy Using Deep Learning, Cai, J. C. et al., 2(5): p. e190183, , 2020

31. Time between milestone events in the Alzheimer's disease amyloid cascade, Insel, P. S., et al., 227: p. 117676, , 2021

32. 85 FastSurfer A fast and accurate deep learning based neuroimaging pipeline, Henschel, L., et al., 219: p. 117012, , 2020

33. Tau and neuroinflammation: What impact for Alzheimer's Disease and Tauopathies?, L. Buée and, D. Blum, Laurent, C., 41(1): p. 2133, , 2018

34. Considerations and code for partial volume correcting [(18)F]AV1451 tau PET data, W. J. Jagust, Baker, S. L., A. Maass and, 15: p. 648657, , 2017

35. Staging tau pathology with tau PET in Alzheimer’s disease: a longitudinal study, Chen, S.D. et al., 11(1): p. 483, , 2021

36. The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET, Klunk, W. E. et al., 11(1): p. 115. e14, , 2015

37. Family of boundary overlap metrics for the evaluation of medical image segmentation, I. Voiculescu, Yeghiazaryan, V. and, 5(1): p. 015006015006, , 2018

38. Assessing the Risks Associated with MRI in Patients with a Pacemaker or Defibrillator, Russo, R. J., et al., 376(8): p. 755 764, , 2017

39. Clinical significance of focal ssamyloid deposition measured by (18)Fflutemetamol PET, Kim, S. E. et al., 12(1): p. 6, , 2020

40. Comparison of multiple tauPET measures as biomarkers in aging and Alzheimer's disease, Maass, A. et al., 157: p. 448463, , 2017

41. White matter hyperintensities are associated with amyloid burden in APOE4 noncarriers, Noh, Y. et al., 40(4): p. 877 86, , 2014

42. nnUNet: a selfconfiguring method for deep learningbased biomedical image segmentation, Isensee, F., et al., 18(2): p. 203211, , 2021

43. A Nomogram for Predicting Amyloid PET Positivity in Amnestic Mild Cognitive Impairment, Kim, S. E. et al., 66(2): p. 681691, , 2018

44. Cascaded volumetric convolutional network for kidney tumor segmentation from CT volumes, Zhang, Y. et al., arXiv preprint arXiv:1910.02235, , 2019

45. Statistical methods for assessing agreement between two methods of clinical measurement, Bland, J. M. and, D. G. Altman, 1(8476): p. 30710, , 1986

46. Statistical methods for assessing agreement between two methods of clinical measurement, D. Altman, Bland, J. M. and, The lancet 327(8476): p. 307310, , 1986

47. 18F T807, a novel tau positron emission tomography imaging agent for Alzheimer's disease, Xia, C.F. et al., 9(6): p. 666676, , 2013

48. SynthSeg: Segmentation of brain MRI scans of any contrast and resolution without retraining, Billot, B. et al., 2023 86: p. 102789, , 2023

49. Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space, Collins, D. L. et al., 18(2): p. 192205, , 1994

50. Comparison of metrics for the evaluation of medical segmentations using prostate MRI dataset, Nai, Y.H. et al., 134: p. 104497, , 2021

51. Florbetaben PET imaging to detect amyloid beta plaques in Alzheimer's disease: phase 3 study, Sabri, O. et al., 11(8): p. 96474, , 2015

52. 108 Towards cascaded VNet for automatic accurate kidney segmentation from abdominal CT images, Luo, X., et al, in Medical Imaging 2021: Image Processing SPIE, , 2021

53. A new Centiloid method for (18)Fflorbetaben and (18)Fflutemetamol PET without conversion to PiB, Cho, S. H. et al., 47(8): p. 19381948, , 2020

54. Postmortem correlates of in vivo PiBPET amyloid imaging in a typical case of Alzheimer's disease, Ikonomovic, M. D. et al., 131(Pt 6): p. 163045, , 2008

55. Automatic pipeline analysis of 3D MRI data for clinical trials: application to multiple sclerosis, A. C. Evans, R. Forghani and, Zijdenbos, A. P., 21(10): p. 128091, , 2002

56. Relative impact of amyloidbeta, lacunes, and downstream imaging markers on cognitive trajectories, Kim, H. J. et al., 139(Pt 9): p. 251627, , 2016

57. Standardized Expression of 18FNAV4694 and 11CPiB betaAmyloid PET Results with the Centiloid Scale, Rowe, C. C. et al., 57(8): p. 12337, , 2016

58. Postmortem histopathology underlying betaamyloid PET imaging following flutemetamol F 18 injection, Ikonomovic, M. D. et al., 4(1): p. 130, , 2016

59. Spatial Normalization of (18)FFlutemetamol PET Images Using an Adaptive PrincipalComponent Template, Lilja, J. et al., 60(2): p. 285291, , 2019

60. Deep learning from MRIderived labels enables automatic brain tissue classification on human brain CT., Srikrishna, M., et al., 244: p. 118606, , 2021

61. Phase 3 trial of flutemetamol labeled with radioactive fluorine 18 imaging and neuritic plaque density, Curtis, C., et al., 72(3): p. 28794, , 2015

62. Detection of Striatal Amyloid Plaques with [18F]flutemetamol: Validation with Postmortem Histopathology, Beach, T. G. et al., 52(3): p. 86373, , 2016

63. Alzheimer's neurofibrillary pathology and the spectrum of cognitive function: findings from the Nun Study, D. A. Snowdon and, Riley, K. P., W. R. Markesbery, 51(5): p. 567577, , 2002

64. Deep learningbased automated segmentation of eight brain anatomical regions using head CT images in PET/CT, Wang, T., et al., 22(1): p. 99, , 2022

65. Automated PETonly quantification of amyloid deposition with adaptive template and empirically predefined ROI, Akamatsu, G., et al., 61(15): p. 576880, , 2016

66. Clinical significance of amyloid beta positivity in patients with probable cerebral amyloid angiopathy markers, Jang, H. et al., 46(6): p. 12871298, , 2019

67. Correlation of Alzheimer Disease Neuropathologic Staging with Amyloid and Tau Scintigraphic Imaging Biomarkers, Koychev, I., N. Friedman, M. Hofer and, 61(10): p. 14131418, , 2020

68. Automatic liver and tumor segmentation of CT and MRI volumes using cascaded fully convolutional neural networks, Christ, P. F., et al., arXiv preprint arXiv:1702.05970, , 2017

69. Validating novel tau positron emission tomography tracer [F‐18]‐AV‐1451 (T807) on postmortem brain tissue, Marquié, M. et al., 78(5): p. 787800, , 2015

70. Centiloid scaling for quantification of brain amyloid with [(18)F]flutemetamol using multiple processing methods, Battle, M. R. et al., 8(1): p. 107, , 2018

71. Staging of Alzheimer diseaseassociated neurofibrillary pathology using paraffin sections and immunocytochemistry, Braak, H. et al., 112(4): p. 389404, , 2006

72. Standardization of amyloid quantitation with florbetapir standardized uptake value ratios to the Centiloid scale, Navitsky, M. et al., 14(12): p. 15651571, , 2018

73. In vivo staging of regional amyloid progression in healthy middleaged to older people at risk of Alzheimer's disease, Levin, F. et al., 13(1): p. 178, , 2021

74. Earliest accumulation of betaamyloid occurs within the defaultmode network and concurrently affects brain connectivity, Palmqvist, S. et al., 8(1): p. 1214, , 2017

75. Regional profiles of the candidate tau PET ligand 18FAV1451 recapitulate key features of Braak histopathological stages, Schwarz, A. J. et al., 139(Pt 5): p. 153950, , 2016

76. Regional Distribution, Asymmetry, and Clinical Correlates of Tau Uptake on [18F]AV1451 PET in Atypical Alzheimer's Disease, Tetzloff, K. A. et al., 62(4): p. 17131724, , 2018

77. Interlaboratory comparison of neuropathological assessments of betaamyloid protein: a study of the BrainNet Europe consortium, Alafuzoff, I. et al., 115(5): p. 53346, , 2008

78. Regional correlations between [(11)C]PIB PET and postmortem burden of amyloidbeta pathology in a diverse neuropathological cohort, Seo, S. W. et al., 13: p. 130137, , 2017

79. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI singlesubject brain, TzourioMazoyer, N. et al., 15(1): p. 27389, , 2002

80. Phase 1 study of the Pittsburgh compound B derivative 18Fflutemetamol in healthy volunteers and patients with probable Alzheimer disease, Nelissen, N. et al., 50(8): p. 12519, , 2009

81. Whole‐body tumor segmentation from PET/CT images using a two‐stage cascaded neural network with camouflaged object detection mechanisms, He, J. et al., 2023, , 2023

82. Cerebral PET with florbetapir compared with neuropathology at autopsy for detection of neuritic amyloidbeta plaques: a prospective cohort study, Clark, C. M. et al., 11(8): p. 66978, , 2012

83. Fullyautomated approach to hippocampus segmentation using a graphcuts algorithm combined with atlasbased segmentation and morphological opening, Kwak, K. et al., 31(7): p. 1190 6, , 2013

84. Cerebral amyloidbeta PET with florbetaben (18F) in patients with Alzheimer's disease and healthy controls: a multicentre phase 2 diagnostic study, Barthel, H. et al., 10(5): p. 42435, , 2011

85. Multisite study of the relationships between antemortem [(11)C]PIBPET Centiloid values and postmortem measures of Alzheimer's disease neuropathology, La Joie, R. et al., 15(2): p. 205216, , 2019

86. Association between in vivo fluorine 18labeled flutemetamol amyloid positron emission tomography imaging and in vivo cerebral cortical histopathology, Wolk, D. A. et al., 68(11): p. 1398 403, , 2011

87. Comparison of (18)Fflorbetaben quantification results using the standard Centiloid, MRbased, and MRless CapAIBL((R)) approaches: Validation against histopathology, Dore, V. et al., 15(6): p. 807816, , 2019

88. 18F]flutemetamol amyloid positron emission tomography in preclinical and symptomatic Alzheimer's disease: specific detection of advanced phases of amyloidbeta pathology, Thal, D. R. et al., 11(8): p. 97585, , 2015

89. Automatic Brain Tumor Segmentation Using Cascaded Anisotropic Convolutional Neural Networks in Brainlesion Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries, Wang, G. et al, Cham: Springer International Publishing, , 2018

90. The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease, McKhann, G. M. et al., 7(3): p. 2639, , 2011