Digitally quantified neuropathological correlates of structural and functional imaging biomarkers in progressive supranuclear palsy

Neuroimaging measures are increasingly useful as *in vivo* biomarkers for differential diagnosis in neurodegenerative tauopathies. However, the relationship between imaging changes and neuropathology requires more thorough validation and beyond Alzheimer’s disease. In tauopathies, tau is the key protein where specific pattern of tau aggregation and distribution can distinguish between different tauopathies. With this, progressive supranuclear palsy (PSP) is a prime disease for investigating the relationship between imaging changes and tau burden. Early studies have revealed that semi-quantitative pathological tau measures are predictive of *in vivo* atrophy but is less predictive of intrinsic functional connectivity. However, the semi-quantitative pathological measures used in these studies have limitations such that they are prone to inter-rater variability and may be insensitive to subtle pathological patterns. Therefore, in this thesis, I first attempted to develop an automated pipeline for quantifying total and tau positive cell density for neuronal and glial cells in *post mortem* samples. This promises to be a more objective, detailed, and scalable solution for pathological assessment. However, cell classification with high accuracy has proven to be very challenging; I then adapted the pipeline to only focus on quantifying tau type-specific aggregates in PSP. I used the pipeline to quantify PSP-related tau aggregates across cortical and subcortical regions and found a strong correspondence between the digitally quantified tau burden and the current consensus PSP staging scheme. I further assessed its clinicopathologic predictive power and found that total cortical tau and subcortical neurofibrillary densities correlate with clinical severity measured prior to death. Lastly, I assessed the relationship between both structural and functional imaging measures and *post mortem* tau type-specific density. I found that total tau density was associated with *in vivo* functional connectivity but not atrophy, suggesting that structural measure may be a less reliable marker of tau burden than functional measure in PSP. Overall, I have demonstrated that digitally quantified tau pathology can be a powerful tool to deepen our understanding of the role of tau for mechanistic studies and diagnosis in neurodegenerative tauopathies.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2023

Author dc:creator

• Pansuwan, Tanrada

Advisor dc:contributor.advisor

• Rittman, Timothy

• Digital pathology
• Machine learning
• Neuroimaging
• PSP
• Random forest
• Tau
• Tauopathies

Licence

• https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/4ad09a30-d338-43d8-b884-47bd9aeb7540/download
• https://creativecommons.org/licenses/by/4.0/

Language dc:language

eng

DOI dc:identifier.doi

https://doi.org/10.17863/CAM.107343

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/366405