edoc

Postmortem imaging using CT and MRI for process automation and biomarker development

Neuhaus, Dominique Louis. Postmortem imaging using CT and MRI for process automation and biomarker development. 2024, Doctoral Thesis, University of Basel, Faculty of Medicine.

[img] PDF
Restricted to Repository staff only until 1 January 2026.
Available under License CC BY (Attribution).

10Mb

Official URL: https://edoc.unibas.ch/96846/

Downloads: Statistics Overview

Abstract

Several methods exist for the forensic identification of unknown deceased, while radiologic identification (RADid) might be the method of choice. Typically, antemortem (AM) and postmortem (PM) computed tomography (CT) data are manually compared to assess similarity. However, this approach is time-consuming, examiner-dependent, and qualitative, therefore, the outcome cannot be supported by statistical evidence. Similarly, several challenges are faced in the diagnosis and prognosis of neurodegenerative diseases, such as Amyotrophic Lateral Sclerosis (ALS). Reasons for these challenges arise due to the heterogeneity of symptoms within one disease and the overlap of symptoms between different diseases. Therefore, there is a great need for magnetic resonance imaging (MRI) biomarkers to aid in diagnosis and prognosis. However, potential new MRI biomarkers need to be validated. Validation usually incorporates ex-situ MRI acquisitions of formalin fixed tissue along with histological evaluation. Nevertheless, performing ex-situ MRI scans is methodologically challenging and histology is fundamentally a qualitative technique, limiting a correlation with MRI findings. The aim of this thesis was therefore to develop forensic and diagnostic techniques based on PM imaging, specifically by automating the identification process of unknown deceased using postmortem computed tomography (PMCT) and by exploring potential biomarkers for ALS via postmortem magnetic resonance imaging (PMMRI).
The process of identifying unknown deceased individuals was automated by developing scripts that perform registration and assess the similarity between AM and PM CT data. In an initial study, two-dimensional (2D) maximum intensity projection (MIP) of CT data from the sternal bone was used, similarly to the current manual procedure (Publication 1; status accepted). In this work, an accuracy of 91.2 % was achieved for the identification process. In a second study, to take advantage of the benefits of CT, three-dimensional (3D) data was used, including both the sternal bone and the fifth thoracic (T5) vertebra (Publication 2; status submitted). This resulted in an accuracy of 97.8 %. These automated approaches are easy to use, examiner-independent, reliable, and efficient. Regarding the MRI biomarker development, an effective method was developed for ex-situ PMMRI including brain-specific 3D printed support plates and the effect of brain extraction and formalin fixation on MRI parameters was assessed by comparing ex-situ to in-situ data (Publication 3; status accepted). Changes in volume, diffusivity, and relaxation parameters were observed. This method along with the findings are crucial prerequisites in biomarker development and validation. In-situ PMMRI was further used to compare MRI parameters between brains of patients with ALS and of controls, exploring potential MRI biomarkers (Publication 4; status submitted). Moreover, a tool was developed to automatically quantify the myelin density in histological data, which was used for the validation of MRI findings. Differences in diffusivity and relaxometry in various brain regions were detected. However, larger sample sizes along with histological analyses are required to confirm the current results.
The tools and techniques developed in this thesis are a valuable contribution to process automation and biomarker development, improving efficiency, reliability, and reproducibility. Moreover, the observed MRI parameter alterations in patients with ALS might lead to the development of novel diagnostic and prognostic biomarkers.
Advisors:Scheurer, Eva
Committee Members:Schläger, Regina and Ropele, Stefan and Lenz, Claudia and Deigendesch, Nikolaus
Faculties and Departments:03 Faculty of Medicine > Departement Biomedical Engineering > Imaging and Computational Modelling > Forensic Medicine (Scheurer)
UniBasel Contributors:Scheurer, Eva and Lenz, Claudia
Item Type:Thesis
Thesis Subtype:Doctoral Thesis
Thesis no:15617
Thesis status:Complete
Number of Pages:XXIII, 171
Language:English
Identification Number:
  • urn: urn:nbn:ch:bel-bau-diss156178
edoc DOI:
Last Modified:05 Feb 2025 05:30
Deposited On:04 Feb 2025 11:42

Repository Staff Only: item control page