Work

Advanced Image Processing Concepts for Characterizing Movement Disorders.

Year: 2024
Type: dissertation
Abstract: This study presents clinical and methodological advancements in fMRI studies, particularly in investigating motor activity in Parkinson's Disease (PD) and Gilles de la Tourette Syndrome (GTS). For PD, the motivation is to provide robust alternatives to the intrinsic difficulty to account for exact task movements performed by PD patients inside the scanner and correctly model it for analysis. Modeling the signal in fMRI means to build a function that will most closely mimic the corresponding brain activity emerging from the task. This PD study demonstrated that the traditional block-based design (which is a unitary step function where task activity is one and rest is zero) results in a considerable sensitivity improvement when patients used a non magnetic glove while performing the task. The detection of primary motor cortex was specifically improved when the gadget was incorporated to the experiment. Furthermore, this study also demonstrated that modelling the onset of tapping and rest blocks (a step function with a very small duration instead of one comprising the entire duration of the task before a rest block) detected putamen’s activation a lot better. Given that putamen is a key area involved in motor control affected by L-DOPA, this sensitivity improvement is very useful to monitor drug intake in Parkinson. These findings underscore the importance of customizing fMRI analysis strategies to the specific roles of individual brain regions and the usage of auxiliar technologies to account for real time movement inside the scanner. For GTS, the first motivation was to propose a brain network characterization of GTS relying on the power of a multicentric study. The study managed to characterize the syndrome as a hyper-connectivity network involving six interconnected brain functions: Motor Control, Sensorimotor Control, Speech Control, Proprioception, Behavior Inhibition, and Salience-Attention. The findings offer a deeper understanding of the neural underpinnings of GTS, proposing that the observed connectivity changes are consistent with the syndrome's clinical manifestations. On the methodological perspective, the research addresses the challenges of multicentric functional connectivity analysis, where variability in acquisition settings, preprocessing methods, and analysis pipelines can introduce inconsistencies, making it difficult to homogenize data across studies. This variability can significantly affect connectivity metrics, emphasizing the need for standardized protocols and robust correction methods to achieve reproducible and generalizable findings. To ensure the generalizability of these findings for other resting state fMRI (rsfMRI) researches, this study developed a logical sequence of data quality assessments and proposed a segregation criterion. Through a series of evaluations—ranging from classical signal-to-noise ratio computations to more subtle metrics like anatomical and functional overlap—the study identified that variance in white matter effectively segregated cohorts and correlated with key quality properties such as bilateral correlation of basal ganglia and edge count (number of pairwise connections). This approach allowed us to create a meaningful hierarchy of data quality, ultimately resulting in improved sensitivity in fMRI studies after applying the segregation criterion. Overall, the study proposes significant clinical and methodological improvements that enhance the reliability of fMRI studies in both PD and GTS, with broader implications for other fields utilizing task and resting state fMRI.:CHAPTER 1. Overview.……………………………………………………………18 1.1. fMRI………………………………………………………………….…………..20 1.2. Preprocessing…………………………………………………………….…….26 1.2.1. Slice-timing correction………………………………………….………..26 1.2.2. Motion correction………………………………………………….……..26 1.2.3. Co-registration……………………………………………………………26 1.2.4. Normalization……………………………………………………………..27 1.2.5. Smoothing………………………………………………………………...27 1.3. The use of GLM in fMRI……………………………………………………… .27 1.4. Functional connectivity………………………………………………… ……..32 1.4.1. Functional centrality measures, pairwise correlation and Edge count…………………………………………………………..33 1.4.1.1. Intrinsic Connectivity……………………………………………..34 1.4.1.2. ROI-to-ROI Connectivity…………………………………………35 1.4.1.3. Edge Count………………………………………………………..35 2. CHAPTER 2. Movement disorders………………………………………………..36 2.1. Gilles de la Tourette syndrome………………………………………………..37 2.2. Parkinson´s disease……………………………………………………………39 3. CHAPTER 3. Improving fMRI in Parkinson’s disease by accounting for brain region-specific activity patterns. 3.1. Introduction……..………………………………………………………………40 3.2. Material and Methods. 3.2.1. Subjects…………………………………………………………………...42 3.2.2. Experimental Design and Data Acquisition……………………………43 3.2.3. Medication Effect on UPDRS-III Scores and Kinematics…………….45 3.2.4. fMRI Data Analysis……………………………………………………….45 3.3. Parkinson´s disease results. 3.3.1. Medication Effect on Clinical Scores……………………………………47 3.3.2. fMRI Main Effects of Tapping versus Rest…………………………….49 3.3.3. Effect of l-DOPA Medication on fMRI Results………………………...54 3.4. Discussion………………………………………………………………………55 3.5. Conclusion………………………………………………………………………61 4. CHAPTER 4. Multicentric GTS fMRI functional connectivity study. 4.1. Introduction……………………………………………………………………..61 4.2. Material and Methods 4.2.1. Multicentric study: 5 Cohorts……………………………………………62 4.2.2. Preprocessing. 4.2.3. Experimental Design, Data Acquisition and Resting-state fMRI analysis……………………………………………………………………63 4.2.4. White matter signal as physiological noise index……………………..64 4.3. Results………………………………………………………………………….65 4.3.1. Clinical Results…………………………………………………………...66 4.3.1.1. ICC GTS patients vs Controls individual cohorts……………..68 4.3.1.2. ICC GTS patients vs Controls combined cohorts……………..69 4.3.1.3. ROI to ROI GTS patients vs Controls individual cohorts……..69 4.3.1.4. ROI to ROI GTS patients vs Controls combined cohorts…….72 4.3.1.5. Edge count………………………………………………………..73 4.3.2. Methodological Findings………………………………………………...73 4.3.2.1. Physiological Noise Index (PNI)………………………………...74 4.3.2.2. Edge counts for all cohorts………………………………………78 4.3.2.3. A connectivity pattern among all cohorts……………………….81 4.3.2.4. How different are the cohorts? ANOVA………………………..83 4.3.2.5. Leave-one-out and Leave-two-out analyses..………………...85 4.3.2.6. Gradual Inclusion of cohorts…………………………………….86 4.3.2.7. The Inclusion Criterion…………………………………………...87 4.4. Discussion 4.4.1. ICC results’ discussion…………………………………………………..94 4.4.2. RRC results’ discussion………………………………………………….98 5. Conclusion……………………………………………………………………………99 6. Bibliography…………………………………………………………………………101 7. Appendix 7.1. Curriculum……………………………………………………………………...116 7.2. Supplementary Tables………………………………………………………..119 7.3. Supplementary Figures……………………………………………………….122
Author: Torrecuso, Renzo
Institution: Leipzig University
Language: English
Open Access status: green
OpenAlex ID: W6979786072
Cites: 189
Cited by:
Related to: 10