13. Quantitative Susceptibility Mapping (QSM) Analysis of Multiple sclerosis lesions.
The video recording of this talk can be viewed here.
Organizer: Sandra Hurtado Rúa
Cleveland State University
Email: s.hurtadorua@csuohio.edu
Chair: Sandra Hurtado Rúa
Cleveland State University
Email: s.hurtadorua@csuohio.edu
Speakers:
1. Susan Gauthier
Weill Cornell Medicine
Email: sag2015@med.cornell.edu
Title: The clinical translation of QSM as a new imaging biomarker for disease progression and treatment response in Multiple Sclerosis
Time: 3:30pm-3:50pm
Abstract:
Chronic CNS inflammation in the multiple sclerosis (MS) lesions is maintained with iron-containing pro-inflammatory microglia and macrophages at the rim of chronic active MS lesions. Histologically, these lesions demonstrate ongoing demyelination and expansion, which may play an essential role in the pathogenesis of progressive clinical decline. Quantitative susceptibility mapping (QSM) is an imaging technique that provides efficient in vivo quantification of susceptibility changes related to iron deposition. We have demonstrated that QSM can detect iron at MS lesion rims and have provided in-vivo validation that these lesions have more inflammation and tissue damage. We have also demonstrated that QSM rim lesions (rim+) have a temporal trajectory with an increase and subsequent decrease in susceptibility, which is consistent with the transition from a chronic active to a chronic inactive lesion. We have preliminary data demonstrating the influence of rim+ lesions on clinical disability, including cognition, promoting the potential for QSM as a tool to understand mechanisms of injury leading to disease progression. Our next stage is to determine the utility of QSM as a biomarker for treatment response. Monitoring chronic MS lesions in response to treatment would provide a novel and essential therapeutic strategy to reduce tissue injury, neuronal degeneration and clinical disability.
2. Elizabeth Sweeney
Weill Cornell Medicine
Email: ems4003@med.cornell.edu
Title: QSM Image Analysis: Automated Lesion Type Identification and Lesion Age Estimation
Time: 3:50pm-4:10pm
Abstract:
Quantitative susceptibility mapping (QSM) rim (rim +) positive multiple sclerosis (MS) lesions and their longitudinal behavior have the potential to serve as a biomarker of chronic inflammation and to be utilized to monitor disease progression and evaluate disease-modifying therapies. Here we introduce the image analysis tools that will enable us to use QSM rim+ lesions for this purpose: an automated method for identifying QSM lesion type and methods for determining the accurate inflammatory stage or age of a lesion. We first introduce an automated algorithm for identifying QSM rim+ MS lesions in order to reduce the bias and burden of manual identification. This algorithm utilizes first-order radiomic features calculated over a lesion and a random forest classification model to classify lesions as QSM rim+. In a validation set, the algorithm obtained an area under the receiver operating characteristic curve (AUC) of 0.88 and an accuracy of 81%. We next introduce methodology for determining the accurate inflammatory stage or age of a QSM rim+ lesion in both cross sectional and longitudinal settings. This is crucial for evaluating the impact of disease-modifying therapies on the longitudinal behavior of these lesions, as temporal misalignment of lesions may obscure treatment effects. We first introduce a random forest model using radiomic features from multi-sequence MRI to classify lesions as less than a year or greater than a year old in a cross-sectional setting. In a validation set we obtain an AUC of 0.89 and an accuracy of 82%. We next move to a longitudinal setting where we employ curve-registration techniques to temporally align longitudinal information from these lesions.
3. Sandra Hurtado Rúa
Cleveland State University
Email: s.hurtadorua@csuohio.edu
Title: Statistical challenges in the analysis of QSM maps
Time: 4:10pm-4:30pm
Abstract:
Quantitative susceptibility maps (QSM) have the potential to be a biomarker in Multiple Sclerosis with the ability to inform clinical management of disease progression and therapy. In this talk, we introduce a few statistical models for the analysis of group data in the context of QSM with a clinical translation goal. We first address the multiplicity data problem in the context of mixed-effects models with applications to the identification of inflammation in a subset of chronic multiple sclerosis lesions. We then introduce a multilevel growth curve model to compare longitudinal susceptibility among rim+ and rim− lesions. Finally, we explore the advantages and disadvantages of structural equations models in analysis of QSM as a longitudinal biomarker in Multiple Sclerosis.