Beyond lesion-language mapping in aphasia: A novel imaging-based prediction model

对脑卒中后失语症的语言康复反应不一，主要是因为很少

Response to post-stroke aphasia language rehabilitation is variable, mainly because there are few predictors that can help identify individualized treatment options. Imaging techniques, such as Voxel-based Lesion Symptom Mapping (VLSM) have been useful in linking specific brain areas to language behavior; however, further development is required to optimize use of structural and functional information in guiding individualized treatment for our Veterans with aphasia. This CDA1 addresses this gap through development of a novel technique that improves prediction of language behavior using anatomical measure of gliosis+ as well as physiological measures such as Cerebral Blood Flow (CBF) and Glucose Extraction Fraction (GEF). In the first aim, we test the sensitivity of our novel anatomical measure, gliosis+, in relating to confrontation naming of nouns and verbs, thereby advancing current VLSM techniques. Our approach to testing the working hypothesis will be to recruit patients with aphasia on whom we will test confrontation naming of nouns and verbs and acquire high-resolution structural MRI scans. The study of nouns and verbs has multiple avenues of clinical significance, including (1) anomia is the most pervasive deficit in aphasia, (2) noun and verb naming can be differentially impaired in aphasia, and (3) nouns and verbs are often targets in impairment-based treatment. We plan to determine the structure-behavior association with three different methodologies (binary lesion maps, continuous T1w signal maps, gliosis + maps), and statistically compare the amount of behavioral variance that each VLSM technique accounts for. After completing Aim 1, it is our expectation that we will have identified which VLSM methodology accounts for the most behavioral variance. In the second aim, we develop a novel methodology called Voxel-based Lesion and Physiology Symptom Mapping (VLPSM) by combining gliosis+ VLSM with regional CBF to improve the predictive capability of structure-behavior association maps, and expand them to structure-function-behavior association maps. We will then compare VLPSM to VLSM, and determine which methodology accounts for more variance in language behavior, with the expectation that combining structural and physiological information will account for more language behavior variance across subjects than when considering structural information alone. After completing Aim 2, it is our expectation that we will have a robust and novel method of mapping language behavior to brain areas utilizing information from both structure and physiology. In the third aim, we will develop regional GEF measures on a Magnetic Resonance Imaging (MRI) system, which can be used as a better marker of neuronal health in VLPSM. Our approach will be to develop the theory of the pulse sequence, ensure proper MRI signal formation via Bloch equation modeling, simulate different conditions of the MRI signal formation, and implement the pulse sequence on a 3T Siemens Prisma platform. We will perform extensive quality and reproducibility tests on phantoms. If our target sensitivity, spatial resolution, and reproducibility are not met, we will identify the limiting factors and overcome them through improved pulse sequence design and potentially improvements in RF hardware. Finally, we will test the novel GEF technique on neurologically healthy young and old subjects to determine reproducibility and sensitivity to physiological changes due to age, and compare these results with CBF measures. After completing Aim 3, it is our expectation that we will have a fully developed methodology to non-invasively measure regional glucose extraction fraction in the brain. The over-arching objective of this application is to create a map that contains both anatomical and physiological correlates of language behavior, highlighting key target brain areas for rehabilitation. The long- term goal of this line of research is to determine how VLPSM can be best used to provide clinicians with individualized maps to predict aphasia treatment outcome, thereby assisting in treatment choices.