Postlesional plasticity in central auditory processing: MRI and FMRI image registration and statistical signal processing

The most general and striking evidence related with brain injury is that of the restoration of function. Recovery of motor and somatosensory functions has shown to commonly occur after stroke, but not all individuals show improvement. Clinical studies have shown the capacity of pharmacological and rehabilitative interventions to accelerate and/or augment recovery after stroke. However, the mechanisms underlying the post-stroke brain functional reorganization, i.e., the brain plasticity, are still not well established. Therefore, therapeutic trials of agents or rehabilitative procedures targeting stroke recovery may benefit from brain mapping studies that may aid to better understand this functional reorganization. Thus the research aimed at the identification of the mechanisms underlying functional recovery should be given high priority, particularly with regard to environmental enrichment, rehabilitation and pharmacological interventions. Prior to investigation of post-stroke functional reorganization, two important conditions have to be gathered: having knowledge of the brain function under normal conditions, i.e., in normal subjects without brain lesions, and having devices adapted to study brain region of interest. This thesis addresses the post-stroke functional reorganization in auditory processing with a cross-sectional study in patients with unilateral hemispheric lesions and a longitudinal study in a patient with a lesion of a right acoustic radiation. Before performing these two studies with patients, normal function of the auditory processing is assessed and appropriate tools are developed. Audition is the key to language processing, the most important communication system in man. Hearing impairments arising from pathology of the brain injury may have detrimental consequences on the quality of the patient life, restricting our ability to interact with others, causing misunderstandings and fatigue, heightening stress and filtering out the myriad of sound experiences that give pleasure and meaning to life. The perception of an auditory scene in everyday acoustic environments involves identifying the content ("what") and the location ("where") of sound. Evidence indicates that sound recognition and sound localization are processed by at least partially independent anatomically distinct networks. In humans, activation studies have suggested existence of a ventral, temporo-frontal, "what" and dorsal, parieto-prefrontal, "where" pathways on the convexities. However, no studies have been able to clearly demonstrate "what" and "where" specialization in early stage auditory areas. This is mainly due to high local interindividual sulcal variability. Indeed, the precise realignment of anatomical landmarks on the supratemporal plane could not be achieved with current registration methods, thus providing less detailed and accurate functional maps. After a short introduction on brain variability, a brief description of auditory cortex structure and function is presented. High variability of the auditory cortex is a major problem when performing activation group studies. For this reason, the interindividual comparison of auditory activations necessitates appropriate method for brain superposition, i.e., brain registration. Two registration approaches are described, voxel-based and feature-based approaches, with demonstrating the benefit of feature-based approaches for interindividual superposition of highly variable cortical structures. Therefore we developed a local landmark-based registration algorithm. This algorithm mainly consisted in semi-automatical extraction of the sulci delimiting Heschl's gyrus and in their realignment, using thin-plate splines, with the corresponding landmarks of the reference brain. We employed this algorithm on "what" and "where" functional data in 18 normal subjects acquired with 1.5 Tesla MR scanner (from [1]) and 15 normal subjects acquired with 3 Tesla MR scanner. Our results have shown that a precise realignment of anatomical structures on the supratemporal plane yielded more detailed functional maps of a group of subjects, which were more consistent with cytoarchitectonically defined auditory areas, than widely used global non-rigid voxel-based registration methods. Moreover, our results demonstrated a specialization of anterior and anterior lateral auditory areas in sound recognition, and caudomedial and posterior auditory areas in sound localization. A first patient study performed on patients with unilateral hemispheric lesion compared "what" and "where" activations on the supratemporal plane between each patient and a group of normal subjects (both acquired with 1.5 Tesla). Our results demonstrated that a unilateral hemispheric lesion, left or right, disturbs auditory processing in both ipsilesional and contralesional hemispheres. This disruption of auditory processing may be an increase or decrease of activation patterns in an area known to be activated for conjoint activations by sound recognition and sound localization. Moreover, our results demonstrated that specialized networks were disturbed in patients with unilateral hemispheric lesions in two different ways: (i) changes in specialized task-related activations, and (ii) increase in specialized task-related networks in the intact hemisphere associated with normal performance in the corresponding task. In a second patient study that is a longitudinal study performed on a patient with a lesion of a right acoustic radiation, we investigated the effect of a unilateral subcortical lesion on the early cortical auditory processing within intact ipsilesional and contralesional hemispheres. Our results demonstrated the contralesional increase in activation of homologous auditory areas in the early stage of recovery, when auditory function was impaired, while at later stage the ipsilesional regions were more activated, as auditory function recovered.


Related material