Evaluating the Relationship between Motor and Cognitive Impairment and Lesion Size and Location Post-Stroke Using CT Scans

Introduction: Stroke is one of the most common causes of death and disability in America and in other countries around the world[1]. Stroke causes brain damage in the form of a lesion, from which lesion location and lesion volume can be harnessed. To visualize the lesion, a CT scan is completed, and CT scans are universally available in both low and high-resource settings. In addition to lesion formation, a person suffering from a stroke will have impairment in their motor and/or cognitive functions. Furthermore, the brain is deeply interconnected. Studies show that 77% of stroke survivors have upper limb motor deficits and about 40% also have at least mild cognitive deficits[2]. Despite this, there are few studies that look at the prevalence of concurrent motor and cognitive deficits, but of the studies that have, they found that motor performance is affected by cognition and low/moderate cognition can mask high motor performance[3-4]. Stroke rehabilitation traditionally relies on “impairment-level treatments” based solely on their motor and cognitive performances, with an emphasis on motor performance[5]. Given that cognitive impairments may mask motor high motor performance, a subject may incorrectly categorize a patient’s impairment level and possibly, therapeutic strategy. Thus, relying on behavior alone to assess function assumes a simplistic and unrealistic view of the brain. This study’s goal is to use affordable imaging tools to gain a priori knowledge of how the lesion(s) effected brain networks would further improve the initial stratification of patients into treatment groups and may better predict patients’ recovery trajectories.

Materials and

Methods: ITK-SNAP, an interactive tool developed for medical imaging segmentation, was used to create the lesion masks. The Statistical Parametric Mapping (SPM Version 12) Clinical Anatomy toolbox in MATLAB[6] is used to register individual CT scans to a uniform space. After normalizing the images to a uniform space, MRIcron, a brain imaging software, to compute lesion volume is used. To accomplish the goal of capturing Brodmann areas and subcortical brain regions, the Jülich Probabilistic Cytoarchitectonic Atlas and the Automated Anatomical Labeling Atlas 3 (AAL 3) were used. The Julich atlas combines cytoarchitectonic maps with microstructural and connectivity brain data into one common reference space, replacing historical maps of the brain’s microstructure (Brodmann areas) while the AAL 3 helps capture the subcortical brain regions. The AAL 3 provides structural parcellations of some subcortical brain nuclei (such as the thalamus) and allows on to examine extensive subcortical involvement. MRIcron was used to yoke the lesioned areas to the anatomical landmark on the atlas. Region of interest involvement was calculated by taking the region of interest involvement divided by total regions involved over all the slices.
Lesion location was computed at four whole-brain categories: by hemispheres, by brain depth, by lobes, and by Brodmann areas. Motor and cognitive behavior was determined by performance tests conducted during patient evaluation: motor and cognitive function were assessed by the Fugl Mayer Assessment (FM) and Montreal Cognitive Assessment (MOCA) respectively.

Results, Conclusions, and Discussions: Results and

Discussion: Lesion metrics were harnessed from the radiological images of thirteen subjects (57 ± 11 years old; male: 53.8%) manually. All subjects were 18 years or older, clinically diagnosed with stroke, and able to voluntary move their upper extremity. Although all subjects were clinically diagnosed with stroke, lesion metrics could not be harnessed from the acute CT Scan for 1 participant. The remaining participants had diverse lesions, spanning coverage across all brain lobes, except the occipital lobe. 67% of the identified lesions were characterized on the left brain, 25% on the right brain, and 8.3% of lesions spanned both hemispheres. The lesions also covered bot cortical and subcortical areas of the brain, with 42% of strokes covering just cortical areas, 25% covering just subcortical areas, 25% spanning both cortical and subcortical areas, and 8.3% involving the brainstem area. As for clinical metrics, the participants also displayed varied impairment levels. The participants presented with severe to mild cognitive impairments, with scores on the MoCA ranging from 9-29 (36 total) and an average of 22 +/- 5.9 (moderate impairment). Similarly, the subjects had mild to severe motor impairments, as assessed with the FM assessment, with scores ranging from 8-66 (66 total) and an average of 50 +/- 20.5 (moderate impairment). These preliminary data indicate a positive linear relationship between the insular lobe and cognitive function (r2 = 0.527), frontal lobe and cognitive function (r2 = 0.411), and the insular lobe and FM (r2 = 0.287). Further, the data indicate a positive linear relationship between motor function and brainstem lesions (r2 = 0.338).

Conclusion: These preliminary results suggest that we can harness lesion metrics from acute CT scans, a contested topic in the literature. These results also suggest possible relationships between structure and function; however, a larger sample size is needed. Additionally, a second independent tracer is necessary to validate the reliability of the metrics.

Acknowledgements (Optional): This work was made possible through the core services and support from the NIH-funded parent and diversity supplement grants R42 HD104325, University of Pennsylvania Fontaine Society, and the Departments of Bioengineering and Physical Medicine and Rehabilitation.