The Heart-Brain Connection is a multicenter study assessing a wide range of parameters of cardiovascular, hemodynamic, and cognitive functioning. The aim of this pilot analysis is to show that we are able to bring together imaging and non-imaging data from four sites, using multiple centralized platforms, and to run statistical analyses. We examined (for pilot purposes only) the relationship between cardiac output, cerebral blood flow (phase-contrast MRI, arterial spin labeling), white matter hyperintensities and memory and executive functioning in patients with vascular cognitive impairment (VCI), heart failure (HF), carotid occlusion disease (COD), and controls (CN).
Randomly select 10 participants per center.
Quality assessment (QA) heart scans
QA brain scans
Processing heart scans
Processing brain scans
Phase-contrast MRI (QFlow) analysis
Arterial spin labeling (ASL) and region-based analysis pipeline
QA results image processing pipelines
Upload imaging parameters to the imaging platform XNAT into sessions of participants
Download imaging parameters from XNAT into combined file for all participants
Download cognitive parameters and demographics from OpenClinica
Combine parameters and perform analysis in SPSS
Cardiac pipeline: Stroke volume (SV) [ml], Ejection Fraction (EF) [%], Pulse Wave Velocity (PWV) [m/s] (22% missing, because of missing scans or scanning artefacts)
QFlow analysis: total cerebral blood flow (tCBF) [ml/min] (5% missing, because of missing scans)
Tissue pipeline: volume of white matter hyperintensities (WMH) [% of intracranial volume] (7% missing, because of missing scans)
ASL + regions pipeline: cerebral blood flow in the global gray matter (CBFGM) [ml/100g/min] (10% missing, because of missing scans and artefacts)
Cognition: Attention & executive functioning domain and Memory domain (z-scores) (0% missing on cognitive domains, 7% missing on separate test scores)
We randomly selected 10 subjects from each site (Table 1). Total pilot population n = 40.
Associations between imaging markers and cognitive domains.
Table 2 shows associations between cognitive domain z-scores and imaging markers of the different pipelines. Although analyses were performed on a limited sample (n=40), expected trends can be observed, e.g. white matter hyperintensity volume relates to a decrease in cognitive functioning (Fig. 1). As Fig. 2 shows, imaging markers can also be associated with demographic parameters such as systolic blood pressure.
This pilot study showed that we are able to assess the quality of imaging data, perform automated analysis of imaging data, and bring together imaging and non-imaging data from multiple sites and platforms, and run statistical analyses. All imaging data, non-imaging data and processed imaging markers are centrally stored and easily accessible within the project. With this result we have shown that the Heart-Brain consortium is in a good position to perform the main analyses. Also, we have established that the collaboration between clinical and technical work packages is successful.