The hemodynamic response function as a type 2 diabetes biomarker: a data-driven approach

Abstract

Introduction: There is a need to better understand the neurophysiological changes associated with early brain dysfunction in Type 2 diabetes mellitus (T2DM) before vascular or structural lesions. Our aim was to use a novel unbiased data-driven approach to detect and characterize hemodynamic response function (HRF) alterations in T2DM patients, focusing on their potential as biomarkers.

Methods: We meshed task-based event-related (visual speed discrimination) functional magnetic resonance imaging with DL to show, from an unbiased perspective, that T2DM patients' blood-oxygen-level dependent response is altered. Relevance analysis determined which brain regions were more important for discrimination. We combined explainability with deconvolution generalized linear model to provide a more accurate picture of the nature of the neural changes.

Results: The proposed approach to discriminate T2DM patients achieved up to 95% accuracy. Higher performance was achieved at higher stimulus (speed) contrast, showing a direct relationship with stimulus properties, and in the hemispherically dominant left visual hemifield, demonstrating biological interpretability. Differences are explained by physiological asymmetries in cortical spatial processing (right hemisphere dominance) and larger neural signal-to-noise ratios related to stimulus contrast. Relevance analysis revealed the most important regions for discrimination, such as extrastriate visual cortex, parietal cortex, and insula. These are disease/task related, providing additional evidence for pathophysiological significance. Our data-driven design allowed us to compute the unbiased HRF without assumptions.

Conclusion: We can accurately differentiate T2DM patients using a data-driven classification of the HRF. HRF differences hold promise as biomarkers and could contribute to a deeper understanding of neurophysiological changes associated with T2DM.

Keywords: deep learning; functional magnetic resonance imaging; hemodynamic response; neuroimaging; type 2 diabetes.

Figure 1

Training, tuning, and testing workflow. Dataset 2 (DS2) subjects were age and gender-matched guarantying results validity. Dataset 1 (DS1), composed of the remainder 80% of the subjects, was used for training and tuning (90%/10% split).

Figure 2

Type 2 diabetes mellitus classification performance. On (A), area under the receiver operating characteristic curve (AUC), accuracy, and F1-score boxplots. Median, first and third quartile represented. Whiskers at each quartile plus 1.5 times the inter quartile range. On (B), discriminated values per hemifield and stimulus condition (at psychophysical threshold level, or submaximal motion contrast level) as indicated. Note the better performance for the left hemifield, corresponding to the right hemisphere. To understand the influence of pre-processing and/or in-scanner head movement, a neural network was trained on raw/unprocessed data and then used to classify both unprocessed (C) and motion-corrected data (D).

Figure 3

Clustering of relevance heatmaps computed by deep Taylor decomposition to define regions of interest. (A) - Individual pixel contributions were backpropagated over the convolutional neural network (CNN) to generate relevance heatmaps. (B-G) - Averaged heatmaps were thresholded and labeled to create relevance clusters. Sagittal (SAG), coronal (COR), and transverse (TRA) planes are shown for each cluster. These are located approximately in the superior frontal gyrus (B), angular gyrus (C), extrastriate visual and parietal cortex (D), insula (E), cerebellum (F), and thalamus (G), left to right and, top to bottom, respectively. Clusters are represented with distinct colors for the sake of convenience. P – Posterior; A – Anterior; S -Superior; I – Inferior.

Figure 4

Estimated hemodynamic response function by deconvolution-based analysis. The hemodynamic response function was computed over the six indicated clusters for controls and type 2 diabetes patients (T2DM). These are (left to right, top to bottom) the superior frontal gyrus (A and D), angular gyrus (B and E), visual cortex (C and F), insula (G and J), cerebellum (H and K), and thalamus (I and L), respectively. Average and standard deviation are represented (line and shaded area, respectively), for each condition (threshold and sub-maximum as indicated). Dashed line highlights the time of peak response of control subjects.