can classify paretic hand outcomes with an accuracy of 0.88 in chronic stroke patients.
ALFF outperformed other metrics in distinguishing between partially and completely paretic hands.
Key brain regions linked to ALFF included the precentral gyrus, posterior cerebellum, and parietal lobule.
Neuroimaging biomarkers from ALFF may relate to the G protein-coupled receptor signaling pathway.
These biomarkers could be significantly expressed in astrocytes and early fetal brain stages.
Noradrenaline levels showed a positive correlation with ALFF biomarker distribution.
AI simplified
BACKGROUND: Hand dysfunction significantly affects independence after stroke, with outcomes varying across individuals. Exploring biomarkers associated with the paretic hand can improve the prognosis and guide personalized rehabilitation. However, whether biomarkers derived from resting-state fMRI (rs-fMRI) can effectively classify and predict different hand outcomes and their biological mechanisms remain unclear.
METHODS: This study analyzed 65 patients with chronic subcortical stroke, including 32 patients with partially paretic hand (PPH) and 33 patients with completely paretic hand (CPH). For patients with PPH and CPH respectively, the age was 56.19 ± 10.53 and 55.60 ± 9.00 years, disease duration was 15.31 ± 14.87 and 14.12 ± 17.36 months, lesion volume was 9.45 ± 5.57 and 16.00 ± 11.33 ml, (FMA-HW) was 11.25 ± 6.15 and 1.24 ± 1.22. Four rs-fMRI metrics were analyzed, including (ALFF), regional homogeneity (ReHo), degree centrality (DC), and voxel-mirrored homotopic connectivity (VMHC). Multivariate pattern analysis was used to classify and predict paretic hand performance. To explore the biological mechanisms of neuroimaging biomarkers, partial least squares regression was conducted to associate gene expression data (from Allen Human Brain Atlas), neurotransmitter maps, neuron types and developmental stages with the classification weight maps of rs-fMRI metrics.
RESULTS: ALFF achieved a higher classification accuracy of 0.88 in differentiating PPH from CPH, outperforming the other three rs-fMRI metrics. Machine learning further identified the top contributing regions from the ALFF classification weight maps, such as the ipsilesional precentral gyrus, contralesional cerebellum posterior lobe, and ipsilesional parietal lobule. Neuroimaging-transcriptome analysis revealed that macroscopic biomarkers from the ALFF were associated with the G protein-coupled receptor signaling pathway and the detection of chemical stimuli involved in sensory perception. Additionally, these neuroimaging biomarkers from ALFF showed prominent expression in astrocytes and early fetal stages. Most importantly, the neurotransmitter noradrenaline positively correlated with the distribution of ALFF biomarkers.
CONCLUSIONS: The ALFF is an effective macroscopic biomarker for classifying and predicting paretic hand outcomes in individuals following chronic stroke. These neuroimaging biomarkers correspond to molecular transcriptional profiles and neurotransmitter distributions, offering insights into the potential of personalized stroke rehabilitation.
Key numbers
0.88
Classification Accuracy of
Accuracy in differentiating from patients
9.45 ± 5.57 ml vs. 16.00 ± 11.33 ml
Lesion Volume Comparison
Lesion volume for vs. patients
Full Text
We can’t show the full text here under this license. Use the link below to read it at the source.
