Diffusion kurtosis imaging as a neuroimaging biomarker in patients with carbon monoxide intoxication

Attempting suicide by burning charcoal can lead to carbon monoxide (CO) intoxication and cognitive deficits. Changes in white matter (WM) quantified by diffusion tensor imaging (DTI)-derived parameters have been validated to reflect cognitive test scores. As diffusion kurtosis imaging (DKI) measures biological microstructures using non-Gaussian diffusivity, we assessed the added-information of DKI with neuropsychological test scores as the major outcome measure. A total of 45 patients were enrolled and compared with 30 age-matched controls. The patients were stratified into acute or chronic phase according to the intervals of intoxication and assessments. WM status was assessed using tract-based spatial statistics for DKI and DTI topographies, and the sensitivity/specificity of either model was tested using area under the curve (AUC) analysis. To evaluate their clinical significance, values of DKI- and DTI-derived parameters were extracted from seven regions of interest (ROI) and correlated with neuropsychiatric scores. The kurtosis parameters were lower in the patients than in the controls but none of the parameters provided differentiations between the acute or chronic phase. Kurtosis fractional anisotropy (KFA) had a higher AUC than fractional anisotropy while the other 3 DTI parameters had higher AUC than the corresponding DKI ones. In clinical correlations, KFA value of right posterior WM correlated with visual memory (r = 0.326, p = 0.029), and KFA values of bilateral posterior WM correlated with the digit forward score (right: r = 0.302, p = 0.043; left: r = 0.314, p = 0.036). Although DTI was more sensitive in reflecting disease status, KFA may be more sensitive and specific than fractional anisotropy in cognitive test score predictions.