PerAF-based resting-state fMRI classifier for minimal hepatic encephalopathy

Zhijian Liang^1,2,3*Yunli Zhang^1,4Zhai Huang⁴Bin Qin⁵Shiting Tang⁶Peng Huang⁴Xiaogang Tang⁴Xuezhen Lin⁴Liu Yayuan⁷Xuhui Deng⁸Feiqi Zhu⁹Shihui Xing¹⁰

• ¹Department of Neurology, First Affiliated Hospital, Guangxi Medical University, Nanning, China
• ²Guangxi Key Laboratory of Precision Medicine in Cardio-cerebrovascular Diseases Control and Prevention, First Affiliated Hospital, Guangxi Medical University, Nanning, Guangxi Zhuang Region, China
• ³Guangxi Clinical Research Center for Cardio-cerebrovascular Diseases, Nanning, People’s Republic of China,, nanning, China
• ⁴Department of Intensive Care Unit, Guangxi Academy of Medical sciences, the People's Hospital of Guangxi Zhuang Autonomous Region, Nanning, China
• ⁵Department of Neurology, Liuzhou People's Hospital, Liuzhou, Guangxi Zhuang Region, China
• ⁶Department of Neurology, Second Affiliated Hospital of Guangxi Medical University, Nanning, China
• ⁷Department of Neurology, Minzu Hospital of Guangxi Zhuang Autonomous Region, Nanning, China
• ⁸Department of neurology, Yuebei people's hospital, Guangdong, China, Guangzhou, China
• ⁹Department of neurology, Luohu Hospital of Shenzhen, Guangdong, China
• ¹⁰Department of Neurology, Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou, China

Background: Minimal hepatic encephalopathy (MHE) is a common cognitive impairment in patients with end-stage liver cirrhosis. However, the selection of sensitive biomarkers and the establishment of reliable diagnostic methods are currently challenging. To explore the abnormal brain spontaneous activity in patients with MHE, and to evaluate the clinical diagnostic value of four indicators for MHE by support vector machine (SVM) method. Methods: A total of 45 MHE patients and 40 healthy controls were enrolled. Amplitude of low frequency fluctuation (ALFF), fractional amplitude of low frequency fluctuation (fALFF), percentage amplitude of low frequency fluctuation (PerAF) and regional homogeneity (ReHo) were used to evaluate local spontaneous brain activity. SVM analysis was used to construct the classification model and evaluate the diagnostic value. Results: Two-sample t-test and SVM results showed that, compared with the healthy control group, MHE patients had decreased ALFF values in the left angular gyrus, right inferior temporal gyrus, left postcentral gyrus, precentral gyrus, and right supplementary motor area. These regions demonstrated moderate classification efficacy (AUC=0.75). Decreased ReHo metrics in the right anterior cingulate and paracingulate gyri also showed general discriminative power (AUC=0.72). fALFF metrics, whether analyzed independently or combined with other indicators, exhibited limited classification performance (AUC<0.70). Decreased PerAF metrics in the right superior parietal lobule, right dorsolateral prefrontal cortex, and right middle frontal gyrus achieved a good classification accuracy rate (AUC value 0.83; accuracy 81.18%; sensitivity 75.56%; specificity 87.50%), outperforming other functional metrics. 4 Conclusion: We found that decreased mean PerAF in the right supramarginal gyrus, right dorsolateral superior frontal gyrus, and right middle frontal gyrus may serve as potential neuroimaging indicators for early identification of cognitive impairment in MHE patients, providing critical evidence for clinical screening protocols.