Diagnostic identification of chronic insomnia using ALFF and FC features of resting-state functional MRI and logistic regression approach
Ning Yang,Shuyi Yuan,Chunlong Li,Wenqing Xiao,Shuangcong Xie,Liming Li,Guihua Jiang,Xiaofen Ma
DOI: https://doi.org/10.1038/s41598-022-24837-8
IF: 4.6
2023-01-09
Scientific Reports
Abstract:Abstract This study investigated whether the amplitude of low-frequency fluctuation (ALFF) and functional connectivity (FC) features could be used as potentially neurological markers to identify chronic insomnia (CI) using resting-state functional MRI and machine learning method logistic regression (LR). This study included 49 CI patients and 47 healthy controls (HC). Voxel-wise features, including the amplitude of low-frequency fluctuations (ALFF) and functional connectivity (FC), were extracted from resting-state functional magnetic resonance brain images. Then, we divided the data into two independent cohorts for training (44 CI patients and 42 HC patients), and independent validation (5 CI patients and 5 HC patients) by using logistic regression. The model was evaluated using 20 rounds of fivefold cross‐validation for training. In particular, a two-sample t-test (GRF corrected, p-voxel < 0.001, p-cluster < 0.05) was used for feature selection during the model training. Finally, single‐shot testing of the final model was performed on the independent validation cohort. A correlation analysis (Bonferroni correction, p < 0.05/4) was also conducted to determine whether the features contributing to the prediction were correlated with clinical characteristics, including the Insomnia Severity Index (ISI), Pittsburgh sleep quality index (PSQI), self-rating anxiety scale (SAS), and self-rating depression scale (SDS). Results showed that resting-state features had a discrimination accuracy of 86.40%, with a sensitivity of 93.00% and specificity of 79.80%. The area under the curve (AUC) was 0.89 (all $${P}_{\mathrm{permutation}}$$ P permutation < 0.001). The ALFF and FC features showed significant differences between the CI patients and HC. The regions contributing to the prediction mainly included the anterior cingulate, prefrontal cortex, orbital part of the frontal lobe, angular gyrus, cingulate gyrus, praecuneus, parietal lobe, temporal gyrus, superior temporal gyrus, and middle temporal gyrus. Furthermore, some specific functional connectivity among related regions was positively correlated with the ISI, and also negatively related to the SDS in correlation analysis. Our current study suggested that ALFF and FC in the regions contributing to diagnostic identification might serve as potential neuromarkers for CI.
multidisciplinary sciences