Resting-state functional MRI (rsfMRI) is one of the most important neuroimaging modalities for investigating alterations in the resting-state networks of the human brain, given that abnormal neural activity during the resting state is associated with neurological disorders. However, neuroimaging results obtained from rsfMRI have rarely been replicated with repeated measurements. Therefore, we aimed to develop new measures to extract highly reliable and reproducible functional neuroimaging metrics from rsfMRI data. Preprocessed rsfMRI data from 30 patients with 10 sessions of rsfMRI scans taken within 1 month were obtained from the Consortium for Reliability and Reproducibility. We developed a time-domain measure to capture low-frequency fluctuation (LFF) using a general linear model with three different periodic regressors: boxcar, triangular, and sinusoidal functions. Then, test-retest reliability for the proposed methods was evaluated using the intraclass correlation (ICC). Our approaches for evaluating LFF from rsfMRI data significantly identified the default mode network areas (corrected P<0.05). The regression model with the sinusoidal basis function produced the most reliable results (ICC=0.6) compared with the boxcar (ICC=0.32) or triangular (ICC=0.34) functions. Taken together, the proposed methods successfully identified the default mode network regions. In addition, our results suggest that new functional metrics aiming to extract LFF components by modeling rsfMRI time-series data might provide a reliable biomarker to identify neurological disorders accompanying abnormal functional activity.