Carpal tunnel syndrome (CTS) is the most common peripheral neuropathy and is characterized by median nerve entrapment at the wrist and the resulting median nerve dysfunction. CTS is diagnosed clinically as the gold standard and confirmed with nerve conduction studies (NCS). Complementing NCS, ultrasound imaging could provide additional anatomical information on pathological and motion changes of the median nerve. The purpose of this study was to estimate the transverse sliding patterns of the median nerve during finger movements by analyzing ultrasound dynamic images to distinguish between normal subjects and CTS patients. Transverse ultrasound images were acquired, and a speckle-tracking algorithm was used to determine the lateral displacements of the median nerve in radial-ulnar plane in B-mode images utilizing the multilevel block-sum pyramid algorithm and averaging. All of the averaged lateral displacements at separate acquisition times within a single flexion-extension cycle were accumulated to obtain the cumulative lateral displacements, which were curve-fitted with a second-order polynomial function. The fitted curve was regarded as the transverse sliding pattern of the median nerve. The R2 value, curvature, and amplitude of the fitted curves were computed to evaluate the goodness, variation and maximum value of the fit, respectively. Box plots, the receiver operating characteristic (ROC) curve, and a fuzzy c-means clustering algorithm were utilized for statistical analysis. The transverse sliding of the median nerve during finger movements was greater and had a steeper fitted curve in the normal subjects than in the patients with mild or severe CTS. The temporal changes in transverse sliding of the median nerve within the carpal tunnel were found to be correlated with the presence of CTS and its severity. The representative transverse sliding patterns of the median nerve during finger movements were demonstrated to be useful for quantitatively estimating median nerve dysfunction in CTS patients.