摘要： Background: The annular ligament of the human stapes constitutes a compliant connection between the stapes footplate and the peripheral cochlear wall at the oval window. The cross section of the human annular ligament is characterized by a three-layered structure, which resembles a sandwich-shaped composite structure. As accurate and precise descriptions of the middle-ear behavior are constrained by lack of information on the complex geometry of the annular ligament, this study aims to obtain comprehensive geometrical data of the annular ligament via multiphoton imaging. Methods: The region of interest containing the stapes and annular ligament was harvested from a fresh-frozen human temporal bone of a 46-years old female. Multiphoton imaging of the unstained sample was performed by detecting the second-harmonic generation of collagen and the autofluorescence of elastin, which are constituents of the annular ligament. The multiphoton scans were conducted on the middle-ear side and cochlear side of the annular ligament to obtain accurate images of the face layers on both sides. The face layers of the annular ligament were manually segmented on both multiphoton scans, and then registered to high-resolution mCT images. Results: Multiphoton scans of the annular ligament revealed 1) relatively large thickness of the core layer compared to the face layers, 2) asymmetric geometry of the face layers between the middle-ear side and cochlear side, and variation of their thickness and width along the footplate boundary, 3) divergent relative alignment of the two face layers, and 4) different fiber composition of the face layers along the boundary with a collagen-reinforcement near the anterior pole on the middle-ear side. Conclusion and outlook: Multiphoton microscopy is a feasible approach to obtain the detailed three-dimensional features of the human stapedial annular ligament along its full boundary. The detailed description of the sandwich-shaped structures of the annular ligament is expected to contribute to modeling of the human middle ear for precise simulation of middle-ear behavior. Further, established methodology in this study may be applicable to imaging of other middle-ear structures.