作者: Ge Wang , Margaret W. Skinner , Michael W. Vannier
DOI: 10.1016/S1076-6332(05)80069-8
关键词: Image processing 、 Artificial intelligence 、 Voxel 、 Spiral 、 Computer vision 、 Point spread function 、 Wiener filter 、 Deblurring 、 Image quality 、 Computer science 、 Deconvolution
摘要: Rationale and Objectives. We developed a method for volumetric image deblurring in spiral (helical) computed tomography (CT) scanning with three-dimensional (3D) Gaussian point spread function (PSF) to improve the quality of temporal bone CT images assessing position cochlear implants electrodes. Methods. A patient was scanned after implantation, reconstructed into volume 128 voxels per dimension, 0.1 mm voxel side, ×10 gray-scale expansion. The 3D PSF imaging assumed be separable transversely longitudinally. Standard deviations were derived subjectively adjusted. then deconvolved using Wiener filtering maximum-likelihood deconvolution methods. Image assessed both visually quantitatively cross-sectional area at half maximum (CAHM) implanted array as figure merit. Results. Substantial achieved via deconvolution. Subjectively, anatomic structures more clearly shown. Deconvolution reduced CAHM by approximately one third, on average. Three-dimensional had better than two-dimensional produced superior but took longer process relative filtering. Conclusion. Volumetric is practical PSF. preferred if time permits. facilitates study fine details implant.