普通CBCT评价上颌窦外提升后新生骨量 可靠性的初步研究：与Micro-CT真实微结构 参数的相关性分析

A Preliminary Study on the Reliability of Conventional Cone-Beam Computed Tomography in Evaluating New Bone Volume after Maxillary Sinus Floor Elevation: Correlation Analysis with True Microstructural Parameters Derived from Micro-CT

摘要: 为探讨普通锥形束计算机断层扫描(cone beam computed tomography, CBCT)在评价上颌窦外提升术后新生骨量及骨微结构方面的可靠性，并以micro-CT测得的真实骨微结构参数作为参照，分析普通CBCT影像参数与真实骨量之间的相关性。通过纳入接受上颌窦外提升术并延期种植的病例资料。术后种植位点于种植体植入同期采集骨芯样本，并进行micro-CT扫描，获得骨组织体积(TV)、骨体积(BV)、骨体积分数(BV/TV)、骨小梁数量(Tb.N)、骨小梁厚度(Tb.Th)及骨小梁间隙(Tb.Sp)等真实微结构参数。同时，基于术前或术后CBCT数据，在对应种植位点建立感兴趣区域(region of interest, ROI)或感兴趣体积(volume of interest, VOI)，测量CBCT HU值及相应骨微结构参数。采用Spearman秩相关分析评价CBCT相关参数与micro-CT真实微结构参数之间的相关性。根据数据结果显示，术前黏膜厚度、剩余骨高度、上颌窦宽度及提升高度等影像学变量与micro-CT测得的骨微结构参数均无显著相关性。CBCT不同方向ROI的HU值与micro-CT微结构参数之间整体缺乏稳定相关性，仅横断位HU值与Tb.Th呈弱负相关，但其相关方向与骨质量评价的理论预期不一致。描述性统计结果显示，CBCT测得的BV、BV/TV、Tb.Th及Tb.Sp等参数明显高于micro-CT测的值，而Tb.N明显低于micro-CT测的值，提示CBCT在骨微结构重建中存在系统性偏差。进一步Spearman相关分析显示，CBCT与micro-CT测得的TV、BV、BV/TV、Tb.N、Tb.Th、Tb.Sp之间均无显著相关性。普通CBCT在未进行骨替代材料分割或误差校正的情况下，不能准确反映上颌窦外提升术后真实新生骨量及骨微结构情况。CBCT分辨率较低、体素较大以及部分容积效应可能导致其HU值及微结构参数偏离真实骨质情况，尤其容易受到残留骨替代材料的干扰。后续研究应在不增加辐射剂量的前提下，通过灰度阈值分割、图像重建优化、人工智能算法或多参数校正模型提高CBCT对新生骨量变化的真实反映能力。

Abstract: The reliability of conventional cone-beam computed tomography (CBCT) in assessing new bone volume and microstructure following lateral sinus floor elevation (LSFE), using micro-computed tomography (micro-CT)-derived true bone microstructural parameters as the reference standard, was investigated. Clinical cases undergoing LSFE with delayed implant placement were enrolled. At the time of implant placement, bone core samples were harvested from the implant site and scanned using micro-CT to obtain true microstructural parameters, including total volume (TV), bone volume (BV), bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp). Concurrently, preoperative or postoperative conventional CBCT datasets were used to define regions of interest (ROI) or volumes of interest (VOI) at corresponding implant sites, and Hounsfield unit (HU) values and derived bone microstructural parameters were measured. Spearman rank correlation analysis was performed to assess associations between CBCT-derived parameters and micro-CT—derived true microstructural parameters. Results showed that preoperative imaging variables—including Schneiderian membrane thickness, residual alveolar ridge height, maxillary sinus width, and elevation height—exhibited no significant correlations with any micro-CT—derived microstructural parameters. Overall, HU values from CBCT ROIs across different anatomical planes demonstrated unstable correlations with micro-CT parameters; only the transverse-plane HU value showed a weak negative correlation with Tb.Th—contradicting theoretical expectations for bone quality assessment. Descriptive statistics revealed that CBCT-derived BV, BV/TV, Tb.Th, and Tb.Sp were consistently overestimated, whereas Tb.N was significantly underestimated compared with micro-CT measurements—indicating systematic bias in CBCT-based microstructural reconstruction. Further Spearman correlation analyses confirmed no statistically significant correlations between CBCT- and micro-CT—derived TV, BV, BV/TV, Tb.N, Tb.Th or Tb.Sp. Thus, conventional CBCT—without segmentation of residual bone graft materials or error correction—fails to accurately reflect true new bone volume and microstructure after LSFE. Limitations such as low spatial resolution, large voxel size, and partial volume effects likely cause CBCT HU values and microstructural parameters to deviate from true bone properties, particularly in the presence of residual bone substitute materials. Future studies should enhance the fidelity of CBCT in capturing new bone changes without increasing radiation dose—through gray-scale threshold segmentation, optimized image reconstruction, artificial intelligence—based algorithms, or multi-parameter calibration models.