Radiomics of Bone Analysis in Micro-CT Imaging with FDK Reconstruction and Modified Simultaneous Algebraic Reconstruction Technique

摘要

Micro-CT provides high resolution image of small objects that can be used in various applications. One of the significant advantages of micro-CT lies in high sensitivity to bone structure. With micro-CT images, the degradation of bone structure induced by chronic kidney disease (CKD) being associated with gradual bone loss can be quantified. The bone turnover occurs from the failure of the kidneys to regulate bone mineralization. The current methods of quantitative imaging typically use a single region of interest (ROI) that segments the whole trabecular region and obtain bone parameters, which usually are not homogenous across such a large ROI. Here we introduce a novel method of quantifying bone parameters that can be used to determine overall bone health. This method analyzes sequential regions on the trabecular bone with multiple small ROIs and evaluates the gradients of bone parameters across these ROIs. Two C57Bl/6J mice femur groups were prepared: a control and CKD groups. All femurs were scanned with a Micro-CT system using tube voltage of 60 kV and current of 0.667 mA. Femur volumes were reconstructed with the Feldkamp-Davis-Kress algorithm and were imported into MicroView to perform bone analysis. Six different sequential ROIs were selected at different distances from the growth plate (0.5mm increments). The gradients of bone parameters along the ROI distance for the control and CKD group were compared. Significant differences were found between two groups in the gradients of bone volume density (P = 0.0002), connective density ( P = 0.0003), trabecular spacing (P = 0.001), and trabecular number (P = 0.01). As a result, the gradient method identified a significant change in several parameters representing a novel and biologically significant strategy.;The bone parameters obtained via two methods were used as features in Gaussian Naive Bayes (GNB) classifier. In order to validate the parameters prior to be successfully used in the classifier, the receiver operating characteristic (ROC) curves were plotted for every bone parameters. The area under the curves (AUCs) above 0.9 were observed in bone volume density, connective density, trabecular spacing and trabecular number acquired by the gradient method. Corresponding parameters were used as features to train and test the classifier. The prediction accuracy of the GNB classifier was in a range of 97~100% when the classifier was trained with 40% of the sample set (200 samples). However, the classification accuracy dropped to 78~85% when the classifier was trained and tested with the parameters obtained by single ROI method. Especially, larger prediction accuracy gap between using the gradient and single ROI parameters was observed in the case of the classifier being trained with 10% of the sample set. The classifier attained 95~98% of accuracy with the gradient parameters while 63~77% of accuracy with the single ROI parameters.;The bone study was conducted on 3D volume reconstructed by FDK reconstruction. To evaluate the effects of alternative reconstruction methods on the bone parameters, simultaneous algebraic reconstruction technique (SART) was investigated and modified. The traditional SART demands large number of iterations due to its lack of regularization term in the method. To overcome such limitation of the SART, conventional FDK was implemented to the SART. The FDK-SART have shown considerably higher SNR in the reconstructed data of Shepp-Logan phantom (SNR1: 24.8, SNR2: 12.2, SNR3: 63.9) compared to the standard SART and FDK. Further, it has shown its capability of reconstructing a complex trabecular structures with sharp edges.