ABSTRACT
Background: Osteoporosis, a disease stemming from bone metabolism irregularities, affects approximately 200 million people worldwide. Timely detection of osteoporosis is pivotal in grappling with this public health challenge. Deep learning (DL), emerging as a promising methodology in the field of medical imaging, holds considerable potential for the assessment of bone mineral density (BMD). This study aimed to propose an automated DL framework for BMD assessment that integrates localization, segmentation, and ternary classification using various dominant convolutional neural networks (CNNs). Methods: In this retrospective study, a cohort of 2,274 patients underwent chest computed tomography (CT) was enrolled from January 2022 to June 2023 for the development of the integrated DL system. The study unfolded in 2 phases. Initially, 1,025 patients were selected based on specific criteria to develop an automated segmentation model, utilizing 2 VB-Net networks. Subsequently, a distinct cohort of 902 patients was employed for the development and testing of classification models for BMD assessment. Then, 3 distinct DL network architectures, specifically DenseNet, ResNet-18, and ResNet-50, were applied to formulate the 3-classification BMD assessment model. The performance of both phases was evaluated using an independent test set consisting of 347 individuals. Segmentation performance was evaluated using the Dice similarity coefficient; classification performance was appraised using the receiver operating characteristic (ROC) curve. Furthermore, metrics such as the area under the curve (AUC), accuracy, and precision were meticulously calculated. Results: In the first stage, the automatic segmentation model demonstrated excellent segmentation performance, with mean Dice surpassing 0.93 in the independent test set. In the second stage, both the DenseNet and ResNet-18 demonstrated excellent diagnostic performance in detecting bone status. For osteoporosis, and osteopenia, the AUCs were as follows: DenseNet achieved 0.94 [95% confidence interval (CI): 0.91-0.97], and 0.91 (95% CI: 0.87-0.94), respectively; ResNet-18 attained 0.96 (95% CI: 0.92-0.98), and 0.91 (95% CI: 0.87-0.94), respectively. However, the ResNet-50 model exhibited suboptimal diagnostic performance for osteopenia, with an AUC value of only 0.76 (95% CI: 0.69-0.80). Alterations in tube voltage had a more pronounced impact on the performance of the DenseNet. In the independent test set with tube voltage at 100 kVp images, the accuracy and precision of DenseNet decreased on average by approximately 14.29% and 18.82%, respectively, whereas the accuracy and precision of ResNet-18 decreased by about 8.33% and 7.14%, respectively. Conclusions: The state-of-the-art DL framework model offers an effective and efficient approach for opportunistic osteoporosis screening using chest CT, without incurring additional costs or radiation exposure.
ABSTRACT
Background: Many patients with malignant tumors require chemotherapy and radiation therapy, which can result in a decline in physical function and potentially influence bone mineral density (BMD). Furthermore, these treatments necessitate enhanced computed tomography (CT) scans for determining disease staging or treatment outcomes, and opportunistic screening with available imaging data is beneficial for patients at high risk for osteoporosis if existing imaging data can be used. The study aimed to investigate the feasibility of opportunistic screening for osteoporosis using enhanced CT based on a dual-energy CT (DECT) material decomposition technique. Methods: We prospectively enrolled 346 consecutive patients who underwent abdominal unenhanced and triphasic contrast-enhanced CT (arterial, portal venous, and delayed phases) between June 2021 and June 2022. The BMD, and the density of hydroxyapatite (HAP) on HAP-iodine images and calcium (Ca) on Ca-iodine images were measured on the L1-L3 vertebral bodies. The iodine intake was recorded. Pearson analysis was conducted to assess the correlation between iodine intake and the density values in three phases and the correlation between BMD and the densities of HAP and Ca. Furthermore, linear regression was employed for quantitative evaluation. Bland-Altman analysis was used to evaluate the agreement between calculated BMD derived from DECT (BMD-DECT) and reference BMD derived from quantitative CT (BMD-QCT). Receiver operating characteristic (ROC) analysis was applied to assess the diagnostic efficacy. Results: The HAP and Ca density of the L1-L3 vertebral bodies did not differ significantly among the three phases of contrast-enhanced CT (F=0.001-0.049; P>0.05). Significant positive correlations were found between HAP, Ca densities, and BMD (HAP-BMD: r=0.9472, R2=0.8973; Ca-BMD: r=0.9470, R2=0.8968; all P<0.001). Bland-Altman plots showed high agreement between BMD-DECT and BMD-QCT. The area under the curve (AUC) using HAP and Ca measurements was 0.963 [95% confidence interval (CI): 0.937-0.980] and 0.964 (95% CI: 0.939-0.981), respectively, for diagnosing osteoporosis and was 0.951 (95% CI: 0.917-0.973) and 0.950 (95% CI: 0.916-0.973), respectively, for diagnosing osteopenia. Conclusions: The HAP and Ca density measurements generated through the material decomposition technique in DECT have good diagnostic performances in assessing BMD, which offers a new perspective for opportunistic screening of osteoporosis on contrast-enhanced CT.
ABSTRACT
Background: The early detection and treatment of osteoporosis can help prevent osteoporosis-related fractures, especially in patients who undergo enhanced computed tomography (CT) scans for disease diagnosis or evaluation of treatment outcomes. Although Hounsfield unit (HU) measurement of the vertebral body has been shown to have a strong positive correlation with bone mineral density (BMD), the contrast media will impact the CT value of the vertebral body and decrease the accuracy. This study is aimed to examine the distinctions in vertebral body CT attenuation measurement on true unenhanced (TUE) and virtual unenhanced (VUE) images generated from triphasic enhanced dual-energy CT (DECT) scans and to determine the feasibility of assessing BMD and detecting osteoporosis on VUE images as compared to quantitative CT (QCT). Methods: A total of 235 patients underwent abdominal CT examinations that included unenhanced (with 120 kVp and Smart mA) and triphasic enhanced DECT scans. The BMD and CT attenuation values of the L1-L2 vertebrae were measured on TUE and VUE images reconstructed from the triphasic enhanced CT. The differences and associations between TUE and VUE generated from triphasic enhanced CT were analyzed. The diagnostic performances of HU measurements obtained from TUE and VUE images were evaluated using receiver operating characteristic curve. Results: The BMD and HU measurements of the vertebrae showed good interobserver repeatability on both TUE and VUE images (all intercorrelation coefficients >0.92). The CT attenuation values of L1 and L2 and their average value showed no statistically significant difference among the triphasic VUE images (F=0.121, F=0.061, F=0.090; all P values >0.05) but were significantly lower than those obtained from the TUE images. HU measurements in both the TUE and triphasic VUE images, along with the reference BMD derived from QCT, demonstrated a strong positive correlation (rTUE =0.981, rVUEa =0.966, rVUEv =0.962, rVUEd =0.964; all P values <0.05), with excellent diagnostic performance for the diagnoses of osteoporosis and osteopenia (all areas under curve >0.95). The Bland-Altman scatter plot exhibited good agreement, as the deviations between the reference BMD and the calculated BMD were evenly distributed around 0. Conclusions: Although the attenuation values of the vertebrae on the VUE images were underestimated compared to those on the TUE images, the HU measurement on VUE image was effective in assessing BMD and detecting osteoporosis and osteopenia with good diagnostic performance.
ABSTRACT
RATIONALE AND OBJECTIVES: To develop an intelligent diagnostic model for osteoporosis screening based on low-dose chest computed tomography (LDCT). The model incorporates automatic deep-learning thoracic vertebrae of cancellous bone (TVCB) segmentation model and radiomics analysis. MATERIALS AND METHODS: A total of 442 participants who underwent both LDCT and quantitative computed tomography (QCT) examinations were enrolled and were randomly allocated to the training, internal testing, and external testing cohorts. The TVCB automatic segmentation model was trained using VB-Net. The accuracy of the segmentation was evaluated using the Dice coefficient. Predictive models for assessing bone mineral density (BMD) were constructed utilizing radiomics analysis based on automatic segmentation (ASeg model) and manual segmentation (MSeg model), respectively. The BMD predictive model based on ASeg and MSeg included the identification of normal and abnormal BMD (first-level model), and osteopenia and osteoporosis (second-level model). The diagnostic performance of the radiomics models were evaluated using the area under the curve (AUC), sensitivity and specificity. RESULTS: The Dice coefficients of the TVCB segmentation model in the internal and external testing cohorts were found to be 0.988 ± 0.014 and 0.939 ± 0.034, respectively. In the first-level model, the AUC of the ASeg model exhibited comparable performance to that of the MSeg model for both the internal (0.985 vs. 0.946, P = 0.080) and external (0.965 vs. 0.955, P = 0.724) testing cohorts. Similarly, in the second-level model, the AUC of the ASeg model was found to be comparable to that of the MSeg model for both the internal (0.933 vs. 0.920, P = 0.794) and external (0.907 vs. 0.892, P = 0.805) testing cohorts. CONCLUSION: A fully automated pipeline for TVCB segmentation and BMD assessment with radiomics analysis can be used for opportunistic BMD screening in chest LDCT.
