Opportunistic osteoporosis screening in multi-detector CT images via local classification of textures

Abstract

Summary Our study proposed an automatic pipeline for opportunistic osteoporosis screening using 3D texture features and regional vBMD using multi-detector CT images. A combination of different local and global texture features outperformed the global vBMD and showed high discriminative power to identify patients with vertebral fractures. Introduction Many patients at risk for osteoporosis undergo computed tomography (CT) scans, usable for opportunistic (non-dedicated) screening. We compared the performance of global volumetric bone mineral density (vBMD) with a random forest classifier based on regional vBMD and 3D texture features to separate patients with and without osteoporotic fractures. Methods In total, 154 patients (mean age 64 ± 8.5, male; n = 103) were included in this retrospective single-center analysis, who underwent contrast-enhanced CT for other reasons than osteoporosis screening. Patients were dichotomized regarding prevalent vertebral osteoporotic fractures (noFX, n = 101; FX, n = 53). Vertebral bodies were automatically segmented, and trabecular vBMD was calculated with a dedicated phantom. For 3D texture analysis, we extracted gray-level co-occurrence matrix Haralick features (HAR), histogram of gradients (HoG), local binary patterns (LBP), and wavelets (WL). Fractured vertebrae were excluded for texture-feature and vBMD data extraction. The performance to identify patients with prevalent osteoporotic vertebral fractures was evaluated in a fourfold cross-validation. Results The random forest classifier showed a high discriminatory power (AUC = 0.88). Parameters of all vertebral levels significantly contributed to this classification. Importantly, the AUC of the proposed algorithm was significantly higher than that of volumetric global BMD alone (AUC = 0.64). Conclusion The presented classifier combining 3D texture features and regional vBMD including the complete thoracolumbar spine showed high discriminatory power to identify patients with vertebral fractures and had a better diagnostic performance than vBMD alone. Electronic supplementary material The online version of this article (10.1007/s00198-019-04910-1) contains supplementary material, which is available to authorized users.

Radiomics for classification of bone mineral loss: A machine learning study

“…Radiomics is commonly used in the context of clinical oncology, yet the number of studies using this approach focusing on bone diseases is limited. However, a considerable body of literature has been published on the use of textural features extracted from radiography (digital or conventional), computed tomography (CT) and magnetic resonance imaging (MRI) for osteoporosis detection, diagnosis, assessment and automatic bone disorder classification [10][11][12][13][14][15][16]. Yet, reports on the use of BMD radiomics for bone disease diagnosis and prognosis is still lacking.…”

Detection of Vertebral Fractures in CT Using 3D Convolutional Neural Networks

“…Histogram of Oriented Gradients,...) and volumetric Bone Mineral Density (vBMD) to finally apply a Random Forest classifier for patient-level fracture detection. Their experimental results show that combining multiple features calculated for each vertebra along the spine yields superior results [12]. Bar et al does not first segment the spine before extracting features, but uses a Convolutional Neural Network (CNN) to directly map input images to output fracture classes.…”

Local Bone Mineral Density, Subcutaneous and Visceral Adipose Tissue Measurements in Routine Multi Detector Computed Tomography—Which Parameter Predicts Incident Vertebral Fractures Best?

“…The strength of our study lies in the level-dependent analysis of BMD distribution with regard to prediction of prospective incidental fractures, which has not been investigated before and has the possibility for implementation in clinical routine. Although, there have been approaches to identify patients at the risk for vertebral fractures with texture analysis derived from MDCT scans, so far the impact of regional BMD heterogeneities in this context remains an issue of high scientific and clinical interest [ 27 , 28 ]. The presented results show the potential of regional BMD measurements and highlight the clinical applicability in routine MDCT scans with immediate implications for patient care.…”

“…Odds ratios with 95% Confidence Interval (CI) as well as receiver operating characteristics (ROC) analyses were performed and the area under the curve (AUC) and its standard error (SE) were calculated to evaluate the diagnostic performance of BMD, SAT, VAT, and VAT/SAT ratio to differentiate patients with incident fractures from controls. The analyses have been done similar to Muehlematter et al and Valentinitsch et al [ 27 , 28 ]. Furthermore, multivariate logistic regression models were used to determine significant predictors of incident fractures.…”