Texture analysis of vertebral bone marrow using chemical shift encoding-based water-fat MRI: a feasibility study

Predicting Osteoporosis and Osteopenia by Fusing Deep Transfer Learning Features and Classical Radiomics Features Based on Single-Source Dual-energy CT Imaging

RATIONALE AND OBJECTIVES

To develop and validate a predictive model for osteoporosis and osteopenia prediction by fusing deep transfer learning (DTL) features and classical radiomics features based on single-source dual-energy computed tomography (CT) virtual monochromatic imaging.

METHODS

A total of 606 lumbar vertebrae with dual-energy CT imaging and quantitative CT (QCT) evaluation were included in the retrospective study and randomly divided into the training (n = 424) and validation (n = 182) cohorts. Radiomics features and DTL features were extracted from 70-keV monochromatic CT images, followed by feature selection and model construction, radiomics and DTL features models were established. Then, we integrated the selected two types of features into a features fusion model. We developed a two-level classifier for the hierarchical pairwise classification of each vertebra. All the vertebrae were first classified into osteoporosis and non-osteoporosis groups, then non-osteoporosis group was classified into osteopenia and normal groups. QCT was used as reference. The predictive performance and clinical usefulness of three models were evaluated and compared.

RESULTS

The area under the curve (AUC) of the features fusion, radiomics and DTL models for the classification between osteoporosis and non-osteoporosis were 0.981, 0.999, 0.997 in the training cohort and 0.979, 0.943, 0.848 in the validation cohort. Furthermore, the AUCs of the previously mentioned models for the differentiation between osteopenia and normal were 0.994, 0.971, 0.996 in the training cohort and 0.990, 0.968, 0.908 in the validation cohort. The overall accuracy of the previously mentioned models for two-level classifications was 0.979, 0.955, 0.908 in the training cohort and 0.918, 0.885, 0.841 in the validation cohort. Decision curve analysis showed that all models had high clinical value.

CONCLUSION

The feature fusion model can be used for osteoporosis and osteopenia prediction with improved predictive ability over a radiomics model or a DTL model alone.

MRI-based radiomics assessment of the imminent new vertebral fracture after vertebral augmentation

BACKGROUND

Imminent new vertebral fracture (NVF) is highly prevalent after vertebral augmentation (VA). An accurate assessment of the imminent risk of NVF could help to develop prompt treatment strategies.

PURPOSE

To develop and validate predictive models that integrated the radiomic features and clinical risk factors based on machine learning algorithms to evaluate the imminent risk of NVF.

MATERIALS AND METHODS

In this retrospective study, a total of 168 patients with painful osteoporotic vertebral compression fractures treated with VA were evaluated. Radiomic features of L1 vertebrae based on lumbar T2-weighted images were obtained. Univariate and LASSO-regression analyses were applied to select the optimal features and construct radiomic signature. The radiomic signature and clinical signature were integrated to develop a predictive model by using machine learning algorithms including LR, RF, SVM, and XGBoost. Receiver operating characteristic curve and calibration curve analyses were used to evaluate the predictive performance of the models.

RESULTS

The radiomic-XGBoost model with the highest AUC of 0.93 of the training cohort and 0.9 of the test cohort among the machine learning algorithms. The combined-XGBoost model with the best performance with an AUC of 0.9 in the training cohort and 0.9 in the test cohort. The radiomic-XGBoost model and combined-XGBoost model achieved better performance to assess the imminent risk of NVF than that of the clinical risk factors alone (p < 0.05).

CONCLUSION

Radiomic and machine learning modeling based on T2W images of preoperative lumbar MRI had an excellent ability to evaluate the imminent risk of NVF after VA.

Opportunistic Evaluation of Trabecular Bone Texture by MRI Reflects Bone Mineral Density and Microarchitecture

CONTEXT

Many individuals at high risk for fracture are never evaluated for osteoporosis and subsequently do not receive necessary treatment. Utilization of magnetic resonance imaging (MRI) is burgeoning, providing an ideal opportunity to use MRI to identify individuals with skeletal deficits. We previously reported that MRI-based bone texture was more heterogeneous in postmenopausal women with a history of fracture compared to controls.

OBJECTIVE

The present study aimed to identify the microstructural characteristics that underlie trabecular texture features.

METHODS

In a prospective cohort, we measured spine volumetric bone mineral density (vBMD) by quantitative computed tomography (QCT), peripheral vBMD and microarchitecture by high-resolution peripheral QCT (HRpQCT), and areal BMD (aBMD) by dual-energy x-ray absorptiometry. Vertebral trabecular bone texture was analyzed using T1-weighted MRIs. A gray level co-occurrence matrix was used to characterize the distribution and spatial organization of voxelar intensities and derive the following texture features: contrast (variability), entropy (disorder), angular second moment (ASM; uniformity), and inverse difference moment (IDM; local homogeneity).

RESULTS

Among 46 patients (mean age 64, 54% women), lower peripheral vBMD and worse trabecular microarchitecture by HRpQCT were associated with greater texture heterogeneity by MRI-higher contrast and entropy (r ∼ -0.3 to 0.4, P < .05), lower ASM and IDM (r ∼ +0.3 to 0.4, P < .05). Lower spine vBMD by QCT was associated with higher contrast and entropy (r ∼ -0.5, P < .001), lower ASM and IDM (r ∼ +0.5, P < .001). Relationships with aBMD were less pronounced.

CONCLUSION

MRI-based measurements of trabecular bone texture relate to vBMD and microarchitecture, suggesting that this method reflects underlying microstructural properties of trabecular bone. Further investigation is required to validate this methodology, which could greatly improve identification of patients with skeletal fragility.

Associations between vertebral bone marrow fat and sagittal spine alignment as assessed by chemical shift-encoding-based water-fat MRI

BACKGROUND

The relationship between sagittal spine alignment and vertebral bone marrow fat is unknown. We aimed to assess the relationship between vertebral bone marrow fat and sagittal spine alignment using chemical shift-encoding-based water-fat magnetic resonance imaging (MRI).

METHODS

A total of 181 asymptomatic volunteers were recruited for whole spine X-ray and lumbar MRI. Spine typing was performed according to the Roussouly classification and measurement of vertebral fat fraction based on the chemical shift-encoding-based water-fat MRI. One-way analysis of variance (ANOVA) was used to analyze the differences in vertebral fat fraction between spine types. The post hoc least significant difference (LSD) test was utilized for subgroup comparison after ANOVA.

RESULTS

Overall, the vertebral fat fraction increased from L1 to L5 and was the same for each spine type. The vertebral fat fraction was the highest in type 1 and lowest in type 4 at all levels. ANOVA revealed statistically significant differences in fat fraction among different spine types at L4 and L5 (P < .05). The post hoc LSD test showed that the fat fraction of L4 was significantly different (P < .05) between type 1 and type 4 as well as between type 2 and type 4. The fat fraction of L5 was significantly different between type 1 and type 3, between type 1 and type 4, and between type 2 and type 4 (P < .05).

CONCLUSION

Our study found that vertebral bone marrow fat is associated with sagittal spine alignment, which may serve as a new additional explanation for the association of sagittal alignment with spinal degeneration.

Features of lumbar spine texture extracted from routine MRI correlate with bone mineral density and can potentially differentiate patients with and without fragility fractures in the spine

The use of routine magnetic resonance imaging (MRI) to potentially assess skeletal fragility has been widely studied in osteoporosis. The aim of this study was to evaluate bone texture attributes (TA) from routine lumbar spine (LS) MRI and their correlation with vertebral fragility fractures (VFF) and bone mineral density (BMD). Sixty-four post-menopausal women were submitted to LS densitometry, total spine radiographs, and routine T2-weighted LS MRI. Twenty-two TA were extracted with the platform IBEX from L3 vertebra. The statistical difference was evaluated using ANOVA and Duncan's post-test. Correlation analyses were performed using Spearman's coefficient. Statistical significance was considered when P<0.05. The results did not show a significant difference in BMD between the women with and without fractures. Two bone TA (cluster tendency and variance) were significantly lower in the fracture group. Cluster tendency with VFF in osteopenia was 1.54±1.37 and in osteoporosis was 1.11±58. Cluster tendency without VFF in osteopenia was 2.23±1.38 and in osteoporosis was 1.88±1.14). Variance with VFF in osteopenia was 1.44±1.37 and in osteoporosis was 1.13±59. Variance without VFF in osteopenia was 2.34±1.38 and in osteoporosis was 1.89±1.14. There was a significant correlation between BMD and cluster prominence (r=0.409), cluster tendency (r=0.345), correlation (r=0.570), entropy (r=0.364), information measure corr1 (r=0.378), inverse variance (r=0.449), sum entropy (r=0.320), variance (r=0.338), sum average (r=-0.274), and sum variance (r=-0.266). Our results demonstrated the potential use of TA extracted from routine MRI as a biomarker to assess osteoporosis and identify the tendency of skeletal fragility vertebral fractures.

Associations of Texture Features of Proton Density Fat Fraction Maps between Lumbar Vertebral Bone Marrow and Paraspinal Musculature

Chemical shift encoding-based water−fat MRI (CSE-MRI)-derived proton density fat fraction (PDFF) has been used for non-invasive assessment of regional body fat distributions. More recently, texture analysis (TA) has been proposed to reveal even more detailed information about the vertebral or muscular composition beyond PDFF. The aim of this study was to investigate associations between vertebral bone marrow and paraspinal muscle texture features derived from CSE-MRI-based PDFF maps in a cohort of healthy subjects. In this study, 44 healthy subjects (13 males, 55 ± 30 years; 31 females, 39 ± 17 years) underwent 3T MRI including a six-echo three-dimensional (3D) spoiled gradient echo sequence used for CSE-MRI at the lumbar spine and the paraspinal musculature. The erector spinae muscles (ES), the psoas muscles (PS), and the vertebral bodies L1-4 (LS) were manually segmented. Mean PDFF values and texture features were extracted for each compartment. Features were compared between males and females using logistic regression analysis adjusted for age and body mass index (BMI). All texture features of ES except for Sum Average were significantly (p < 0.05) different between men and women. The three global texture features (Variance, Skewness, Kurtosis) for PS as well as LS showed a significant difference between male and female subjects (p < 0.05). Mean PDFF measured in PS and ES was significantly higher in females, but no difference was found for the vertebral bone marrow’s PDFF. Partial correlation analysis between the texture features of the spine and the paraspinal muscles revealed a highly significant correlation for Variance(global) (r = 0.61 for ES, r = 0.62 for PS; p < 0.001 respectively). Texture analysis using PDFF maps based on CSE-MRI revealed differences between healthy male and female subjects. Global texture features in the lumbar vertebral bone marrow allowed for differentiation between men and women, when the overall PDFF was not significantly different, indicating that PDFF maps may contain detailed and subtle textural information beyond fat fraction. The observed significant correlation of Variance(global) suggests a metabolic interrelationship between vertebral bone marrow and the paraspinal muscles.

Development and validation of a machine learning-derived radiomics model for diagnosis of osteoporosis and osteopenia using quantitative computed tomography

Background

To develop and validate a quantitative computed tomography (QCT) based radiomics model for discriminating osteoporosis and osteopenia.

Methods

A total of 635 patients underwent QCT were retrospectively included from November 2016 to November 2019. The patients with osteopenia or osteoporosis (N = 590) were divided into a training cohort (N = 414) and a test cohort (N = 176). Radiomics features were extracted from the QCT images of the third lumbar vertebra. Minimum redundancy and maximum relevance and least absolute shrinkage and selection operator were used for data dimensional reduction, features selection and radiomics model building. Multivariable logistic regression was applied to construct the combined clinical-radiomic model that incorporated radiomics signatures and clinical characteristics. The performance of the combined clinical-radiomic model was evaluated by the area under the curve of receiver operator characteristic curve (ROC–AUC), accuracy, specificity, sensitivity, positive predictive value, and negative predictive value.

Results

The patients with osteopenia or osteoporosis were randomly divided into training and test cohort with a ratio of 7:3. Six more predictive radiomics signatures, age, alkaline phosphatase and homocysteine were selected to construct the combined clinical-radiomic model for diagnosis of osteoporosis and osteopenia. The AUC of the combined clinical-radiomic model was 0.96 (95% confidence interval (CI), 0.95 to 0.98) in the training cohort and 0.96 (95% CI 0.92 to 1.00) in the test cohort, which were superior to the clinical model alone (training-AUC = 0.81, test-AUC = 0.79). The calibration curve demonstrated that the radiomics nomogram had good agreement between prediction and observation and decision curve analysis confirmed clinically useful.

Conclusions

The combined clinical-radiomic model that incorporates the radiomics score and clinical risk factors, can serve as a reliable and powerful tool for discriminating osteoporosis and osteopenia.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-022-00868-5.

Predicting Lumbar Vertebral Osteopenia Using LvOPI Scores and Logistic Regression Models in an Exploratory Study of Premenopausal Taiwanese Women

Purpose

To propose hybrid predicting models integrating clinical and magnetic resonance imaging (MRI) features to diagnose lumbar vertebral osteopenia (LvOPI) in premenopausal women.

Methods

This prospective study enrolled 101 Taiwanese women, including 53 before and 48 women after menopause. Clinical information, including age, body height, body weight and body mass index (BMI), were recorded. Bone mineral density (BMD) was measured by the dual-energy X-ray absorptiometry. Lumbar vertebral fat fraction (LvFF) was measured by MRI. LvOPI scores (LvOPISs) comprising different clinical features and LvFF were constructed to diagnose LvOPI. Statistical analyses included normality tests, linear regression analyses, logistic regression analyses, group comparisons, and diagnostic performance. A P value less than 0.05 was considered as statistically significant.

Results

The post-menopausal women had higher age, body weight, BMI, LvFF and lower BMD than the pre-menopausal women (all P < 0.05). The lumbar vertebral osteoporosis group had significantly higher age, longer MMI, and higher LvFF than the LvOPI group (all P < 0.05) and normal group (all P < 0.005). LvOPISs (AUC, 0.843 to 0.864) outperformed body weight (0.747; P = 0.0566), BMI (0.737; P < 0.05), age (0.649; P < 0.05), and body height (0.5; P < 0.05) in diagnosing LvOPI in the premenopausal women. Hybrid predicting models using logistic regression analysis (0.894 to 0.9) further outperformed all single predictors in diagnosing LvOPI in the premenopausal women (P < 0.05).

Conclusion

The diagnostic accuracy of the LvOPI can be improved by using our proposed hybrid predicting models in Taiwanese premenopausal women.

Novel Radiomics-Clinical Model for the Noninvasive Prediction of New Fractures After Vertebral Augmentation

PURPOSE

To investigate the noninvasive prediction model for new fractures after percutaneous vertebral augmentation (PVA) based on radiomics signature and clinical parameters.

METHODS

Data from patients who were diagnosed with osteoporotic vertebral compression fracture (OVCF) and treated with PVA in our hospital between May 2014 and April 2019 were retrospectively analyzed. Radiomics features were extracted from T1-weighted magnetic resonance imaging (MRI) of the T11-L5 segments taken before PVA. Different radiomics models was developed by using linear discriminant analysis (LDA), multilayer perceptron (MLP), and stochastic gradient descent (SGD) classifiers. A nomogram was constructed by integrating clinical parameters and Radscore that calculated by the best radiomics model. The model performance was quantified in terms of discrimination, calibration and clinical usefulness.

RESULT

Four clinical parameters and 16 selected radiomics features were used for model development. The clinical model showed poor discrimination capability with area under the curves (AUCs) yielding of 0.522 in the training dataset and 0.517 in the validation dataset. The LDA, MLP and SGD classifier-based radiomics model had achieved AUCs of 0.793, 0.810, and 0.797 in the training dataset, and 0.719, 0.704, and 0.725 in the validation dataset, respectively. The nomogram showed the best performance with AUCs achieving 0.810 and 0.754 in the training and validation datasets, respectively. The decision curve analysis demonstrated the net benefit of the nomogram was higher than that of other models.

CONCLUSION

Our findings indicate that combining clinical features with radiomics features from pre-augmentation T1-weighted MRI can be used to develop a nomogram that can predict new fractures in patients after PVA.

Imaging of the Osteoporotic Spine - Quantitative Approaches in Diagnostics and for the Prediction of the Individual Fracture Risk

Osteoporosis is a highly prevalent systemic skeletal disease that is characterized by low bone mass and microarchitectural bone deterioration. It predisposes to fragility fractures that can occur at various sites of the skeleton, but vertebral fractures (VFs) have been shown to be particularly common. Prevention strategies and timely intervention depend on reliable diagnosis and prediction of the individual fracture risk, and dual-energy X-ray absorptiometry (DXA) has been the reference standard for decades. Yet, DXA has its inherent limitations, and other techniques have shown potential as viable add-on or even stand-alone options. Specifically, three-dimensional (3 D) imaging modalities, such as computed tomography (CT) and magnetic resonance imaging (MRI), are playing an increasing role. For CT, recent advances in medical image analysis now allow automatic vertebral segmentation and value extraction from single vertebral bodies using a deep-learning-based architecture that can be implemented in clinical practice. Regarding MRI, a variety of methods have been developed over recent years, including magnetic resonance spectroscopy (MRS) and chemical shift encoding-based water-fat MRI (CSE-MRI) that enable the extraction of a vertebral body's proton density fat fraction (PDFF) as a promising surrogate biomarker of bone health. Yet, imaging data from CT or MRI may be more efficiently used when combined with advanced analysis techniques such as texture analysis (TA; to provide spatially resolved assessments of vertebral body composition) or finite element analysis (FEA; to provide estimates of bone strength) to further improve fracture prediction. However, distinct and experimentally validated diagnostic criteria for osteoporosis based on CT- and MRI-derived measures have not yet been achieved, limiting broad transfer to clinical practice for these novel approaches. KEY POINTS::   · DXA is the reference standard for diagnosis and fracture prediction in osteoporosis, but it has important limitations.. · CT- and MRI-based methods are increasingly used as (opportunistic) approaches.. · For CT, particularly deep-learning-based automatic vertebral segmentation and value extraction seem promising.. · For MRI, multiple techniques including spectroscopy and chemical shift imaging are available to extract fat fractions.. · Texture and finite element analyses can provide additional measures for vertebral body composition and bone strength.. CITATION FORMAT: · Sollmann N, Kirschke JS, Kronthaler S et al. Imaging of the Osteoporotic Spine - Quantitative Approaches in Diagnostics and for the Prediction of the Individual Fracture Risk. Fortschr Röntgenstr 2022; DOI: 10.1055/a-1770-4626.

Radiomic Modeling of Bone Density and Rib Fracture Risk After Stereotactic Body Radiation Therapy for Early-Stage Non-Small Cell Lung Cancer

Purpose

Our purpose was to determine whether bone density and bone-derived radiomic metrics in combination with dosimetric variables could improve risk stratification of rib fractures after stereotactic body radiation therapy (SBRT) for early-stage non-small cell lung cancer (NSCLC).

Methods and Materials

A retrospective analysis was conducted of patients with early-stage NSCLC treated with SBRT. Dosimetric data and rib radiomic data extracted using PyRadiomics were used for the analysis. A subset of patients had bone density scans that were used to create a predicted bone density score for all patients. A 10-fold cross validated approach with 10 resamples was used to find the top univariate logistic models and elastic net regression models that predicted for rib fracture.

Results

A total of 192 treatment plans were included in the study with a rib fracture rate of 16.1%. A predicted bone density score was created from a multivariate model with vertebral body Hounsfield units and patient weight, with an R-squared of 0.518 compared with patient dual-energy x-ray absorptiometry T-scores. When analyzing all patients, a low predicted bone density score approached significance for increased risk of rib fracture (P = .07). On competing risk analysis, when stratifying patients based on chest wall V30 Gy and bone density score, those with a V30 Gy ≥30 cc and a low bone density score had a significantly higher risk of rib fracture compared with all other patients (P < .001), with a predicted 2-year risk of rib fracture of 28.6% (95% confidence interval, 17.2%-41.1%) and 4.9% (95% confidence interval, 2.3%-9.0%), respectively. Dosimetric variables were the primary drivers of fracture risk. A multivariate elastic net regression model including all dosimetric variables was the best predictor of rib fracture (area under the curve [AUC], 0.864). Bone density variables (AUC, 0.618) and radiomic variables (AUC, 0.617) have better predictive power than clinical variables that exclude bone density (AUC, 0.538).

Conclusion

Radiomic features, including a bone density score that includes vertebral body Hounsfield units and radiomic signatures from the ribs, can be used to stratify risk of rib fracture after SBRT for NSCLC.

Using radiomic features of lumbar spine CT images to differentiate osteoporosis from normal bone density

OBJECTIVE

This study aimed to develop a predictive model to detect osteoporosis using radiomic features from lumbar spine computed tomography (CT) images.

METHODS

A total of 133 patients were included in this retrospective study, 41 men and 92 women, with a mean age of 65.45 ± 9.82 years (range: 31-94 years); 53 had normal bone mineral density, 32 osteopenia, and 48 osteoporosis. For each patient, the L1-L4 vertebrae on the CT images were automatically segmented using SenseCare and defined as regions of interest (ROIs). In total, 1,197 radiomic features were extracted from these ROIs using PyRadiomics. The most significant features were selected using logistic regression and Pearson correlation coefficient matrices. Using these features, we constructed three linear classification models based on the random forest (RF), support vector machine (SVM), and K-nearest neighbor (KNN) algorithms, respectively. The training and test sets were repeatedly selected using fivefold cross-validation. The model performance was evaluated using the area under the receiver operator characteristic curve (AUC) and confusion matrix.

RESULTS

The classification model based on RF had the highest performance, with an AUC of 0.994 (95% confidence interval [CI]: 0.979-1.00) for differentiating normal BMD and osteoporosis, 0.866 (95% CI: 0.779-0.954) for osteopenia versus osteoporosis, and 0.940 (95% CI: 0.891-0.989) for normal BMD versus osteopenia.

CONCLUSIONS

The excellent performance of this radiomic model indicates that lumbar spine CT images can effectively be used to identify osteoporosis and as a tool for opportunistic osteoporosis screening.

Fully automated segmentation of lumbar bone marrow in sagittal, high-resolution T1-weighted magnetic resonance images using 2D U-NET

BACKGROUND

We investigated a 2-dimensional (2D) U-Net model to delineate lumbar bone marrow (BM) using a high resolution T1-weighted magnetic resonance imaging.

METHOD

Healthy controls (n = 44, 836 images) and patients with hematologic diseases (n = 56, 1064 images) received MRI of the lumbar spines. Lumbar BM on each image was manually delineated by an experienced radiologist as a ground-truth. The 2D U-Net models were trained using a healthy lumbar BM only, diseased BM only, and using healthy and diseased BM combined, respectively. The models were validated using healthy and diseased subjects, separately. A repeated-measures analysis of variance was performed to compare segmentation accuracies with 2 validation cohorts among U-Net trained with healthy subjects (UNET_HC), U-Net trained with diseased subjects (UNET_HD), U-Net trained with all subjects including both healthy and diseased subjects (UNET_HCHD), and 3-dimensional Grow-Cut algorithm (3DGC).

RESULTS

When validated with the healthy subjects, UNET_HC, UNET_HD, UNET_HCHD and 3DGC achieved the mean and standard deviation of the Dice Similarity Coefficient (DSC) of 0.9415 ± 0.07056, 0.9583 ± 0.05146, 0.9602 ± 0.0486 and 0.9139 ± 0.2039, respectively. When validated with the diseased subjects, DSCs of UNET_HC, UNET_HD, UNET_HCHD and 3DGC were 0.8303 ± 0.1073, 0.9502 ± 0.0217, 0.9502 ± 0.0217 and 0.8886 ± 0.2179, respectively. The U-Net models segmented BM better than the semi-automatic 3DGC (P < 0.0001), and UNET_HD produced better results than UNET_HC (P < 0.0001).

CONCLUSIONS

We successfully constructed a fully automatic lumbar BM segmentation model for a high-resolution T1-weighted MRI using U-Net, which outperformed most of the previously reported approaches and the existing semi-automatic algorithm.

Radiomics and stacking regression model for measuring bone mineral density using abdominal computed tomography

BACKGROUND

Measurement of bone mineral density (BMD) is the most important method to diagnose osteoporosis. However, current BMD measurement is always performed after a fracture has occurred.

PURPOSE

To explore whether a radiomic model based on abdominal computed tomography (CT) can predict the BMD of lumbar vertebrae.

MATERIAL AND METHODS

A total of 245 patients who underwent both dual-energy X-ray absorptiometry (DXA) and abdominal CT examination (training cohort, n = 196; validation cohort, n = 49) were included in our retrospective study. In total, 1218 image features were extracted from abdominal CT images for each patient. Combined with clinical information, three steps including least absolute shrinkage and selection operator (LASSO) regression were used to select key features. A two-tier stacking regression model with multi-algorithm fusion was used for BMD prediction, which can integrate the advantages of linear model and non-linear model. The prediction results of this model were compared with those using a single regressor. The degree-of-freedom adjusted coefficient of determination (Adjusted-R²), root mean square error (RMSE), and mean absolute error (MAE) were used to evaluate the regression performance.

RESULTS

Compared with other regression methods, the two-tier stacking regression model has a higher regression performance, with Adjusted-R², RMSE, and MAE of 0.830, 0.077, and 0.06, respectively. Pearson correlation analysis and Bland-Altman analysis showed that the BMD predicted by the model had a high correlation with the DXA results (r = 0.932, difference = -0.01 ± 0.1412 mg/cm²).

CONCLUSION

Using radiomics, the BMD of lumbar vertebrae could be predicted from abdominal CT images.

T2 -weighted magnetic resonance imaging texture as predictor of low back pain: A texture analysis-based classification pipeline to symptomatic and asymptomatic cases

Low back pain is a very common symptom and the leading cause of disability throughout the world. Several degenerative imaging findings seen on magnetic resonance imaging are associated with low back pain but none of them is specific for the presence of low back pain as abnormal findings are prevalent among asymptomatic subjects as well. The purpose of this population-based study was to investigate if more specific magnetic resonance imaging predictors of low back pain could be found via texture analysis and machine learning. We used this methodology to classify T₂ -weighted magnetic resonance images from the Northern Finland Birth Cohort 1966 data to symptomatic and asymptomatic groups. Lumbar spine magnetic resonance imaging was performed using a fast spin-echo sequence at 1.5 T. Texture analysis pipeline consisting of textural feature extraction, principal component analysis, and logistic regression classifier was applied to the data to classify them into symptomatic (clinically relevant pain with frequency ≥30 days and intensity ≥6/10) and asymptomatic (frequency ≤7 days, intensity ≤3/10, and no previous pain episodes in the follow-up period) groups. Best classification results were observed applying texture analysis to the two lowest intervertebral discs (L4-L5 and L5-S1), with accuracy of 83%, specificity of 83%, sensitivity of 82%, negative predictive value of 94%, precision of 56%, and receiver operating characteristic area-under-curve of 0.91. To conclude, textural features from T₂ -weighted magnetic resonance images can be applied in low back pain classification.

Association of Cervical and Lumbar Paraspinal Muscle Composition Using Texture Analysis of MR-Based Proton Density Fat Fraction Maps

In this study, the associations of cervical and lumbar paraspinal musculature based on a texture analysis of proton density fat fraction (PDFF) maps were investigated to identify gender- and anatomical location-specific structural patterns. Seventy-nine volunteers (25 men, 54 women) participated in the present study (mean age ± standard deviation: men: 43.7 ± 24.6 years; women: 37.1 ± 14.0 years). Using manual segmentations of the PDFF maps, texture analysis was performed and texture features were extracted. A significant difference in the mean PDFF between men and women was observed in the erector spinae muscle (p < 0.0001), whereas the mean PDFF did not significantly differ in the cervical musculature and the psoas muscle (p > 0.05 each). Among others, Variance(global) and Kurtosis(global) showed significantly higher values in men than in women in all included muscle groups (p < 0.001). Not only the mean PDFF values (p < 0.001) but also Variance(global) (p < 0.001), Energy (p < 0.001), Entropy (p = 0.01), Homogeneity (p < 0.001), and Correlation (p = 0.037) differed significantly between the three muscle compartments. The cervical and lumbar paraspinal musculature composition seems to be gender-specific and has anatomical location-specific structural patterns.

MRI-based Texture Analysis of Trabecular Bone for Opportunistic Screening of Skeletal Fragility

CONTEXT

Many individuals at high risk for osteoporosis and fragility fracture are never screened by traditional methods. Opportunistic use of imaging obtained for other clinical purposes is required to foster identification of these patients.

OBJECTIVE

The aim of this pilot study was to evaluate texture features as a measure of bone fragility, by comparing clinically acquired magnetic resonance imaging (MRI) scans from individuals with and without a history of fragility fracture.

METHODS

This study retrospectively investigated 100 subjects who had lumbar spine MRI performed at our institution. Cases (n = 50) were postmenopausal women with osteoporosis and a confirmed history of fragility fracture. Controls (n = 50) were age- and race-matched postmenopausal women with no known fracture history. Trabecular bone from the lumbar vertebrae was segmented to create regions of interest within which a gray level co-occurrence matrix was used to quantify the distribution and spatial organization of voxel intensity. Heterogeneity in the trabecular bone texture was assessed by several features, including contrast (variability), entropy (disorder), and angular second moment (homogeneity).

RESULTS

Texture analysis revealed that trabecular bone was more heterogeneous in fracture patients. Specifically, fracture patients had greater texture variability (+76% contrast; P = 0.005), greater disorder (+10% entropy; P = 0.005), and less homogeneity (-50% angular second moment; P = 0.005) compared with controls.

CONCLUSIONS

MRI-based textural analysis of trabecular bone discriminated between patients with known osteoporotic fractures and controls. Further investigation is required to validate this promising methodology, which could greatly expand the number of patients screened for skeletal fragility.

Radiomics Based on Lumbar Spine Magnetic Resonance Imaging to Detect Osteoporosis

RATIONALE AND OBJECTIVES

Signal intensity of the lumbar spine in magnetic resonance imaging (MRI) correlates to bone mineral density (BMD). This study aims to explore a lumbar spine magnetic resonance imaging based on the radiomics model for detecting osteoporosis.

MATERIALS AND METHODS

A total of 109 patients, who underwent both dual-energy X-ray absorptiometry (DEXA) and MRI of the lumbar spine, were recruited. Among these patients, 38 patients were normal, 32 patients had osteopenia, and 39 patients had osteoporosis, according to the DEXA results. A total of 396 × 2 radiomic features were extracted from the T1WI and T2WI images of the segmentation images in the lumbar magnetic resonance imaging. The correlated radiomic features were selected to establish the radiomic classification model. Then, the classification models (based on T1WI, T2WI, and T1WI+T2WI) of normal vs. osteopenia, normal vs. osteoporosis, and osteopenia vs. osteoporosis were established. The performance of the classification models was evaluated through the estimated area under the receiver operating characteristic curve.

RESULTS

The area under the receiver operating characteristic curves based on T1WI, T2WI, and T1WI+T2WI were 0.772, 0.772, and 0.810, respectively, for the models of normal vs. osteopenia, 0.724, 0.682, and 0.797, respectively, for the models of normal vs. osteoporosis, and 0.730, 0.734, and 0.769, respectively, for the models of osteopenia vs. osteoporosis.

CONCLUSION

Radiomic models established based on lumbar spine MRI can be used to detect osteoporosis.

Combined radiomics-clinical model to predict malignancy of vertebral compression fractures on CT

OBJECTIVES

To develop and validate a combined radiomics-clinical model to predict malignancy of vertebral compression fractures on CT.

METHODS

One hundred sixty-five patients with vertebral compression fractures were allocated to training (n = 110 [62 acute benign and 48 malignant fractures]) and validation (n = 55 [30 acute benign and 25 malignant fractures]) cohorts. Radiomics features (n = 144) were extracted from non-contrast-enhanced CT images. Radiomics score was constructed by applying least absolute shrinkage and selection operator regression to reproducible features. A combined radiomics-clinical model was constructed by integrating significant clinical parameters with radiomics score using multivariate logistic regression analysis. Model performance was quantified in terms of discrimination and calibration. The model was internally validated on the independent data set.

RESULTS

The combined radiomics-clinical model, composed of two significant clinical predictors (age and history of malignancy) and the radiomics score, showed good calibration (Hosmer-Lemeshow test, p > 0.05) and discrimination in both training (AUC, 0.970) and validation (AUC, 0.948) cohorts. Discrimination performance of the combined model was higher than that of either the radiomics score (AUC, 0.941 in training cohort and 0.852 in validation cohort) or the clinical predictor model (AUC, 0.924 in training cohort and 0.849 in validation cohort). The model stratified patients into groups with low and high risk of malignant fracture with an accuracy of 98.2% in the training cohort and 90.9% in the validation cohort.

CONCLUSIONS

The combined radiomics-clinical model integrating clinical parameters with radiomics score could predict malignancy in vertebral compression fractures on CT with high discriminatory ability.

KEY POINTS

• A combined radiomics-clinical model was constructed to predict malignancy of vertebral compression fractures on CT by combining clinical parameters and radiomics features. • The model showed good calibration and discrimination in both training and validation cohorts. • The model showed high accuracy in the stratification of patients into groups with low and high risk of malignant vertebral compression fractures.

Association of Thigh Muscle Strength with Texture Features Based on Proton Density Fat Fraction Maps Derived from Chemical Shift Encoding-Based Water-Fat MRI

Purpose: Based on conventional and quantitative magnetic resonance imaging (MRI), texture analysis (TA) has shown encouraging results as a biomarker for tissue structure. Chemical shift encoding-based water–fat MRI (CSE-MRI)-derived proton density fat fraction (PDFF) of thigh muscles has been associated with musculoskeletal, metabolic, and neuromuscular disorders and was demonstrated to predict muscle strength. The purpose of this study was to investigate PDFF-based TA of thigh muscles as a predictor of thigh muscle strength in comparison to mean PDFF. Methods: 30 healthy subjects (age = 30 ± 6 years; 15 females) underwent CSE-MRI of the lumbar spine at 3T, using a six-echo 3D spoiled gradient echo sequence. Quadriceps (EXT) and ischiocrural (FLEX) muscles were segmented to extract mean PDFF and texture features. Muscle flexion and extension strength were measured with an isokinetic dynamometer. Results: Of the eleven extracted texture features, Variance(global) showed the highest significant correlation with extension strength (p < 0.001, R²_adj = 0.712), and Correlation showed the highest significant correlation with flexion strength (p = 0.016, R²_adj = 0.658). Multivariate linear regression models identified Variance(global) and sex, but not PDFF, as significant predictors of extension strength (R²_adj = 0.709; p < 0.001), while mean PDFF, sex, and BMI, but none of the texture features, were identified as significant predictors of flexion strength (R²_adj = 0.674; p < 0.001). Conclusions: Prediction of quadriceps muscle strength can be improved beyond mean PDFF by means of TA, indicating the capability to quantify muscular fat infiltration patterns.

Texture Features of Proton Density Fat Fraction Maps from Chemical Shift Encoding-Based MRI Predict Paraspinal Muscle Strength

Texture analysis (TA) has shown promise as a surrogate marker for tissue structure, based on conventional and quantitative MRI sequences. Chemical-shift-encoding-based MRI (CSE-MRI)-derived proton density fat fraction (PDFF) of paraspinal muscles has been associated with various medical conditions including lumbar back pain (LBP) and neuromuscular diseases (NMD). Its application has been shown to improve the prediction of paraspinal muscle strength beyond muscle volume. Since mean PDFF values do not fully reflect muscle tissue structure, the purpose of our study was to investigate PDFF-based TA of paraspinal muscles as a predictor of muscle strength, as compared to mean PDFF. We performed 3T-MRI of the lumbar spine in 26 healthy subjects (age = 30 ± 6 years; 15 females) using a six-echo 3D spoiled gradient echo sequence for chemical-shift-encoding-based water–fat separation. Erector spinae (ES) and psoas (PS) muscles were segmented bilaterally from level L2–L5 to extract mean PDFF and texture features. Muscle flexion and extension strength was measured with an isokinetic dynamometer. Out of the eleven texture features extracted for each muscle, Kurtosis(global) of ES showed the highest significant correlation (r = 0.59, p = 0.001) with extension strength and Variance(global) of PS showed the highest significant correlation (r = 0.63, p = 0.001) with flexion strength. Using multivariate linear regression models, Kurtosis(global) of ES and BMI were identified as significant predictors of extension strength (R²_adj = 0.42; p < 0.001), and Variance(global) and Skewness(global) of PS were identified as significant predictors of flexion strength (R²_adj = 0.59; p = 0.001), while mean PDFF was not identified as a significant predictor. TA of CSE-MRI-based PDFF maps improves the prediction of paraspinal muscle strength beyond mean PDFF, potentially reflecting the ability to quantify the pattern of muscular fat infiltration. In the future, this may help to improve the pathophysiological understanding, diagnosis, monitoring and treatment evaluation of diseases with paraspinal muscle involvement, e.g., NMD and LBP.

Texture Analysis Using CT and Chemical Shift Encoding-Based Water-Fat MRI Can Improve Differentiation Between Patients With and Without Osteoporotic Vertebral Fractures

PURPOSE

Osteoporosis is a highly prevalent skeletal disease that frequently entails vertebral fractures. Areal bone mineral density (BMD) derived from dual-energy X-ray absorptiometry (DXA) is the reference standard, but has well-known limitations. Texture analysis can provide surrogate markers of tissue microstructure based on computed tomography (CT) or magnetic resonance imaging (MRI) data of the spine, thus potentially improving fracture risk estimation beyond areal BMD. However, it is largely unknown whether MRI-derived texture analysis can predict volumetric BMD (vBMD), or whether a model incorporating texture analysis based on CT and MRI may be capable of differentiating between patients with and without osteoporotic vertebral fractures.

MATERIALS AND METHODS

Twenty-six patients (15 females, median age: 73 years, 11 patients showing at least one osteoporotic vertebral fracture) who had CT and 3-Tesla chemical shift encoding-based water-fat MRI (CSE-MRI) available were analyzed. In total, 171 vertebral bodies of the thoracolumbar spine were segmented using an automatic convolutional neural network (CNN)-based framework, followed by extraction of integral and trabecular vBMD using CT data. For CSE-MRI, manual segmentation of vertebral bodies and consecutive extraction of the mean proton density fat fraction (PDFF) and T2* was performed. First-order, second-order, and higher-order texture features were derived from texture analysis using CT and CSE-MRI data. Stepwise multivariate linear regression models were computed using integral vBMD and fracture status as dependent variables.

RESULTS

Patients with osteoporotic vertebral fractures showed significantly lower integral and trabecular vBMD when compared to patients without fractures (p<0.001). For the model with integral vBMD as the dependent variable, T2* combined with three PDFF-based texture features explained 40% of the variance (adjusted R²[Ra2] = 0.40; p<0.001). Furthermore, regarding the differentiation between patients with and without osteoporotic vertebral fractures, a model including texture features from CT and CSE-MRI data showed better performance than a model based on integral vBMD and PDFF only ( Ra2 = 0.47 vs. Ra2  = 0.81; included texture features in the final model: integral vBMD, CT_Short-run_emphasis, CT_Varianceglobal, and PDFF_Variance).

CONCLUSION

Using texture analysis for spine CT and CSE-MRI can facilitate the differentiation between patients with and without osteoporotic vertebral fractures, implicating that future fracture prediction in osteoporosis may be improved.

Cervical spine osteoradionecrosis or bone metastasis after radiotherapy for nasopharyngeal carcinoma? The MRI-based radiomics for characterization

Background

To develop and validate an MRI-based radiomics nomogram for differentiation of cervical spine ORN from metastasis after radiotherapy (RT) in nasopharyngeal carcinoma (NPC).

Methods

A radiomics nomogram was developed in a training set that comprised 46 NPC patients after RT with 95 cervical spine lesions (ORN, n = 51; metastasis, n = 44), and data were gathered from January 2008 to December 2012. 279 radiomics features were extracted from the axial contrast-enhanced T1-weighted image (CE-T1WI). A radiomics signature was created by using the least absolute shrinkage and selection operator (LASSO) algorithm. A nomogram model was developed based on the radiomics scores. The performance of the nomogram was determined in terms of its discrimination, calibration, and clinical utility. An independent validation set contained 25 consecutive patients with 47 lesions (ORN, n = 25; metastasis, n = 22) from January 2013 to December 2015.

Results

The radiomics signature that comprised eight selected features was significantly associated with the differentiation of cervical spine ORN and metastasis. The nomogram model demonstrated good calibration and discrimination in the training set [AUC, 0.725; 95% confidence interval (CI), 0.622–0.828] and the validation set (AUC, 0.720; 95% CI, 0.573–0.867). The decision curve analysis indicated that the radiomics nomogram was clinically useful.

Conclusions

MRI-based radiomics nomogram shows potential value to differentiate cervical spine ORN from metastasis after RT in NPC.

Differentiation of usual vertebral compression fractures using CT histogram analysis as quantitative biomarkers: A proof-of-principle study

PURPOSE

To investigate the utility of CT histogram analysis (CTHA) for discrimination of traumatic, osteoporotic and malignant fractures in patients with vertebral compression fractures (VCFs). To evaluate the feasibility and accuracy of CTHA in differentiating non-malignant (traumatic and osteoporotic) from malignant VCFs.

MATERIALS AND METHODS

Totally, 235 patients with VCFs were enrolled in the current experimental study. There were 132 patients with traumatic VCFs, 51 with osteoporotic VCFs and 52 with malignant VCFs, with MRI and histology as the standard references. All the patients underwent unenhanced CT scans. Nineteen histogram-based parameters were derived using Omni-Kinetics software (Omni-Kinetics, GE Healthcare). The reproducibility of those parameters was evaluated using two independent delineations conducted by two observers. These histogram parameters were compared among the three different VCFs using Kruskal-Wallis H test. Traumatic VCFs and osteoporotic VCFs were combined as non-malignant VCFs and compared with malignant VCFs using Mann-Whitney U test Multivariable logistic regression analysis was performed on the significantly different features and built a diagnosis model. Receiver operating characteristic (ROC) curve was carried out to observe the difference of diagnostic performance between the single positive parameter and the combination of parameters.

RESULTS

All the 19 parameters presented excellent reproducibility, with intraclass correlation coefficient values from 0.789 to 0.997. At quantitative evaluation, the best predictive histogram parameters in discrimination of the three different types of VCFs were relative min intensity (p = 0.022), relative entropy (p = 0.043), and relative frequency size (p < 0.001). Relative frequency size (p < 0.001) and relative quantile5 (p = 0.012) resulted in statistically significant difference between non-malignant and malignant VCFs. The area under ROC curve indicated that relative frequency size combined with relative quantile5 (0.754; 95 % confidence intervals: 0.661∼0.829; p < 0.001) was of best performance in differentiating malignant from non-malignant VCFs.

CONCLUSIONS

Our results are encouraging and suggest that histogram parameters derived from unenhanced CT could be reliable quantitative biomarkers for diff ;erential diagnosis of usual VCFs.

MRI-Based Quantitative Osteoporosis Imaging at the Spine and Femur

Osteoporosis is a systemic skeletal disease with a high prevalence worldwide, characterized by low bone mass and microarchitectural deterioration, predisposing an individual to fragility fractures. Dual-energy X-ray absorptiometry (DXA) has been the clinical reference standard for diagnosing osteoporosis and for assessing fracture risk for decades. However, other imaging modalities are of increasing importance to investigate the etiology, treatment, and fracture risk. The purpose of this work is to review the available literature on quantitative magnetic resonance imaging (MRI) methods and related findings in osteoporosis at the spine and proximal femur as the clinically most important fracture sites. Trabecular bone microstructure analysis at the proximal femur based on high-resolution MRI allows for a better prediction of osteoporotic fracture risk than DXA-based bone mineral density (BMD) alone. In the 1990s, T₂ * mapping was shown to correlate with the density and orientation of the trabecular bone. Recently, quantitative susceptibility mapping (QSM), which overcomes some of the limitations of T₂ * mapping, has been applied for trabecular bone quantifications at the spine, whereas ultrashort echo time (UTE) imaging provides valuable surrogate markers of cortical bone quantity and quality. Magnetic resonance spectroscopy (MRS) and chemical shift encoding-based water-fat MRI (CSE-MRI) enable the quantitative assessment of the nonmineralized bone compartment through extraction of the bone marrow fat fraction (BMFF). Furthermore, CSE-MRI allows for the differentiation of osteoporotic vs. pathologic fractures, which is of high clinical relevance. Lastly, advanced postprocessing and image analysis tools, particularly considering statistical parametric mapping and region-specific BMFF distributions, have high potential to further improve MRI-based fracture risk assessments at the spine and hip. LEVEL OF EVIDENCE: 5 TECHNICAL EFFICACY STAGE: 2.

Radiographic Image Radiomics Feature Reproducibility: A Preliminary Study on the Impact of Field Size

RATIONALE AND OBJECTIVES

Radiomics is an approach to quantifying diseases. Recently, several studies have indicated that radiomics features are vulnerable against imaging parameters. The aim of this study is to assess how radiomics features change with radiographic field sizes, positions in the field size, and mAs.

MATERIALS AND METHODS

A large and small wood phantom and a cotton phantom were prepared and imaged in different field sizes, mAs, and placement in the radiographic field size. A region of interest was drawn on the image features, and twenty two features were extracted. Radiomics feature reproducibility was obtained based on coefficient of variation, Bland-Altman analysis, and intraclass correlation coefficient. Features with coefficient of variation ≤ 5%, intraclass correlation coefficient ≤ 90%, and 1% ≤ U/LRL ≤30% were introduced as robust features. U/LRL is upper/lower reproducibility limits in Bland-Altman.

RESULTS

For all field sizes and all phantoms, features including Difference Variance, Inverse Different Moment, Fraction, Long Run Emphasis, Run Length Non Uniformity, and Short Run Emphasis were found as highly reproducible features. For change in the position of field size, Fraction was the most reproducible in all field sizes and all phantoms. On the mAs change, we found that feature, Short Run Emphasis field 15 × 15 for small wood phantom, and Correlation in all field sizes for Cotton are the most reproducible features.

CONCLUSION

We demonstrated that radiomics features are strongly vulnerable against radiographic field size, positions in the radiation field, mAs, and phantom materials, and reproducibility analyses should be performed before each radiomics study. Moreover, these changing parameters should be considered, and their effects should be minimized in future radiomics studies.

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

PURPOSE

The purpose of this study was to develop predictive models to classify osteoporosis, osteopenia and normal patients using radiomics and machine learning approaches.

MATERIALS AND METHODS

A total of 147 patients were included in this retrospective single-center study. There were 12 men and 135 women with a mean age of 56.88±10.6 (SD) years (range: 28-87 years). For each patient, seven regions including four lumbar and three femoral including trochanteric, intertrochanteric and neck were segmented on bone mineral densitometry images and 54 texture features were extracted from the regions. The performance of four feature selection methods, including classifier attribute evaluation (CLAE), one rule attribute evaluation (ORAE), gain ratio attribute evaluation (GRAE) and principal components analysis (PRCA) along with four classification methods, including random forest (RF), random committee (RC), K-nearest neighbor (KN) and logit-boost (LB) were evaluated. Four classification categories, including osteopenia vs. normal, osteoporosis vs. normal, osteopenia vs. osteoporosis and osteoporosis+osteopenia vs. osteoporosis were examined for the defined seven regions. The classification model performances were evaluated using the area under the receiver operator characteristic curve (AUC).

RESULTS

The AUC values ranged from 0.50 to 0.78. The combination of methods RF+CLAE, RF+ORAE and RC+ORAE yielded highest performance (AUC=0.78) in discriminating between osteoporosis and normal state in the trochanteric region. The combinations of RF+PRCA and LB+PRCA had the highest performance (AUC=0.76) in discriminating between osteoporosis and normal state in the neck region.

CONCLUSION

The machine learning radiomic approach can be considered as a new method for bone mineral deficiency disease classification using bone mineral densitometry image features.

Vertebral Bone Marrow Heterogeneity Using Texture Analysis of Chemical Shift Encoding-Based MRI: Variations in Age, Sex, and Anatomical Location

Objective: Vertebral bone marrow composition has been extensively studied in the past and shown potential as imaging biomarker for osteoporosis, hematopoietic, and metabolic disorders. However, beyond quantitative assessment of bone marrow fat, little is known about its heterogeneity. Therefore, we investigated bone marrow heterogeneity of the lumbar spine using texture analysis of chemical-shift-encoding (CSE-MRI) based proton density fat fraction (PDFF) maps and its association with age, sex, and anatomical location. Methods: One hundred and fifty-six healthy subjects were scanned (age range: 20-29 years, 12/30 males/females; 30-39, 15/9; 40-49, 5/13; 50-59, 9/27; ≥60: 9/27). A sagittal 8-echo 3D spoiled-gradient-echo sequence at 3T was used for CSE-MRI-based water-fat separation at the lumbar spine. Manual segmentation of vertebral bodies L1-4 was performed. Mean PDFF and texture features (global: variance, skewness, kurtosis; second-order: energy, entropy, contrast, homogeneity, correlation, sum-average, variance, dissimilarity) were extracted at each vertebral level and compared between age groups, sex, and anatomical location. Results: Mean PDFF significantly increased from L1 to L4 (35.89 ± 11.66 to 39.52 ± 11.18%, p = 0.017) and with age (females: 27.19 ± 6.01 to 49.34 ± 7.75%, p < 0.001; males: 31.97 ± 7.96 to 41.83 ± 7.03 %, p = 0.025), but showed no difference between females and males after adjustment for age and BMI (37.13 ± 11.63 vs. 37.17 ± 8.67%; p = 0.199). Bone marrow heterogeneity assessed by texture analysis, in contrast to PDFF, was significantly higher in females compared to males after adjustment for age and BMI (namely contrast and dissimilarity; p < 0.031), demonstrated age-dependent differences, in particular in females (p < 0.05), but showed no statistically significant dependence on vertebral location. Conclusion: Vertebral bone marrow heterogeneity, assessed by texture analysis of PDFF maps, is primarily dependent on sex and age but not on anatomical location. Future studies are needed to investigate bone marrow heterogeneity with regard to aging and disease.