Assessment of osteoporotic vertebral fractures using specialized workflow software for 6-point morphometry

Ground truth generalizability affects performance of the artificial intelligence model in automated vertebral fracture detection on plain lateral radiographs of the spine

BACKGROUND CONTEXT

Computer-aided diagnosis with artificial intelligence (AI) has been used clinically, and ground truth generalizability is important for AI performance in medical image analyses. The AI model was trained on one specific group of older adults (aged≧60) has not yet been shown to work equally well in a younger adult group (aged 18-59).

PURPOSE

To compare the performance of the developed AI model with ensemble method trained with the ground truth for those aged 60 years or older in identifying vertebral fractures (VFs) on plain lateral radiographs of spine (PLRS) between younger and older adult populations.

STUDY DESIGN/SETTING

Retrospective analysis of PLRS in a single medical institution.

OUTCOME MEASURES

Accuracy, sensitivity, specificity, and interobserver reliability (kappa value) were used to compare diagnostic performance of the AI model and subspecialists' consensus between the two groups.

METHODS

Between January 2016 and December 2018, the ground truth of 941 patients (one PLRS per person) aged 60 years and older with 1101 VFs and 6358 normal vertebrae was used to set up the AI model. The framework of the developed AI model includes: object detection with You Only Look Once Version 3 (YOLOv3) at T0-L5 levels in the PLRS, data pre-preprocessing with image-size and quality processing, and AI ensemble model (ResNet34, DenseNet121, and DenseNet201) for identifying or grading VFs. The reported overall accuracy, sensitivity and specificity were 92%, 91% and 93%, respectively, and external validation was also performed. Thereafter, patients diagnosed as VFs and treated in our institution during October 2019 to August 2020 were the study group regardless of age. In total, 258 patients (339 VFs and 1725 normal vertebrae) in the older adult population (mean age 78±10.4; range, 60-106) were enrolled. In the younger adult population (mean age 36±9.43; range, 20-49), 106 patients (120 VFs and 728 normal vertebrae) were enrolled. After identification and grading of VFs based on the Genant method with consensus between two subspecialists', VFs in each PLRS with human labels were defined as the testing dataset. The corresponding CT or MRI scan was used for labeling in the PLRS. The bootstrap method was applied to the testing dataset.

RESULTS

The model for clinical application, Digital Imaging and Communications in Medicine (DICOM) format, is uploaded directly (available at: http://140.113.114.104/vght_demo/svf-model (grading) and http://140.113.114.104/vght demo/svf-model2 (labeling). Overall accuracy, sensitivity and specificity in the older adult population were 93.36% (95% CI 93.34%-93.38%), 88.97% (95% CI 88.59%-88.99%) and 94.26% (95% CI 94.23%-94.29%), respectively. Overall accuracy, sensitivity and specificity in the younger adult population were 93.75% (95% CI 93.7%-93.8%), 65.00% (95% CI 64.33%-65.67%) and 98.49% (95% CI 98.45%-98.52%), respectively. Accuracy reached 100% in VFs grading once the VFs were labeled accurately. The unique pattern of limbus-like VFs, 43 (35.8%) were investigated only in the younger adult population. If limbus-like VFs from the dataset were not included, the accuracy increased from 93.75% (95% CI 93.70%-93.80%) to 95.78% (95% CI 95.73%-95.82%), sensitivity increased from 65.00% (95% CI 64.33%-65.67%) to 70.13% (95% CI 68.98%-71.27%) and specificity remained unchanged at 98.49% (95% CI 98.45%-98.52%), respectively. The main causes of false negative results in older adults were patients' lung markings, diaphragm or bowel airs (37%, n=14) followed by type I fracture (29%, n=11). The main causes of false negatives in younger adults were limbus-like VFs (45%, n=19), followed by type I fracture (26%, n=11). The overall kappa between AI discrimination and subspecialists' consensus in the older and younger adult populations were 0.77 (95% CI, 0.733-0.805) and 0.72 (95% CI, 0.6524-0.80), respectively.

CONCLUSIONS

The developed VF-identifying AI ensemble model based on ground truth of older adults achieved better performance in identifying VFs in older adults and non-fractured thoracic and lumbar vertebrae in the younger adults. Different age distribution may have potential disease diversity and implicate the effect of ground truth generalizability on the AI model performance.

Can a Deep-learning Model for the Automated Detection of Vertebral Fractures Approach the Performance Level of Human Subspecialists?

BACKGROUND

Vertebral fractures are the most common osteoporotic fractures in older individuals. Recent studies suggest that the performance of artificial intelligence is equal to humans in detecting osteoporotic fractures, such as fractures of the hip, distal radius, and proximal humerus. However, whether artificial intelligence performs as well in the detection of vertebral fractures on plain lateral spine radiographs has not yet been reported.

QUESTIONS/PURPOSES

(1) What is the accuracy, sensitivity, specificity, and interobserver reliability (kappa value) of an artificial intelligence model in detecting vertebral fractures, based on Genant fracture grades, using plain lateral spine radiographs compared with values obtained by human observers? (2) Do patients' clinical data, including the anatomic location of the fracture (thoracic or lumbar spine), T-score on dual-energy x-ray absorptiometry, or fracture grade severity, affect the performance of an artificial intelligence model? (3) How does the artificial intelligence model perform on external validation?

METHODS

Between 2016 and 2018, 1019 patients older than 60 years were treated for vertebral fractures in our institution. Seventy-eight patients were excluded because of missing CT or MRI scans (24% [19]), poor image quality in plain lateral radiographs of spines (54% [42]), multiple myeloma (5% [4]), and prior spine instrumentation (17% [13]). The plain lateral radiographs of 941 patients (one radiograph per person), with a mean age of 76 ± 12 years, and 1101 vertebral fractures between T7 and L5 were retrospectively evaluated for training (n = 565), validating (n = 188), and testing (n = 188) of an artificial intelligence deep-learning model. The gold standard for diagnosis (ground truth) of a vertebral fracture is the interpretation of the CT or MRI reports by a spine surgeon and a radiologist independently. If there were any disagreements between human observers, the corresponding CT or MRI images would be rechecked by them together to reach a consensus. For the Genant classification, the injured vertebral body height was measured in the anterior, middle, and posterior third. Fractures were classified as Grade 1 (< 25%), Grade 2 (26% to 40%), or Grade 3 (> 40%). The framework of the artificial intelligence deep-learning model included object detection, data preprocessing of radiographs, and classification to detect vertebral fractures. Approximately 90 seconds was needed to complete the procedure and obtain the artificial intelligence model results when applied clinically. The accuracy, sensitivity, specificity, interobserver reliability (kappa value), receiver operating characteristic curve, and area under the curve (AUC) were analyzed. The bootstrapping method was applied to our testing dataset and external validation dataset. The accuracy, sensitivity, and specificity were used to investigate whether fracture anatomic location or T-score in dual-energy x-ray absorptiometry report affected the performance of the artificial intelligence model. The receiver operating characteristic curve and AUC were used to investigate the relationship between the performance of the artificial intelligence model and fracture grade. External validation with a similar age population and plain lateral radiographs from another medical institute was also performed to investigate the performance of the artificial intelligence model.

RESULTS

The artificial intelligence model with ensemble method demonstrated excellent accuracy (93% [773 of 830] of vertebrae), sensitivity (91% [129 of 141]), and specificity (93% [644 of 689]) for detecting vertebral fractures of the lumbar spine. The interobserver reliability (kappa value) of the artificial intelligence performance and human observers for thoracic and lumbar vertebrae were 0.72 (95% CI 0.65 to 0.80; p < 0.001) and 0.77 (95% CI 0.72 to 0.83; p < 0.001), respectively. The AUCs for Grades 1, 2, and 3 vertebral fractures were 0.919, 0.989, and 0.990, respectively. The artificial intelligence model with ensemble method demonstrated poorer performance for discriminating normal osteoporotic lumbar vertebrae, with a specificity of 91% (260 of 285) compared with nonosteoporotic lumbar vertebrae, with a specificity of 95% (222 of 234). There was a higher sensitivity 97% (60 of 62) for detecting osteoporotic (dual-energy x-ray absorptiometry T-score ≤ -2.5) lumbar vertebral fractures, implying easier detection, than for nonosteoporotic vertebral fractures (83% [39 of 47]). The artificial intelligence model also demonstrated better detection of lumbar vertebral fractures compared with detection of thoracic vertebral fractures based on the external dataset using various radiographic techniques. Based on the dataset for external validation, the overall accuracy, sensitivity, and specificity on bootstrapping method were 89%, 83%, and 95%, respectively.

CONCLUSION

The artificial intelligence model detected vertebral fractures on plain lateral radiographs with high accuracy, sensitivity, and specificity, especially for osteoporotic lumbar vertebral fractures (Genant Grades 2 and 3). The rapid reporting of results using this artificial intelligence model may improve the efficiency of diagnosing vertebral fractures. The testing model is available at http://140.113.114.104/vght_demo/corr/. One or multiple plain lateral radiographs of the spine in the Digital Imaging and Communications in Medicine format can be uploaded to see the performance of the artificial intelligence model.

LEVEL OF EVIDENCE

Level II, diagnostic study.

Differentiation Between Osteoporotic And Neoplastic Vertebral Fractures: State Of The Art And Future Perspectives

Vertebral fractures are a common condition, occurring in the context of osteoporosis and malignancy. These entities affect a group of patients in the same age range; clinical features may be indistinct and symptoms non-existing, and thus present challenges to diagnosis. In this article, we review the use and accuracy of different imaging modalities available to characterize vertebral fracture etiology, from well-established classical techniques, to the role of new and advanced imaging techniques, and the prospective use of artificial intelligence. We also address the role of imaging on treatment. In the context of osteoporosis, the importance of opportunistic diagnosis is highlighted. In the near future, the use of automated computer-aided diagnostic algorithms applied to different imaging techniques may be really useful to aid on diagnosis.

CT based quantitative measures of the stability of fractured metastatically involved vertebrae treated with spine stereotactic body radiotherapy

Mechanical instability secondary to vertebral metastases can lead to pathologic vertebral compression fracture (VCF) mechanical pain, neurological compromise, and the need for surgical stabilization. Stereotactic body radiation therapy (SBRT) as a treatment for spinal metastases is effective for pain and local tumor control, it has been associated with an increased risk of VCF. This study quantified computed tomography (CT) based stability measures in metastatic vertebrae with VCF treated with spine SBRT. It was hypothesized that semi-automated quantification of VCF based on CT metrics would be related to clinical outcomes. 128 SBRT treated spinal metastases patients were identified from a prospective database. Of these, 18 vertebral segments were identified with a VCF post-SBRT. A semi-automated system for quantifying VCF was developed based on CT imaging before and after SBRT. The system identified and segmented SBRT treated vertebral bodies, calculated stability metrics at single time points and changes over time. In the vertebrae that developed a new (n = 7) or progressive (n = 11) VCF following SBRT, the median time to VCF/VCF progression was 1.74 months (range 0.53-7.79 months). Fractured thoracolumbar vertebrae that went on to be stabilized (cemented and/or instrumented), had greater fractured vertebral body volume progression over time (12%) compared to those not stabilized (0.4%, p < 0.05). Neither the spinal instability neoplastic score (SINS) or any single timepoint stability metrics in post-hoc analyses correlated with future stabilization. This pilot study presents a quantitative semi-automated method assessing fractured thoracolumbar vertebrae based on CT. Increased fractured vertebral body volume progression post-SBRT was shown to predict those patients who were subsequently stabilized, motivating study of methods that assess temporal radiological changes toward augmenting existing clinical management in the metastatic spine.

Computational techniques to segment and classify lumbar compression fractures

Vertebral fractures are important indicators of osteoporosis. Fractures with partial collapse of vertebral bodies are referred to as vertebral compression fractures (VCFs) that are usually non-traumatic in nature. Some common causes of VCFs are trauma, bone failure related to osteoporosis (benign) and metastatic cancer (malignant). This paper aims at developing a system for computer-aided diagnosis to help in the detection, labeling and segmentation of lumbar vertebral body (VB) and to further classify each VB into normal, malignant and benign VCFs. After the initial preprocessing, morphological, shape and angular features are used in the detection, labeling and segmentation steps. Various shape and statistical texture features are extracted from the segmented VB and are fed to the classifier for the final decision. The segmentation and classification results obtained were compared with the ground truth manual segmentation of the lumbar VB and the decision labels of the fractures provided by the experts. The dice similarity coefficient (DSC) for segmentation reached up to 94.27%, and the classification results show that shape and texture features together are able to correctly classify with an accuracy rate of 95.34%. The final outcomes are expected to be useful in the analysis of vertebral compression fractures.

Are semi-automated software program designed for adults accurate for the identification of vertebral fractures in children?

Objectives

To assess whether diagnostic accuracy of morphometric vertebral fracture (VF) diagnosis in children can be improved using AVERT™ (a 33-point semi-automated program developed for VF diagnosis in adults) compared with SpineAnalyzer™ (a 6-point program), which has previously been shown to be of insufficient accuracy.

Materials and methods

Lateral spine radiographs (XR) and dual-energy X-ray absorptiometry (DXA) scans of 50 children and young people were analysed by two observers using two different programs (AVERT™ and SpineAnalyzer™). Diagnostic accuracy (sensitivity, specificity, false-negative (FN) and false-positive rates (FP)) was calculated by comparing with a previously established consensus arrived at by three experienced paediatric musculoskeletal radiologists, using a simplified algorithm-based qualitative scoring system. Observer agreement was calculated using Cohen’s kappa.

Results

For XR, overall sensitivity, specificity, FP and FN rates using AVERT™ were 36%, 95%, 5% and 64% respectively and 26%, 98%, 2% and 75% respectively, using SpineAnalyzer™. For DXA, overall sensitivity, specificity, FP and FN rates using AVERT™ were 41%, 91%, 9% and 59% respectively and 31%, 96%, 4% and 69% respectively, using SpineAnalyzer. Reliability (kappa) ranged from 0.34 to 0.37 (95%CI, 0.26–0.46) for AVERT™ and from 0.26 to 0.31 (95%CI, 0.16–0.44) for SpineAnalyzer™. Inter- and intra-observer agreement ranged from 0.41 to 0.47 for AVERT™ and from 0.50 to 0.79 for SpineAnalyzer™.

Conclusion

AVERT™ has slightly higher accuracy but lower observer reliability for the representation of vertebral morphometry in children when compared with SpineAnalyzer™. However, neither software program is satisfactorily reliable for VF diagnosis in children.

Key Points

• SpineAnalyzer™ and AVERT™ have low diagnostic accuracy and observer agreement when compared to three paediatric radiologists’ readings for the diagnosis of vertebral fractures (VF) in children.

• Neither AVERT™ nor SpineAnalyzer™ is satisfactorily reliable for VF diagnosis in children.

• Development of specific paediatric software and normative values (incorporating age-related physiological variation in children) is required.

Vertebral Body Compression Fractures and Bone Density: Automated Detection and Classification on CT Images

Purpose To create and validate a computer system with which to detect, localize, and classify compression fractures and measure bone density of thoracic and lumbar vertebral bodies on computed tomographic (CT) images. Materials and Methods Institutional review board approval was obtained, and informed consent was waived in this HIPAA-compliant retrospective study. A CT study set of 150 patients (mean age, 73 years; age range, 55-96 years; 92 women, 58 men) with (n = 75) and without (n = 75) compression fractures was assembled. All case patients were age and sex matched with control subjects. A total of 210 thoracic and lumbar vertebrae showed compression fractures and were electronically marked and classified by a radiologist. Prototype fully automated spinal segmentation and fracture detection software were then used to analyze the study set. System performance was evaluated with free-response receiver operating characteristic analysis. Results Sensitivity for detection or localization of compression fractures was 95.7% (201 of 210; 95% confidence interval [CI]: 87.0%, 98.9%), with a false-positive rate of 0.29 per patient. Additionally, sensitivity was 98.7% and specificity was 77.3% at case-based receiver operating characteristic curve analysis. Accuracy for classification by Genant type (anterior, middle, or posterior height loss) was 0.95 (107 of 113; 95% CI: 0.89, 0.98), with weighted κ of 0.90 (95% CI: 0.81, 0.99). Accuracy for categorization by Genant height loss grade was 0.68 (77 of 113; 95% CI: 0.59, 0.76), with a weighted κ of 0.59 (95% CI: 0.47, 0.71). The average bone attenuation for T12-L4 vertebrae was 146 HU ± 29 (standard deviation) in case patients and 173 HU ± 42 in control patients; this difference was statistically significant (P < .001). Conclusion An automated machine learning computer system was created to detect, anatomically localize, and categorize vertebral compression fractures at high sensitivity and with a low false-positive rate, as well as to calculate vertebral bone density, on CT images. ^© RSNA, 2017 Online supplemental material is available for this article.

A systematic review of diagnostic accuracy of vertebral fracture assessment (VFA) in postmenopausal women and elderly men

This systematic review was performed to compare the diagnostic accuracy of vertebral fracture assessment (VFA) with that of spinal radiography for identification of vertebral fractures (VFs). VFA appeared to have moderate sensitivity and high specificity for detecting VFs when compared with spinal radiography.

INTRODUCTION

VFs are recognized as the hallmark of osteoporosis, and a previous VF increases the risk of a future fracture. Therefore, the timely detection of VFs is important for prevention of further fractures. This systematic review examined the diagnostic accuracy of VFA using dual X-ray absorptiometry (DXA) to identify VFs.

METHODS

We searched for potentially relevant studies using electronic databases, including Ovid-Medline, Ovid-EMBASE, Cochrane library, and four Korean databases, from their inception to May 2013. We compared the diagnostic accuracy of VFA with that of spinal radiography for detection of VFs by analyzing the sensitivity and specificity using a 2 × 2 contingency table. Subgroup analyses were also performed on studies with a low risk of bias and applicability.

RESULTS

Twelve studies were analyzed for the diagnostic accuracy of VFA. The sensitivity and specificity were 0.70-0.93 and 0.95-1.00, respectively, analyzed on a per-vertebra basis, and 0.65-1.00 and 0.74-1.00 on a per-patient basis. The sensitivity and specificity of five studies in subgroups with a low risk of bias in the intervention test were 0.70-0.84 and 0.96-0.99, respectively. In studies with a low risk of bias in the patient selection, those based on a per-vertebra basis in three studies were 0.70-0.93 and 0.96-1.00, respectively.

CONCLUSIONS

VFA had moderate sensitivity and high specificity for detecting VF when compared with spinal radiography. However, the present findings are insufficient to assess whether spinal radiography should be replaced by VFA.

Shape, texture and statistical features for classification of benign and malignant vertebral compression fractures in magnetic resonance images

PURPOSE

Vertebral compression fractures (VCFs) result in partial collapse of vertebral bodies. They usually are nontraumatic or occur with low-energy trauma in the elderly secondary to different etiologies, such as insufficiency fractures of bone fragility in osteoporosis (benign fractures) or vertebral metastasis (malignant fractures). Our study aims to classify VCFs in T1-weighted magnetic resonance images (MRI).

METHODS

We used the median sagittal planes of lumbar spine MRIs from 63 patients (38 women and 25 men) previously diagnosed with VCFs. The lumbar vertebral bodies were manually segmented and statistical features of gray levels were computed from the histogram. We also extracted texture and shape features to analyze the contours of the vertebral bodies. In total, 102 lumbar VCFs (53 benign and 49 malignant) and 89 normal lumbar vertebral bodies were analyzed. The k-nearest-neighbor method, a neural network with radial basis functions, and a naïve Bayes classifier were used with feature selection. We compared the classification obtained by these classifiers with the final diagnosis of each case, including biopsy for the malignant fractures and clinical and laboratory follow up for the benign fractures.

RESULTS

The results obtained show an area under the receiver operating characteristic curve of 0.97 in distinguishing between normal and fractured vertebral bodies, and 0.92 in discriminating between benign and malignant fractures.

CONCLUSIONS

The proposed classification methods based on shape, texture, and statistical features have provided high accuracy and may assist in the diagnosis of VCFs.

Vertebral Morphometry

There is as yet no agreement about the criteria by which to arrive at an imaging diagnosis of a vertebral fracture. Because high-grade fractures result in alterations in vertebral shape, 1 possible avenue of diagnosis has been to quantitate changes in vertebral shape. The result has been a variety of methods for the relative and absolute measurements of vertebral dimensions. Such measurements have also lent themselves to automated computed analysis. The number of techniques reflects the absence of any consensus about the best. The semiquantitative technique proposed by Genant has become the most widely used and has served the field well for comparative purposes. Its use in higher grade fractures has been widely endorsed, if some concepts (e.g., short vertebral height-vertebrae) are at variance with lower grades of fracturing. Vertebral morphometry may be the only recourse in high volume epidemiological and interventional studies.

Automated Detection, Localization, and Classification of Traumatic Vertebral Body Fractures in the Thoracic and Lumbar Spine at CT

PURPOSE

To design and validate a fully automated computer system for the detection and anatomic localization of traumatic thoracic and lumbar vertebral body fractures at computed tomography (CT).

MATERIALS AND METHODS

This retrospective study was HIPAA compliant. Institutional review board approval was obtained, and informed consent was waived. CT examinations in 104 patients (mean age, 34.4 years; range, 14-88 years; 32 women, 72 men), consisting of 94 examinations with positive findings for fractures (59 with vertebral body fractures) and 10 control examinations (without vertebral fractures), were performed. There were 141 thoracic and lumbar vertebral body fractures in the case set. The locations of fractures were marked and classified by a radiologist according to Denis column involvement. The CT data set was divided into training and testing subsets (37 and 67 subsets, respectively) for analysis by means of prototype software for fully automated spinal segmentation and fracture detection. Free-response receiver operating characteristic analysis was performed.

RESULTS

Training set sensitivity for detection and localization of fractures within each vertebra was 0.82 (28 of 34 findings; 95% confidence interval [CI]: 0.68, 0.90), with a false-positive rate of 2.5 findings per patient. The sensitivity for fracture localization to the correct vertebra was 0.88 (23 of 26 findings; 95% CI: 0.72, 0.96), with a false-positive rate of 1.3. Testing set sensitivity for the detection and localization of fractures within each vertebra was 0.81 (87 of 107 findings; 95% CI: 0.75, 0.87), with a false-positive rate of 2.7. The sensitivity for fracture localization to the correct vertebra was 0.92 (55 of 60 findings; 95% CI: 0.79, 0.94), with a false-positive rate of 1.6. The most common cause of false-positive findings was nutrient foramina (106 of 272 findings [39%]).

CONCLUSION

The fully automated computer system detects and anatomically localizes vertebral body fractures in the thoracic and lumbar spine on CT images with a high sensitivity and a low false-positive rate.

Statistical shape and appearance models in osteoporosis

Statistical models (SMs) of shape (SSM) and appearance (SAM) have been acquiring popularity in medical image analysis since they were introduced in the early 1990s. They have been primarily used for segmentation, but they are also a powerful tool for 3D reconstruction and classification. All these tasks may be required in the osteoporosis domain, where fracture detection and risk estimation are key to reducing the mortality and/or morbidity of this bone disease. In this article, we review the different applications of SSMs and SAMs in the context of osteoporosis, and it concludes with a discussion of their advantages and disadvantages for this application.

Statistical shape and appearance models of bones

When applied to bones, statistical shape models (SSM) and statistical appearance models (SAM) respectively describe the mean shape and mean density distribution of bones within a certain population as well as the main modes of variations of shape and density distribution from their mean values. The availability of this quantitative information regarding the detailed anatomy of bones provides new opportunities for diagnosis, evaluation, and treatment of skeletal diseases. The potential of SSM and SAM has been recently recognized within the bone research community. For example, these models have been applied for studying the effects of bone shape on the etiology of osteoarthritis, improving the accuracy of clinical osteoporotic fracture prediction techniques, design of orthopedic implants, and surgery planning. This paper reviews the main concepts, methods, and applications of SSM and SAM as applied to bone.

Validity and role of vertebral fracture assessment in detecting prevalent vertebral fracture in patients with rheumatoid arthritis

OBJECTIVES

We aimed to identify the validity and the role of vertebral fracture assessment (VFA) for the diagnosis of prevalent vertebral fracture (VF) in rheumatoid arthritis (RA) patients.

METHODS

Total of 100 women with RA who were 50 years or older were enrolled. All participants underwent lateral imaging of the thoraco-lumbar spine by radiography and VFA. All radiographs were analyzed by two radiologists. Discrepancies between radiologists for spine radiography were resolved by consensus and these results were defined as the reference standard. VFA interpretation was done independently by two nuclear medicine physicians. Fracture defined by VFA measure was done only when two physicians both independently reported fracture. The inter-rater agreement for the detection of VF on VFA was evaluated by kappa statistics. We calculated percent values for the diagnostic validity of VFA in detecting VF in the 100 women as a whole and according to the presence of previous fracture or back pain.

RESULTS

The prevalence of VF identified by spine radiography was 47%. Inter-rater agreement of VFA per vertebra by two VFA readers showed moderate agreement (kappa=0.60). The sensitivity, PPV, specificity and NPV of VFA compared to spine radiography were 57.3%, 30.9%, 89.1% and 96.1% for total vertebrae. All patients with history of previous VF (n=13) were visualized with VFA with 100% sensitivity but it has 64.7% sensitivity and 79.3% specificity in patients without previous VF (n=87).

CONCLUSION

VFA is most useful to identify patients without VF because of its high specificity and NPV in all spine level.

High prevalence of vertebral fractures assessed by quantitative morphometry in hemodialysis patients, strongly associated with vascular calcifications

Few studies have provided information on the prevalence of vertebral fractures (VFs) and their risk factors in hemodialysis patients. A multicenter, cross-sectional, observational study was carried out to assess the prevalence of VFs and vascular calcifications (VCs) in 387 hemodialysis patients (mean age 64.2 ± 14.1 years, 63 % males) and in a control group of 51 osteoporotic subjects. Biochemical tests included 25(OH) vitamin D, bone Gla protein (total and undercarboxylated), and total matrix Gla protein. Vertebral quantitative morphometry was carried out centrally for the detection of VF, defined as reduction by ≥20 % of one of the vertebral body dimensions. In the same radiograph, aortic and iliac VC scores were calculated. Prevalence of VF was 55.3 % in hemodialysis patients and 51.0 % in the control group. Multivariate analysis disclosed that male gender (59.8 vs. 47.6 %, p = 0.02; OR = 1.78, 95 % CI 1.15-2.75) and age (mean ± SD 66.7 ± 13.1 vs. 61.0 ± 14.7 years, p < 0.001; OR = 1.03, 95 % CI 1.01-1.05) were significantly associated with VF. The prevalence of aortic VC was significantly higher in hemodialysis patients than in controls (80.6 vs. 68.4 %, p = 0.001). The factors with the strongest association with VC, apart from atrial fibrillation, were serum 25(OH)vitamin D levels below 29 ng/mL for aortic VC (OR = 1.85, 95 % CI 1.04-3.29) and VF both for aortic (OR = 1.77, 95 % CI 1.00-3.14) and iliac (OR = 1.96, 95 % CI 1.27-3.04) VC. In conclusion, the prevalence of VF, especially in males, and VC, in both genders, is high in hemodialysis patients. VF is associated with VC. Vitamin D deficiency is also associated with VC. Further longitudinal studies are warranted to investigate fractures in renal patients.

High precision semiautomated computed tomography measurement of lumbar disk and vertebral heights

PURPOSE

Evaluation of treatments of many spine disorders requires precise measurement of the heights of vertebral bodies and disk spaces. The authors present a semiautomated computer algorithm measuring those heights from spine computed tomography (CT) scans and evaluate its precision.

METHODS

Eight patients underwent two spine CT scans in the same day. In each scan, five thoracolumbar vertebral heights and four disk heights were estimated using the algorithm. To assess precision, the authors computed the differences between the height measurements in the two scans, coefficients of variation (CV), and 95% limits of agreement. Intraoperator and interoperator precisions were evaluated. For local vertebral and disk height measurement (anterior, middle, posterior) the algorithm was compared to a manual mid-sagittal plane method.

RESULTS

The mean (standard deviation) interscan difference was as low as 0.043 (0.031) mm for disk heights and 0.044 (0.043) mm for vertebral heights. The corresponding 95% limits of agreement were [-0.085, 0.11] and [-0.10, 0.12] mm, respectively. Intraoperator and interoperator precision was high, with a maximal CV of 0.30%. For local vertebral and disk heights, the algorithm improved upon the precision of the manual mid-sagittal plane measurement by as much as a factor of 6 and 4, respectively.

CONCLUSIONS

The authors evaluated the precision of a novel computer algorithm for measuring vertebral body heights and disk heights using short term repeat CT scans of patients. The 95% limits of agreement indicate that the algorithm can detect small height changes of the order of 0.1 mm.

Early diagnosis of vertebral fractures

Vertebral fractures are a common clinical entity, caused by trauma or related to osteoporosis (benign). Their recognition is especially important in the post-menopausal female population but also important is their differentiation from pathological (malignant) fractures (1). A vertebral fracture is evidenced by vertebral body deformity or reduction in vertebral body height beyond a certain threshold value in the absence of bone discontinuity. For prognosis and treatment it is extremely important to recognize the cause of the fracture. In contrast to fractures that occur in other locations, vertebral fractures often go unrecognized in the acute phase as the pain may be transient and radiographic and evaluation of the spine may be difficult (2). Objective measurement of the vertebral deformity provides invaluable information to the interpreting physician and helps grade fracture severity. The recognition and diagnosis of vertebral fractures can be performed using additional diagnostic tools.

Agreement between semi-automatic radiographic morphometry and Genant semi-quantitative method in the assessment of vertebral fractures

Semi-automatic morphometry is highly reproducible and not time intensive; however, no study has evaluated agreement between semi-automated morphometric methods and the Genant semi-quantitative method performed as a rule by radiologists. Our study shows substantial agreement between both methods; however, semi-automatic morphometry upgrades mild deformities and overestimates the prevalence of fractures.

INTRODUCTION

The aim of this study was to evaluate the agreement between radiologists using the Genant semi-quantitative (SQ) method and semi-automated morphometry in the diagnosis of vertebral fractures in post-menopausal women.

METHODS

Cross-sectional study was conducted in 2006-2007 in an age-stratified population-based sample of 824 post-menopausal women over the age of 50. From this population two sets of 95 and 50 X-ray were randomly extracted to test inter-rater agreement and agreement between SQ and semi-automated morphometry, and vertebral fractures were classified according to both methods. The Genant method was used to homogenise the diagnosis of fractures. Agreement was evaluated with weighted kappa. We evaluated each vertebral body independently and also the whole vertebral column (T4-L4) classifying women into the worst grade of fracture. For the qualitative interpretation of the agreement, we used the criteria described by Landis and Koch (Biometrics 33:159-174, 1977).

RESULTS

The radiologists' agreement was 98.4% (Kappa, 0.75; 95% CI, 0.42-0.89). Agreement between semi-automated morphometry and SQ reached 97.6% and Kappa was 0.86 (95% CI, 0.66-0.94). In the whole evaluation of the spine semi-automated morphometry overestimates, the prevalence of fractures compared with the radiologists were 15.8% of women with fractures and 7.4% of women with moderate-severe fractures by semi-automated morphometry vs. 8.4% and 3.2% by the SQ method. The negative predictive value for MorphoXpress was 99% while the positive was 40%.

CONCLUSIONS

Semi-automated morphometry shows high reliability and a substantial agreement with the SQ approach but overestimates the prevalence of fractures. Its role in routine clinical practice is limited because positive results should be reassessed by qualitative or semi-quantitative methods.

New dual-energy X-ray absorptiometry equipment in the assessment of vertebral fractures: technical limits and software accuracy

OBJECTIVE

The aim of this study was to investigate software accuracy and influence of body mass index on image quality of Lunar iDXA (Lunar, Madison, WI, USA; software enCORE 12.0) in vertebral fracture (VFs) assessment.

MATERIALS AND METHODS

We enrolled 65 normal or overweight patients (group 1) and 64 obese patients (group 2) with indication for morphometric evaluation of the spine. Patients underwent iDXA, with scans performed in the standard manner by an expert technologist. Lateral images of the spine were subsequently evaluated by a musculoskeletal radiologist as the gold standard. Our analysis considered five points: vertebral bodies missed or not assessable or wrongly labeled on T4-L4 segment, diagnostic performance of the automatic morphometric point-positioning system in the detection of VFs, upgrading and downgrading of fractures, radiologist intervention rate, and BMI influence.

RESULTS

In group 1, 57/845 (6.7%) vertebral bodies and 34/832 (4.1%) in group 2 were not assessable-the upper thoracic spine. enCORE failed to recognize vertebral levels in 5.4% of the patients (7.7% in group 1 vs. 3.1% in group 2). On a lesion-based analysis sensitivity, specificity and accuracy of the software were 81.4, 93.8, and 93.1% in group 1 and 69.1, 88.3, and 86.7% in group 2, respectively. For 52.7% of the vertebrae in group 1 (51/8 upgraded/downgraded) and 70.0% in group 2 (96/26 upgraded/downgraded), a point correction was necessary and this changed the diagnosis respectively in 29.2 and 50.0% of the patients. Differences in diagnostic performance and point correction rate were significantly different between the two groups; however, BMI did not significantly affect vertebral level labeling and was correlated with a better visualization of the whole T4-L4 spine segment.

CONCLUSIONS

This study provides new and interesting information about the accuracy, reliability, and imaging quality provided by iDXA in the assessment of VFs.

Integrated imaging approach to osteoporosis: state-of-the-art review and update

Osteoporosis is the most common of all metabolic bone disorders. It is characterized by low bone mass and microarchitectural deterioration of bone tissue, with a consequent increase in bone fragility and susceptibility to fractures. Because of the increasing aging of the world population, the number of persons affected by osteoporosis is also increasing. Complications related to osteoporosis can create social and economic burdens. For these reasons, the early diagnosis of osteoporosis is crucial. Conventional radiography allows qualitative and semiquantitative evaluation of osteoporosis, whereas other imaging techniques allow quantification of bone loss (eg, dual-energy x-ray absorptiometry and quantitative computed tomography [CT]), assessment for the presence of fractures (morphometry), and the study of bone properties (ultrasonography). In recent years, new imaging modalities such as micro-CT and high-resolution magnetic resonance imaging have been developed in an attempt to help diagnose osteoporosis in its early stages, thereby reducing social and economic costs and preventing patient suffering. The correct diagnosis of osteoporosis results in better management in terms of prevention and adequate pharmacologic or surgical treatment.

Detection of vertebral body fractures based on cortical shell unwrapping

Assessment of trauma patients with multiple injuries can be one of the most clinically challenging situations dealt with by the radiologist. We propose a fully automated method to detect acute vertebral body fractures on trauma CT studies. The spine is first segmented and partitioned into vertebrae. Then the cortical shell of the vertebral body is extracted using deformable dual-surface models. The extracted cortical shell is unwrapped onto a 2D map effectively converting a complex 3D fracture detection problem into a pattern recognition problem of fracture lines on a 2D plane. Twenty-eight features are computed for each fracture line and sent to a committee of support vector machines for classification. The system was tested on 18 trauma CT datasets and achieved 95.3% sensitivity and 1.7 false positives per case by leave-one-out cross validation.

Intra-and inter-reader reliability of semi-automated quantitative morphometry measurements and vertebral fracture assessment using lateral scout views from computed tomography

Intra-and inter-reader reliability of semi-automated quantitative vertebral morphometry measurements was determined using lateral computed tomography (CT) scout views. The method requires less time than conventional morphometry. Reliability was excellent for vertebral height measurements, good for height ratios, and comparable to semi-quantitative grading by radiologists for identification of vertebral fractures.

INTRODUCTION

Underdiagnosis and undertreatment of vertebral fracture (VFx) is a well-known problem worldwide. Thus, new methods are needed to facilitate identification of VFx. This study aimed to determine intra- and inter-reader reliability of semi-automated quantitative vertebral morphometry based on shape-based statistical modeling (SpineAnalyzer, Optasia Medical, Cheadle, UK).

METHODS

Two non-radiologists independently assessed vertebral morphometry from CT lateral scout views at two time points in 96 subjects (50 men, 46 women, 70.3 ± 8.9 years) selected from the Framingham Heart Study Offspring and Third Generation Multi-Detector CT Study. VFxs were classified based solely on morphometry measurements using Genant's criteria. Intraclass correlation coefficients (ICCs), root mean squared coefficient of variation (RMS CV) and kappa (k) statistics were used to assess reliability.

RESULTS

We analyzed 1,246 vertebrae in 96 subjects. The analysis time averaged 5.4 ± 1.7 min per subject (range, 3.2-9.1 min). Intra-and inter-reader ICCs for vertebral heights were excellent (>0.95) for all vertebral levels combined. Intra-and inter-reader RMS CV for height measurements ranged from 2.5% to 3.9% and 3.3% to 4.4%, respectively. Reliability of vertebral height ratios was good to fair. Based on morphometry measurements alone, readers A and B identified 51-52 and 46-59 subjects with at least one prevalent VFx, respectively, and there was a good intra-and inter-reader agreement (k = 0.59-0.69) for VFx identification.

CONCLUSIONS

Semi-automated quantitative vertebral morphometry measurements from CT lateral scout views are convenient and reproducible, and may facilitate assessment of VFx.

Radiation exposure in X-ray-based imaging techniques used in osteoporosis

Recent advances in medical X-ray imaging have enabled the development of new techniques capable of assessing not only bone quantity but also structure. This article provides (a) a brief review of the current X-ray methods used for quantitative assessment of the skeleton, (b) data on the levels of radiation exposure associated with these methods and (c) information about radiation safety issues. Radiation doses associated with dual-energy X-ray absorptiometry are very low. However, as with any X-ray imaging technique, each particular examination must always be clinically justified. When an examination is justified, the emphasis must be on dose optimisation of imaging protocols. Dose optimisation is more important for paediatric examinations because children are more vulnerable to radiation than adults. Methods based on multi-detector CT (MDCT) are associated with higher radiation doses. New 3D volumetric hip and spine quantitative computed tomography (QCT) techniques and high-resolution MDCT for evaluation of bone structure deliver doses to patients from 1 to 3 mSv. Low-dose protocols are needed to reduce radiation exposure from these methods and minimise associated health risks.

Prevalence of vertebral fracture in postmenopausal women with lumbar osteopenia using MorphoXpress® (OSTEOXPRESS Study)

BACKGROUND AND AIMS

Vertebral fracture (VF) is the most common complication of osteoporosis. However, more than half of all VF are asymptomatic and may go unnoticed, even in patients with osteoporosis. Our aim was to assess the prevalence of VF in postmenopausal women with osteopenic lumbar densitometry by means of vertebral morphometry, using the MorphoXpress® software.

PATIENTS AND METHODS

This was an epidemiological, cross-sectional, multicenter study conducted among 289 postmenopausal women (>1 year of amenorrhoea), diagnosed with lumbar osteopenia (not due to chronic treatment with corticosteroids or immobilization). Vertebral deformities ≥20% were considered as VF.

RESULTS

Demographic and clinical characteristics showed mean age (±SD) 64±9 years, body mass index 27±5 kg/m2, and time from diagnosis of 2±3 years. A total of 25% of subjects had a family history of osteoporotic fracture in first-degree relatives, and 23% had previous fragility fracture. The prevalence of VF was 50% (CI 95% 44-56), the most frequent being the dorsal wedge (34%). Previous fragility fracture was a risk factor for VF (OR 3.13, p=0.0004). A total of 76.5% of patients were receiving treatment, mainly calcium and vitamin D supplements (70%) and bisphosphonates (27%).

CONCLUSIONS

MorphoXpress® revealed that 50% of postmenopausal women with osteopenic lumbar densitometry showed VF. This result is important since only 7% of all evaluated subjects had previously been diagnosed with VF.

Underreported vertebral fractures in an Italian population: comparison of plain radiographs vs quantitative measurements

PURPOSE

Vertebral fractures (VFs) are the hallmark of osteoporosis and are responsible for almost 70,000 hospital admissions yearly, implying social costs and impaired quality of life for patients. In recent years, several techniques, both qualitative and quantitative, have been proposed for VF diagnosis, but a gold standard is not yet available and the visual semiquantitative (VSQ) assessment proposed by Genant remains the most validated. However, given the lack of a standardised method, in clinical practice, the diagnosis of VF is often missed, and patients are not correctly assessed. The aim of our study was to estimate the percentage of VFs not detected in clinical practice in italian population using the VSQ method and a new morphometric technique.

MATERIALS AND METHODS

In 283 postmenopausal women referred to our clinic for osteoporosis screening, we performed a clinical examination, plain spinal radiographs (for VSQ assessment) and digital computerised morphometry (DCM) to assess VFs. Bone density was measured using dual-energy X-ray absorptiometry (DXA).

RESULTS

Forty-seven percent of patients had a T score <-2.5 standard deviations (SD), and 35.2% were osteopenic, but no significant correlations between T score and grade or number of fractures were found. DCM identified VFs in 38.5% of patients versus 32.5% using the VSQ method. Overall, 280 VFs were detected by DCM and 236 by VSQ, whereas only 105 were recognised by the reports.

CONCLUSIONS

VFs went undetected in 55.5% according to the VSQ method on standard spinal radiographs. Therefore, the morphometric technique may be helpful when performed with the semiquantitative approach to improve recognition of VFs. However, other studies are needed to further validate the utility of this new morphometric technique in clinical practice.

Vertebral morphometry

Visual semiquantitative (SQ) assessment of the radiographs by a trained and experienced observer is the "gold standard" method to detect vertebral fractures. Vertebral morphometry is a quantitative method to identify osteoporotic vertebral fractures based on the measurement of vertebral heights. Vertebral morphometry may be performed on conventional spinal radiographs (MRX: morphometric x-ray radiography) or on images obtained from dual x-ray absorptiometry (DXA) scans (MXA: morphometric x-ray absorptiometry). Vertebral fracture assessment (VFA) indicates the method for identification of the vertebral fractures using lateral spine views acquired by DXA, with low-dose exposition. For epidemiologic studies and clinical drug trials in osteoporosis research but also in clinical practice, the preferred method is radiographic SQ assessment., because an expert eye can better distinguish between true fractures and vertebral anomalies than can quantitative morphometry. However, vertebral morphometry, calculating the deformity of overall thoracic and lumbar spine, may supply useful data about the vertebral fracture risk. VFA performed during routine densitometry allows identification, by visual or morphometric methods, of most osteoporotic vertebral fractures, even those that are asymptomatic.