Prediction of the Acuity of Vertebral Compression Fractures on CT Using Radiologic and Radiomic Features

Constructing a Deep Learning Radiomics Model Based on X-ray Images and Clinical Data for Predicting and Distinguishing Acute and Chronic Osteoporotic Vertebral Fractures: A Multicenter Study

RATIONALE AND OBJECTIVES

To construct and validate a deep learning radiomics (DLR) model based on X-ray images for predicting and distinguishing acute and chronic osteoporotic vertebral fractures (OVFs).

METHODS

A total of 942 cases (1076 vertebral bodies) with both vertebral X-ray examination and MRI scans were included in this study from three hospitals. They were divided into a training cohort (n = 712), an internal validation cohort (n = 178), an external validation cohort (n = 111), and a prospective validation cohort (n = 75). The ResNet-50 model architecture was used for deep transfer learning (DTL), with pre-training performed on RadImageNet and ImageNet datasets. DTL features and radiomics features were extracted from lateral X-ray images of OVFs patients and fused together. A logistic regression model with the least absolute shrinkage and selection operator was established, with MRI showing bone marrow edema as the gold standard for acute OVFs. The performance of the model was evaluated using receiver operating characteristic curves. Eight machine learning classification models were evaluated for their ability to distinguish between acute and chronic OVFs. The Nomogram was constructed by combining clinical baseline data to achieve visualized classification assessment. The predictive performance of the best RadImageNet model and ImageNet model was compared using the Delong test. The clinical value of the Nomogram was evaluated using decision curve analysis (DCA).

RESULTS

Pre-training resulted in 34 and 39 fused features after feature selection and fusion. The most effective machine learning algorithm in both DLR models was Light Gradient Boosting Machine. Using the Delong test, the area under the curve (AUC) for distinguishing between acute and chronic OVFs in the training cohort was 0.979 and 0.972 for the RadImageNet and ImageNet models, respectively, with no statistically significant difference between them (P = 0.235). In the internal validation cohort, external validation cohort, and prospective validation cohort, the AUCs for the two models were 0.967 vs 0.629, 0.886 vs 0.817, and 0.933 vs 0.661, respectively, with statistically significant differences in all comparisons (P < 0.05). The deep learning radiomics nomogram (DLRN) was constructed by combining the predictive model of RadImageNet with clinical baseline features, resulting in AUCs of 0.981, 0.974, 0.895, and 0.902 in the training cohort, internal validation cohort, external validation cohort, and prospective validation cohort, respectively. Using the Delong test, the AUCs for the fused feature model and the DLRN in the training cohort were 0.979 and 0.981, respectively, with no statistically significant difference between them (P = 0.169). In the internal validation cohort, external validation cohort, and prospective validation cohort, the AUCs for the two models were 0.967 vs 0.974, 0.886 vs 0.895, and 0.933 vs 0.902, respectively, with statistically significant differences in all comparisons (P < 0.05). The Nomogram showed a slight improvement in predictive performance in the internal and external validation cohort, but a slight decrease in the prospective validation cohort (0.933 vs 0.902). DCA showed that the Nomogram provided more benefits to patients compared to the DLR models.

CONCLUSION

Compared to the ImageNet model, the RadImageNet model has higher diagnostic value in distinguishing between acute and chronic OVFs. Furthermore, the diagnostic performance of the model is further improved when combined with clinical baseline features to construct the Nomogram.

Red flags to screen for vertebral fracture in people presenting with low back pain

BACKGROUND

Low back pain is a common presentation across different healthcare settings. Clinicians need to confidently be able to screen and identify people presenting with low back pain with a high suspicion of serious or specific pathology (e.g. vertebral fracture). Patients identified with an increased likelihood of having a serious pathology will likely require additional investigations and specific treatment. Guidelines recommend a thorough history and clinical assessment to screen for serious pathology as a cause of low back pain. However, the diagnostic accuracy of recommended red flags (e.g. older age, trauma, corticosteroid use) remains unclear, particularly those used to screen for vertebral fracture.

OBJECTIVES

To assess the diagnostic accuracy of red flags used to screen for vertebral fracture in people presenting with low back pain. Where possible, we reported results of red flags separately for different types of vertebral fracture (i.e. acute osteoporotic vertebral compression fracture, vertebral traumatic fracture, vertebral stress fracture, unspecified vertebral fracture).

SEARCH METHODS

We used standard, extensive Cochrane search methods. The latest search date was 26 July 2022.

SELECTION CRITERIA

We considered primary diagnostic studies if they compared results of history taking or physical examination (or both) findings (index test) with a reference standard test (e.g. X-ray, magnetic resonance imaging (MRI), computed tomography (CT), single-photon emission computerised tomography (SPECT)) for the identification of vertebral fracture in people presenting with low back pain. We included index tests that were presented individually or as part of a combination of tests.

DATA COLLECTION AND ANALYSIS

Two review authors independently extracted data for diagnostic two-by-two tables from the publications or reconstructed them using information from relevant parameters to calculate sensitivity, specificity, and positive (+LR) and negative (-LR) likelihood ratios with 95% confidence intervals (CIs). We extracted aspects of study design, characteristics of the population, index test, reference standard, and type of vertebral fracture. Meta-analysis was not possible due to heterogeneity of studies and index tests, therefore the analysis was descriptive. We calculated sensitivity, specificity, and LRs for each test and used these as an indication of clinical usefulness. Two review authors independently conducted risk of bias and applicability assessment using the QUADAS-2 tool.

MAIN RESULTS

This review is an update of a previous Cochrane Review of red flags to screen for vertebral fracture in people with low back pain. We included 14 studies in this review, six based in primary care, five in secondary care, and three in tertiary care. Four studies reported on 'osteoporotic vertebral fractures', two studies reported on 'vertebral compression fracture', one study reported on 'osteoporotic and traumatic vertebral fracture', two studies reported on 'vertebral stress fracture', and five studies reported on 'unspecified vertebral fracture'. Risk of bias was only rated as low in one study for the domains reference standard and flow and timing. The domain patient selection had three studies and the domain index test had six studies rated at low risk of bias. Meta-analysis was not possible due to heterogeneity of the data. Results from single studies suggest only a small number of the red flags investigated may be informative. In the primary healthcare setting, results from single studies suggest 'trauma' demonstrated informative +LRs (range: 1.93 to 12.85) for 'unspecified vertebral fracture' and 'osteoporotic vertebral fracture' (+LR: 6.42, 95% CI 2.94 to 14.02). Results from single studies suggest 'older age' demonstrated informative +LRs for studies in primary care for 'unspecified vertebral fracture' (older age greater than 70 years: 11.19, 95% CI 5.33 to 23.51). Results from single studies suggest 'corticosteroid use' may be an informative red flag in primary care for 'unspecified vertebral fracture' (+LR range: 3.97, 95% CI 0.20 to 79.15 to 48.50, 95% CI 11.48 to 204.98) and 'osteoporotic vertebral fracture' (+LR: 2.46, 95% CI 1.13 to 5.34); however, diagnostic values varied and CIs were imprecise. Results from a single study suggest red flags as part of a combination of index tests such as 'older age and female gender' in primary care demonstrated informative +LRs for 'unspecified vertebral fracture' (16.17, 95% CI 4.47 to 58.43). In the secondary healthcare setting, results from a single study suggest 'trauma' demonstrated informative +LRs for 'unspecified vertebral fracture' (+LR: 2.18, 95% CI 1.86 to 2.54) and 'older age' demonstrated informative +LRs for 'osteoporotic vertebral fracture' (older age greater than 75 years: 2.51, 95% CI 1.48 to 4.27). Results from a single study suggest red flags as part of a combination of index tests such as 'older age and trauma' in secondary care demonstrated informative +LRs for 'unspecified vertebral fracture' (+LR: 4.35, 95% CI 2.92 to 6.48). Results from a single study suggest when '4 of 5 tests' were positive in secondary care, they demonstrated informative +LRs for 'osteoporotic vertebral fracture' (+LR: 9.62, 95% CI 5.88 to 15.73). In the tertiary care setting, results from a single study suggest 'presence of contusion/abrasion' was informative for 'vertebral compression fracture' (+LR: 31.09, 95% CI 18.25 to 52.96).

AUTHORS' CONCLUSIONS

The available evidence suggests that only a few red flags are potentially useful in guiding clinical decisions to further investigate people suspected to have a vertebral fracture. Most red flags were not useful as screening tools to identify vertebral fracture in people with low back pain. In primary care, 'older age' was informative for 'unspecified vertebral fracture', and 'trauma' and 'corticosteroid use' were both informative for 'unspecified vertebral fracture' and 'osteoporotic vertebral fracture'. In secondary care, 'older age' was informative for 'osteoporotic vertebral fracture' and 'trauma' was informative for 'unspecified vertebral fracture'. In tertiary care, 'presence of contusion/abrasion' was informative for 'vertebral compression fracture'. Combinations of red flags were also informative and may be more useful than individual tests alone. Unfortunately, the challenge to provide clear guidance on which red flags should be used routinely in clinical practice remains. Further research with primary studies is needed to improve and consolidate our current recommendations for screening for vertebral fractures to guide clinical care.

Differentiation of acute and chronic vertebral compression fractures using conventional CT based on deep transfer learning features and hand-crafted radiomics features

BACKGROUND

We evaluated the diagnostic efficacy of deep learning radiomics (DLR) and hand-crafted radiomics (HCR) features in differentiating acute and chronic vertebral compression fractures (VCFs).

METHODS

A total of 365 patients with VCFs were retrospectively analysed based on their computed tomography (CT) scan data. All patients completed MRI examination within 2 weeks. There were 315 acute VCFs and 205 chronic VCFs. Deep transfer learning (DTL) features and HCR features were extracted from CT images of patients with VCFs using DLR and traditional radiomics, respectively, and feature fusion was performed to establish the least absolute shrinkage and selection operator. The MRI display of vertebral bone marrow oedema was used as the gold standard for acute VCF, and the model performance was evaluated using the receiver operating characteristic (ROC).To separately evaluate the effectiveness of DLR, traditional radiomics and feature fusion in the differential diagnosis of acute and chronic VCFs, we constructed a nomogram based on the clinical baseline data to visualize the classification evaluation. The predictive power of each model was compared using the Delong test, and the clinical value of the nomogram was evaluated using decision curve analysis (DCA).

RESULTS

Fifty DTL features were obtained from DLR, 41 HCR features were obtained from traditional radiomics, and 77 features fusion were obtained after feature screening and fusion of the two. The area under the curve (AUC) of the DLR model in the training cohort and test cohort were 0.992 (95% confidence interval (CI), 0.983-0.999) and 0.871 (95% CI, 0.805-0.938), respectively. While the AUCs of the conventional radiomics model in the training cohort and test cohort were 0.973 (95% CI, 0.955-0.990) and 0.854 (95% CI, 0.773-0.934), respectively. The AUCs of the features fusion model in the training cohort and test cohort were 0.997 (95% CI, 0.994-0.999) and 0.915 (95% CI, 0.855-0.974), respectively. The AUCs of nomogram constructed by the features fusion in combination with clinical baseline data were 0.998 (95% CI, 0.996-0.999) and 0.946 (95% CI, 0.906-0.987) in the training cohort and test cohort, respectively. The Delong test showed that the differences between the features fusion model and the nomogram in the training cohort and the test cohort were not statistically significant (P values were 0.794 and 0.668, respectively), and the differences in the other prediction models in the training cohort and the test cohort were statistically significant (P < 0.05). DCA showed that the nomogram had high clinical value.

CONCLUSION

The features fusion model can be used for the differential diagnosis of acute and chronic VCFs, and its differential diagnosis ability is improved when compared with that when either radiomics is used alone. At the same time, the nomogram has a high predictive value for acute and chronic VCFs and can be a potential decision-making tool to assist clinicians, especially when a patient is unable to undergo spinal MRI examination.

Radiographic assessment of acute vs chronic vertebral compression fractures

PURPOSE

Distinguishing between acute and chronic vertebral compression fractures typically requires advanced imaging techniques such as magnetic resonance imaging (MRI). Recognizing specific radiographic findings associated with fracture acuity may improve the accuracy of radiographic assessment.

METHODS

Patients with compression fractures that had both radiographic and MRI studies of the lumbar spine within a 30-day time frame were retrospectively reviewed. MRI studies were used to determine compression fracture acuity. Radiographs were interpreted by a separate group of radiologists blinded to the MRI results. Radiographic findings of endplate osteophyte, subendplate density, subendplate cleft, and subendplate cyst were recorded as was the overall impression of fracture acuity.

RESULTS

Sensitivity and specificity for radiographic reporting of acute fracture were 0.52 (95% CI: 0.42, 0.61) and 0.95 (95% CI: 0.93, 0.97) respectively. For chronic fractures, the sensitivity and specificity were 0.52 (95% CI: 0.41, 0.63) and 0.94 (95% CI: 0.92, 0.96). The radiographic presence of a subendplate cleft increased the odds of a fracture being acute by a factor of 1.75 (95% CI: 1.09, 2.81; P = 0.0202). The radiographic presence of subendplate density increased the odds of a fracture being acute by a factor of 1.78 (95% CI: 1.21, 2.63; P = 0.0037). The presence of an endplate osteophyte or subendplate cyst was not significantly associated with fracture acuity.

CONCLUSION

Radiographs are relatively insensitive in distinguishing between acute and chronic lumbar compression fractures but the presence of a subendplate cleft or subendplate density increases the likelihood that a given fracture is acute.

A Predictive Clinical-Radiomics Nomogram for Differentiating Tuberculous Spondylitis from Pyogenic Spondylitis Using CT and Clinical Risk Factors

OBJECTIVE

The study aimed to develop and validate a nomogram model with clinical risk factors and radiomic features for differentiating tuberculous spondylitis (TS) from pyogenic spondylitis (PS).

METHODS

A total of 254 patients with TS (n = 141) or PS (n = 113) were randomly divided into training (n = 180) and validation (n = 74) groups. In addition, 43 patients (TS = 22 and PS = 21) were collected to construct a test cohort. t-test analysis, de-redundancy analysis, and minimum absolute shrinkage and selection operator (lasso) algorithm were utilized on the training set to obtain the optimal radiomics features from computed tomography (CT) for constructing the radiomics model and determine the radiomics score (Rad-score). Eight clinical risk predictors were identified to develop the clinical model. Combined with clinical risk predictors and Rad-scores, a nomogram model was constructed using multivariate logistic regression analysis.

RESULTS

A total of 1781 features were extracted, and 12 optimal radiomic features were utilized to construct the radiomic model and determine the Rad-score. The combined clinical radiomics model revealed good discrimination performance in both the training cohort and the validation cohort (AUC = 0.891 and 0.830) and was superior to the clinical (AUC = 0.807 and 0.785) and radiomics (AUC = 0.796 and 0.811) models. The calibration curve and DCA also depicted that the nomogram had better clinical efficacy. The discriminative performance of the model is well validated in the test cohort (AUC=0.877).

CONCLUSION

The clinical radiomic nomogram could serve as a promising predictive tool for differentiating TS from PS, which could be helpful for clinical decision-making.