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

Differential diagnostic value of radiomics models in benign versus malignant vertebral compression fractures: A systematic review and meta-analysis

PURPOSE

Early diagnosis of benign and malignant vertebral compression fractures by analyzing imaging data is crucial to guide treatment and assess prognosis, and the development of radiomics made it an alternative option to biopsy examination. This systematic review and meta-analysis was conducted with the purpose of quantifying the diagnostic efficacy of radiomics models in distinguishing between benign and malignant vertebral compression fractures.

METHODS

Searching on PubMed, Embase, Web of Science and Cochrane Library was conducted to identify eligible studies published before September 23, 2023. After evaluating for methodological quality and risk of bias using the Radiomics Quality Score (RQS) and the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2), we selected studies providing confusion matrix results to be included in random-effects meta-analysis.

RESULTS

A total of sixteen articles, involving 1,519 vertebrae with pathological-diagnosed tumor infiltration, were included in our meta-analysis. The combined sensitivity and specificity of the top-performing models were 0.92 (95 % CI: 0.87-0.96) and 0.93 (95 % CI: 0.88-0.96), respectively. Their AUC was 0.97 (95 % CI: 0.96-0.99). By contrast, radiologists' combined sensitivity was 0.90 (95 %CI: 0.75-0.97) and specificity was 0.92 (95 %CI: 0.67-0.98). The AUC was 0.96 (95 %CI: 0.94-0.97). Subsequent subgroup analysis and sensitivity test suggested that part of the heterogeneity might be explained by differences in imaging modality, segmentation, deep learning and cross-validation.

CONCLUSION

We found remarkable diagnosis potential in correctly distinguishing vertebral compression fractures in complex clinical contexts. However, the published radiomics models still have a great heterogeneity, and more large-scale clinical trials are essential to validate their generalizability.

Radiomics Based on Multimodal magnetic resonance imaging for the Differential Diagnosis of Benign and Malignant Vertebral Compression Fractures

OBJECTIVES

Recent studies have indicated that radiomics may have excellent performance and clinical application prospects in the differential diagnosis of benign and malignant vertebral compression fractures (VCFs). However, multimodal magnetic resonance imaging (MRI)-based radiomics model is rarely used in the differential diagnosis of benign and malignant VCFs, and is limited to lumbar. Herein, this study intends to develop and validate MRI radiomics models for differential diagnoses of benign and malignant VCFs in patients.

METHODS

This cross-sectional study involved 151 adult patients diagnosed with VCF in The First Affiliated Hospital of Soochow University in 2016-2021. The study was conducted in three steps: (i) the original MRI images were segmented, and the region of interest (ROI) was marked out; (ii) among the extracted features, those features with Pearson's correlation coefficient lower than 0.9 and the top 15 with the highest variance and Lasso regression coefficient less than and more than 0 were selected; (iii) MRI images and combined data were studied by logistic regression, decision tree, random forest and extreme gradient boosting (XGBoost) models in training set and the test set (ratio of 8:2), respectively; and the models were further verified and evaluated for the differential diagnosis performance. The evaluated indexes included area under receiver (AUC) of operating characteristic curve, accuracy, sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and 95% confidence intervals (CIs). The AUCs were used to assess the predictive performance of different machine learning modes for benign and malignant VCFs.

RESULTS

A total of 1144 radiomics features, and 14 clinical features were extracted. Finally, 12 radiomics features were included in the radiomics model, and 12 radiomics features with 14 clinical features were included in the combined model. In the radiomics model, the differential diagnosis performance in the logistic regression model with the AUC of 0.905 ± 0.026, accuracy of 0.817 ± 0.057, sensitivity of 0.831 ± 0.065, and negative predictive value of 0.813 ± 0.042, was superior to the other three. In the combined model, XGBoost model had the superior differential diagnosis performance with specificity (0.979 ± 0.026) and positive predictive value (0.971 ± 0.035).

CONCLUSION

The multimodal MRI-based radiomics model performed well in the differential diagnosis of benign and malignant VCFs, which may provide a tool for clinicians to differentially diagnose VCFs.

A Comparative Study of a Nomogram and Machine Learning Models in Predicting Early Hematoma Expansion in Hypertensive Intracerebral Hemorrhage

RATIONALE AND OBJECTIVES

Early identification for hematoma expansion can help improve patient outcomes. Presently, there are many methods to predict hematoma expansion. This study compared a variety of models to find a model suitable for clinical promotion.

MATERIALS AND METHODS

Non-contrast head CT images and clinical data were collected from 203 patients diagnosed with hypertensive intracerebral hemorrhage. Radiomics features were extracted from all CT images, and the dataset was randomly divided into training and validation sets (7:3 ratio) after applying the synthetic minority oversampling method. The radiomics score (Radscore) was calculated using least absolute shrinkage and selection operator (LASSO) regression, combined with selected clinical predictors, to develop a nomogram and four machine learning (ML) models: logistic regression, random forest, support vector machine, and extreme gradient boosting (XGBoost). Discrimination, calibration and clinical usefulness of the nomogram and ML models were assessed.

RESULTS

The nomogram and ML models were integrated with Radscore and clinical predictors. The nomogram demonstrated favorable discriminatory ability in the training set with an AUC of 0.80, which was confirmed in the validation set (AUC=0.76). Among the ML models, the XGBoost model achieved the highest AUC (training set=0.89 and validation set=0.85), surpassing that of the nomogram. The XGBoost model exhibited good clinical usefulness.

CONCLUSION

Both the nomogram and ML models constructed by non-contrast head CT image-based Radscore integrated with clinical predictors can predict early hematoma expansion of hypertensive intracerebral hemorrhage, and the XGBoost model had the highest prediction performance and best clinical usefulness.

Application of bone alkaline phosphatase and 25-oxhydryl-vitamin D in diagnosis and prediction of osteoporotic vertebral compression fractures

BACKGROUND

Osteoporosis is a bone metabolic disease that usually causes fracture. The improvement of the clinical diagnostic efficiency of osteoporosis is of great significance for the prevention of fracture. The predictive and diagnostic values of bone alkaline phosphatase (B-ALP) and 25-oxhydryl-vitamin D (25-OH-VD) for osteoporotic vertebral compression fractures (OVCFs) were evaluated.

METHODS

110 OVCFs patients undergoing percutaneous vertebroplasty were included as subjects and their spinal computed tomography (CT) images were collected. After that, deep convolutional neural network model was employed for intelligent fracture recognition. Next, the patients were randomly enrolled into Ctrl group (65 cases receiving postoperative routine treatment) and VD2 group (65 cases injected with vitamin D2 into muscle after the surgery). In addition, 100 healthy people who participated in physical examination were included in Normal group. The differences in Oswestry dysfunction indexes (ODI), imaging parameters, B-ALP and 25-OH-VD expressions, and quality of life (QOL) scores of patients among the three groups were compared. The values of B-ALP and 25-OH-VD in predicting and diagnosing OVCFs and their correlation with bone density were analyzed.

RESULTS

It was demonstrated that computer intelligent medical image technique was more efficient in fracture CT recognition than artificial recognition. In contrast to those among patients in Normal group, B-ALP rose while 25-OH-VD declined among patients in Ctrl and VD2 groups (P < 0.05). Versus those among patients in Ctrl group, ODI, Cobb angle, and B-ALP reduced, while bone density, the height ratio of the injured vertebrae, 25-OH-VD, and QOL score increased among patients in VD2 group after the treatment (P < 0.05). The critical values, accuracy, and areas under the curve (AUC) of the diagnosis of OVCFs by B-ALP and 25-OH-VD amounted to 87.8 μg/L versus 30.3 nmol/L, 86.7% versus 83.3%, and 0.86 versus 0.82, respectively. B-ALP was apparently negatively correlated with bone density (r =  - 0.602, P < 0.05), while 25-OH-VD was remarkably positively correlated with bone density (r = 0.576, P < 0.05).

CONCLUSION

To sum up, deep learning-based computer CT image intelligent detection technique could improve the diagnostic efficacy of fracture. B-ALP rose while 25-OH-VD declined among patients with OVCFs and OVCFs could be predicted and diagnosed based on B-ALP and 25-OH-VD. Postoperative intramuscular injection of VD2 could effectively improve the therapeutic effect on patients with OVCFs and QOL.

Differential diagnosis of benign and malignant vertebral fracture on CT using deep learning

OBJECTIVES

To evaluate the performance of deep learning using ResNet50 in differentiation of benign and malignant vertebral fracture on CT.

METHODS

A dataset of 433 patients confirmed with 296 malignant and 137 benign fractures was retrospectively selected from our spinal CT image database. A senior radiologist performed visual reading to evaluate six imaging features, and three junior radiologists gave diagnostic prediction. A ROI was placed on the most abnormal vertebrae, and the smallest square bounding box was generated. The input channel into ResNet50 network was 3, including the slice with its two neighboring slices. The diagnostic performance was evaluated using 10-fold cross-validation. After obtaining the malignancy probability from all slices in a patient, the highest probability was assigned to that patient to give the final diagnosis, using the threshold of 0.5.

RESULTS

Visual features such as soft tissue mass and bone destruction were highly suggestive of malignancy; the presence of a transverse fracture line was highly suggestive of a benign fracture. The reading by three radiologists with 5, 3, and 1 year of experience achieved an accuracy of 99%, 95.2%, and 92.8%, respectively. In ResNet50 analysis, the per-slice diagnostic sensitivity, specificity, and accuracy were 0.90, 0.79, and 85%. When the slices were combined to ve per-patient diagnosis, the sensitivity, specificity, and accuracy were 0.95, 0.80, and 88%.

CONCLUSION

Deep learning has become an important tool for the detection of fractures on CT. In this study, ResNet50 achieved good accuracy, which can be further improved with more cases and optimized methods for future clinical implementation.

KEY POINTS

• Deep learning using ResNet50 can yield a high accuracy for differential diagnosis of benign and malignant vertebral fracture on CT. • The per-slice diagnostic sensitivity, specificity, and accuracy were 0.90, 0.79, and 85% in deep learning using ResNet50 analysis. • The slices combined with per-patient diagnostic sensitivity, specificity, and accuracy were 0.95, 0.80, and 88% in deep learning using ResNet50 analysis.