Computer-Aided Diagnosis of Vertebral Compression Fractures Using Convolutional Neural Networks and Radiomics

Development of a deep learning model for detecting lumbar vertebral fractures on CT images: An external validation

OBJECTIVE

To develop and externally validate a binary classification model for lumbar vertebral body fractures based on CT images using deep learning methods.

METHODS

This study involved data collection from two hospitals for AI model training and external validation. In Cohort A from Hospital 1, CT images from 248 patients, comprising 1508 vertebrae, revealed that 20.9% had fractures (315 vertebrae) and 79.1% were non-fractured (1193 vertebrae). In Cohort B from Hospital 2, CT images from 148 patients, comprising 887 vertebrae, indicated that 14.8% had fractures (131 vertebrae) and 85.2% were non-fractured (756 vertebrae). The AI model for lumbar spine fractures underwent two stages: vertebral body segmentation and fracture classification. The first stage utilized a 3D V-Net convolutional deep neural network, which produced a 3D segmentation map. From this map, region of each vertebra body were extracted and then input into the second stage of the algorithm. The second stage employed a 3D ResNet convolutional deep neural network to classify each proposed region as positive (fractured) or negative (not fractured).

RESULTS

The AI model's accuracy for detecting vertebral fractures in Cohort A's training set (n = 1199), validation set (n = 157), and test set (n = 152) was 100.0 %, 96.2 %, and 97.4 %, respectively. For Cohort B (n = 148), the accuracy was 96.3 %. The area under the receiver operating characteristic curve (AUC-ROC) values for the training, validation, and test sets of Cohort A, as well as Cohort B, and their 95 % confidence intervals (CIs) were as follows: 1.000 (1.000, 1.000), 0.978 (0.944, 1.000), 0.986 (0.969, 1.000), and 0.981 (0.970, 0.992). The area under the precision-recall curve (AUC-PR) values were 1.000 (0.996, 1.000), 0.964 (0.927, 0.985), 0.907 (0.924, 0.984), and 0.890 (0.846, 0.971), respectively. According to the DeLong test, there was no significant difference in the AUC-ROC values between the test set of Cohort A and Cohort B, both for the overall data and for each specific vertebral location (all P>0.05).

CONCLUSION

The developed model demonstrates promising diagnostic accuracy and applicability for detecting lumbar vertebral fractures.

Weakly supervised deep learning for diagnosis of multiple vertebral compression fractures in CT

OBJECTIVE

This study aims to develop a weakly supervised deep learning (DL) model for vertebral-level vertebral compression fracture (VCF) classification using image-level labelled data.

METHODS

The training set included 815 patients with normal (n = 507, 62%) or VCFs (n = 308, 38%). Our proposed model was trained on image-level labelled data for vertebral-level classification. Another supervised DL model was trained with vertebral-level labelled data to compare the performance of the proposed model.

RESULTS

The test set included 227 patients with normal (n = 117, 52%) or VCFs (n = 110, 48%). For a fair comparison of the two models, we compared sensitivities with the same specificities of the proposed model and the vertebral-level supervised model. The specificity for overall L1-L5 performance was 0.981. The proposed model may outperform the vertebral-level supervised model with sensitivities of 0.770 vs 0.705 (p = 0.080), respectively. For vertebral-level analysis, the specificities for each L1-L5 were 0.974, 0.973, 0.970, 0.991, and 0.995, respectively. The proposed model yielded the same or better sensitivity than the vertebral-level supervised model in L1 (0.750 vs 0.694, p = 0.480), L3 (0.793 vs 0.586, p < 0.05), L4 (0.833 vs 0.667, p = 0.480), and L5 (0.600 vs 0.600, p = 1.000), respectively. The proposed model showed lower sensitivity than the vertebral-level supervised model for L2, but there was no significant difference (0.775 vs 0.825, p = 0.617).

CONCLUSIONS

The proposed model may have a comparable or better performance than the supervised model in vertebral-level VCF classification.

CLINICAL RELEVANCE STATEMENT

Vertebral-level vertebral compression fracture classification aids in devising patient-specific treatment plans by identifying the precise vertebrae affected by compression fractures.

KEY POINTS

• Our proposed weakly supervised method may have comparable or better performance than the supervised method for vertebral-level vertebral compression fracture classification. • The weakly supervised model could have classified cases with multiple vertebral compression fractures at the vertebral-level, even if the model was trained with image-level labels. • Our proposed method could help reduce radiologists' labour because it enables vertebral-level classification from image-level labels.

Artificial intelligence in fracture detection with different image modalities and data types: A systematic review and meta-analysis

Artificial Intelligence (AI), encompassing Machine Learning and Deep Learning, has increasingly been applied to fracture detection using diverse imaging modalities and data types. This systematic review and meta-analysis aimed to assess the efficacy of AI in detecting fractures through various imaging modalities and data types (image, tabular, or both) and to synthesize the existing evidence related to AI-based fracture detection. Peer-reviewed studies developing and validating AI for fracture detection were identified through searches in multiple electronic databases without time limitations. A hierarchical meta-analysis model was used to calculate pooled sensitivity and specificity. A diagnostic accuracy quality assessment was performed to evaluate bias and applicability. Of the 66 eligible studies, 54 identified fractures using imaging-related data, nine using tabular data, and three using both. Vertebral fractures were the most common outcome (n = 20), followed by hip fractures (n = 18). Hip fractures exhibited the highest pooled sensitivity (92%; 95% CI: 87-96, p< 0.01) and specificity (90%; 95% CI: 85-93, p< 0.01). Pooled sensitivity and specificity using image data (92%; 95% CI: 90-94, p< 0.01; and 91%; 95% CI: 88-93, p < 0.01) were higher than those using tabular data (81%; 95% CI: 77-85, p< 0.01; and 83%; 95% CI: 76-88, p < 0.01), respectively. Radiographs demonstrated the highest pooled sensitivity (94%; 95% CI: 90-96, p < 0.01) and specificity (92%; 95% CI: 89-94, p< 0.01). Patient selection and reference standards were major concerns in assessing diagnostic accuracy for bias and applicability. AI displays high diagnostic accuracy for various fracture outcomes, indicating potential utility in healthcare systems for fracture diagnosis. However, enhanced transparency in reporting and adherence to standardized guidelines are necessary to improve the clinical applicability of AI. Review Registration: PROSPERO (CRD42021240359).

Value of 18F-FDG-PET/CT radiomics combined with clinical variables in the differential diagnosis of malignant and benign vertebral compression fractures

BACKGROUND

Vertebral compression fractures (VCFs) are common clinical problems that arise from various reasons. The differential diagnosis of benign and malignant VCFs is challenging. This study was designed to develop and validate a radiomics model to predict benign and malignant VCFs with 18F-fluorodeoxyglucose-positron emission tomography/computed tomography (18F-FDG-PET/CT).

RESULTS

Twenty-six features (9 PET features and 17 CT features) and eight clinical variables (age, SUVmax, SUVpeak, SULmax, SULpeak, osteolytic destruction, fracture line, and appendices/posterior vertebrae involvement) were ultimately selected. The area under the curve (AUCs) of the radiomics and clinical-radiomics models were significantly different from that of the clinical model in both the training group (0.986, 0.987 vs. 0.884, p < 0.05) and test group (0.962, 0.948 vs. 0.858, p < 0.05), while there was no significant difference between the radiomics model and clinical-radiomics model (p > 0.05). The accuracies of the radiomics and clinical-radiomics models were 94.0% and 95.0% in the training group and 93.2% and 93.2% in the test group, respectively. The three models all showed good calibration (Hosmer-Lemeshow test, p > 0.05). According to the decision curve analysis (DCA), the radiomics model and clinical-radiomics model exhibited higher overall net benefit than the clinical model.

CONCLUSIONS

The PET/CT-based radiomics and clinical-radiomics models showed good performance in distinguishing between malignant and benign VCFs. The radiomics method may be valuable for treatment decision-making.

A software program for automated compressive vertebral fracture detection on elderly women's lateral chest radiograph: Ofeye 1.0

Background

Because osteoporotic vertebral fracture (OVF) on chest radiographs is commonly missed in radiological reports, we aimed to develop a software program which offers automated detection of compressive vertebral fracture (CVF) on lateral chest radiographs, and which emphasizes CVF detection specificity with a low false positivity rate.

Methods

For model training, we retrieved 3,991 spine radiograph cases and 1,979 chest radiograph cases from 16 sources, with among them in total 1,404 cases had OVF. For model testing, we retrieved 542 chest radiograph cases and 162 spine radiograph cases from four independent clinics, with among them 215 cases had OVF. All cases were female subjects, and except for 31 training data cases which were spine trauma cases, all the remaining cases were post-menopausal women. Image data included DICOM (Digital Imaging and Communications in Medicine) format, hard film scanned PNG (Portable Network Graphics) format, DICOM exported PNG format, and PACS (Picture Archiving and Communication System) downloaded resolution reduced DICOM format. OVF classification included: minimal and mild grades with <20% or ≥20-25% vertebral height loss respectively, moderate grade with ≥25-40% vertebral height loss, severe grade with ≥40%-2/3 vertebral height loss, and collapsed grade with ≥2/3 vertebral height loss. The CVF detection base model was mainly composed of convolution layers that include convolution kernels of different sizes, pooling layers, up-sampling layers, feature merging layers, and residual modules. When the model loss function could not be further decreased with additional training, the model was considered to be optimal and termed 'base-model 1.0'. A user-friendly interface was also developed, with the synthesized software termed 'Ofeye 1.0'.

Results

Counting cases and with minimal and mild OVFs included, base-model 1.0 demonstrated a specificity of 97.1%, a sensitivity of 86%, and an accuracy of 93.9% for the 704 testing cases. In total, 33 OVFs in 30 cases had a false negative reading, which constituted a false negative rate of 14.0% (30/215) by counting all OVF cases. Eighteen OVFs in 15 cases had OVFs of ≥ moderate grades missed, which constituted a false negative rate of 7.0% (15/215, i.e., sensitivity 93%) if only counting cases with ≥ moderate grade OVFs missed. False positive reading was recorded in 13 vertebrae in 13 cases (one vertebra in each case), which constituted a false positivity rate of 2.7% (13/489). These vertebrae with false positivity labeling could be readily differentiated from a true OVF by a human reader. The software Ofeye 1.0 allows 'batch processing', for example, 100 radiographs can be processed in a single operation. This software can be integrated into hospital PACS, or installed in a standalone personal computer.

Conclusions

A user-friendly software program was developed for CVF detection on elderly women's lateral chest radiographs. It has an overall low false positivity rate, and for moderate and severe CVFs an acceptably low false negativity rate. The integration of this software into radiological practice is expected to improve osteoporosis management for elderly women.