Sensitivity For Detection Of Fractures Research Articles

Automated detection and classification of acute vertebral body fractures using a convolutional neural network on computed tomography.

Acute vertebral fracture is usually caused by low-energy injury with osteoporosis and high-energy trauma. The AOSpine thoracolumbar spine injury classification system (AO classification) plays an important role in the diagnosis and treatment of the disease. The diagnosis and description of vertebral fractures according to the classification scheme requires a great deal of time and energy for radiologists. To design and validate a multistage deep learning system (multistage AO system) for the automatic detection, localization and classification of acute thoracolumbar vertebral body fractures according to AO classification on computed tomography. The CT images of 1,217 patients who came to our hospital from January 2015 to December 2019 were collected retrospectively. The fractures were marked and classified by 2 junior radiology residents according to the type A standard in the AO classification. Marked fracture sites included the upper endplate, lower endplate and posterior wall. When there were inconsistent opinions on classification labels, the final result was determined by a director radiologist. We integrated different networks into different stages of the overall framework. U-net and a graph convolutional neural network (U-GCN) are used to realize the location and classification of the thoracolumbar spine. Next, a classification network is used to detect whether the thoracolumbar spine has a fracture. In the third stage, we detect fractures in different parts of the thoracolumbar spine by using a multibranch output network and finally obtain the AO types. The mean age of the patients was 61.87 years with a standard deviation of 17.04 years, consisting of 760 female patients and 457 male patients. On vertebrae level, sensitivity for fracture detection was 95.23% in test dataset, with an accuracy of 97.93% and a specificity of 98.35%. For the classification of vertebral body fractures, the balanced accuracy was 79.56%, with an AUC of 0.904 for type A1, 0.945 for type A2, 0.878 for type A3 and 0.942 for type A4. The multistage AO system can automatically detect and classify acute vertebral body fractures in the thoracolumbar spine on CT images according to AO classification with high accuracy.

Ultra-low-dose CT versus radiographs for minor spine and pelvis trauma: a Bayesian analysis of accuracy.

To compare diagnosis performance and effective dose of ultra-low-dose CT (ULD CT) versus radiographs in suspected spinal or pelvic ring or hip fracture for minor trauma. ULD CT, in addition to radiography, was prospectively performed in consecutive patients admitted to the emergency department for minor traumas, during working hours over 2 months. Presence of a recent fracture was assessed by two blind radiologists independently. Sensitivities and specificities were estimated using the best valuable comparator (BVC) as a reference and using a latent class model in Bayesian inference (BLCM). Dosimetric indicators were recorded and effective doses (E) were calculated using conversion coefficient. Eighty areas were analyzed in 69 patients, including 22 dorsal spine, 28 lumbar spine, and 30 pelvic ring/hip. Thirty-six fractures (45%) were observed. Applying the BVC method, depending on location, ULD CT sensitivity was 80 to 100% for reader 1 and 85 to 100% for reader 2, whereas radiographic sensitivity was 60 to 85% for reader 1 and 50 to 92% for reader 2. With BLCM approach for reader 2, ULD CT sensitivity for all locations/dorsal spine/lumbar spine and pelvic ring-hip was 87.1/75.9/84.2/76.9% respectively. Corresponding radiograph sensitivity was 73.8, 54.8, 80.4, and 68.7%. Effective doses of ULD CT were similar to radiographs for dorsal and hip locations whereas for lumbar spine, ULD CT effective dose was 1.83 ± 0.59 mSv compared with 0.96 ± 0.59 mSv (p < 0.001). Sensitivity for fracture detection was higher for ULD CT compared with radiographs with an effective dose comparable to radiographs. • Ultra-low-dose spine and pelvis CT demonstrates better fracture detection when compared with radiographs. • The effective dose of ultra-low-dose spine and pelvis CT scan and radiographs is comparable. • Replacement of radiographs by ULD CT in daily practice for trauma patients is an option to consider and should be evaluated by a randomized trial.

Accuracy of after-hour ‘red dot’ trauma radiograph triage by radiographers in a South African regional hospital

The global demand for diagnostic imaging exceeds the supply of radiologists and is of particular significance in poorly resourced healthcare environments where many radiographs are unreported. Delayed or absent reporting may negatively impact patient management. In well-resourced countries there is recognition that extending the role of radiographers to radiological reporting tasks helps meet service demands. Aim. To determine the accuracy of acute fracture detection by South African radiographers working in an after-hour setting. We performed a retrospective study of radiographers at a Western Cape Regional Hospital over 2 months in 2011. The sensitivity and specificity of radiographers' fracture detection were compared with that of a consultant radiologist. Differences were evaluated using the McNemar chi-squared test, with p<0.05 regarded as significant. A total of 369 radiographs were analysed. The overall accuracy of reporting by radiographers was 93.7%, with 74.4% sensitivity for fracture detection. Experienced radiographers performed better than inexperienced radiographers; adult fractures were more consistently identified than paediatric fractures, and appendicular fractures were better visualised than axial fractures. In all instances there was a significant difference between fracture detection by radiographers and the radiologist. Experienced radiographers evaluating appendicular fractures in adults achieved the highest sensitivity (89.9%), which was not significantly different from that of a consultant radiologist (p=0.88). The performance of experienced radiographers in our study is comparable with that of experienced radiographers internationally, who have no specific training in trauma radiograph reporting. However, additional training is required if role extension is to be considered.

Accuracy of diagnosis of subtle chest disease and subtle fractures with a teleradiology system.

This study compared conventional thoracic and skeletal radiographs with 12-bit digitized images of the same radiographs in terms of subjective image quality and accuracy of diagnosis of subtle disease. Thirty-six chest radiographs with normal findings, 48 chest radiographs with abnormal findings, 29 skeletal radiographs with normal findings, and 26 skeletal radiographs with abnormal findings were included in this study. The images were chosen for the subtlety of their findings (nodules, pneumothoraces, interstitial lung disease, fractures). Each set of images was laser-digitized and viewed independently by three radiologists at a teleradiology workstation with a 1280 x 1024 pixel matrix monitor equipped with an eight-bit-per-pixel gray-scale display. All images were viewed at maximum resolution. After review of the digitized images for image quality and for the presence of abnormalities, each radiologist analyzed the analog images in a similar way. Sensitivities for detecting nodules, pneumothoraces, and interstitial lung disease on digitized chest radiographs were 58%, 75%, and 90%, respectively, compared with 62%, 79%, and 92%, respectively, on the original radiographs. Sensitivity for fracture detection on digitized bone radiographs was 87% compared with 88% on analog radiographs. None of these differences was statistically significant. Digitization of radiographs for primary diagnosis by teleradiology results in a slight decrease in sensitivity for detection of subtle abnormalities, provided that the images are viewed at maximum resolution.

Radiographically occult scaphoid fractures: value of MR imaging in detection.

To evaluate the diagnostic value of magnetic resonance (MR) imaging in patients with clinical suspicion of scaphoid fractures and normal initial plain radiographs. MR imaging was performed within 7 days after trauma in 42 patients with clinical suspicion of scaphoid fractures and normal plain radiographs. T1-weighted spin-echo, T2*-weighted gradient-echo, and short inversion time inversion-recovery (STIR) sequences were performed. MR images were evaluated independently by two radiologists. Six-week follow-up radiographs were used as a standard to diagnose fractures. MR imaging depicted occult fractures of the scaphoid bone in 14 patients (33%), the capitate bone in four (10%), the trapezium in one (2%), and the distal radius in two (5%). All wrist fractures were detected with a combination of STIR and T1-weighted spin-echo sequences. The sensitivity and specificity for detection of radiographically occult fractures of the wrist were 100% each for the first and 95% and 100%, respectively, for the second radiologist with an almost perfect interobserver agreement (K = 0.953). MR imaging has a high sensitivity for detection of fractures of the scaphoid bone and wrist not evident on plain radiographs and may enable early diagnosis and treatment.

Reporting of fracture radiographs by radiographers: the impact of a training programme.

In order to determine what influence training would have on their ability to interpret skeletal radiographs from the accident and emergency department, a 6 months training programme was established for three radiographers in various aspects of the radiology of orthopaedics and skeletal trauma. During the study the radiographers reported on radiographs from the accident and emergency department and each month an evaluation of their accuracy was undertaken. The overall radiographer error rate for fracture detection (false positive and false negative) declined during the training period. This was highly significant (p < 0.001). The sensitivity for fracture detection improved from 81.1% at the commencement of the trial to 95.9% at the end. This was also highly significant (p < 0.001). Radiographer specificity for the exclusion of fractures also improved from 94.4% during the first 2 months to 96.6% in the final 2 months, and this was also significant (p < 0.05). The overall error rate of two of the three radiographers improved significantly (p < 0.001) but for one radiographer the improvement did not reach a level of statistical significance. The difference in sensitivity for fracture detection at the commencement of the trial period between radiologist and radiographer was highly significant (p < 0.001), but there was no statistically significant difference during the last two months of the trial. The difference in specificity between radiologist and radiographer remained highly significant both at the beginning and the end of the trial (p < 0.001). Experienced radiographers who receive supplementary training in the radiology of skeletal trauma can significantly improve their diagnostic skills and can report such radiographs with a high degree of accuracy. A programme of training and certification of radiographers in fracture reporting could help alleviate the diagnostic radiologists' workload of plain film reporting.