Deep learning for osteoporosis screening using an anteroposterior hip radiograph image.

PURPOSE: Osteoporosis is a common bone disorder characterized by decreased bone mineral density (BMD) and increased bone fragility, which can lead to fractures and eventually cause morbidity and mortality. It is of great concern that the one-year mortality rate for osteoporotic hip fractures could be as high as 22%, regardless of the treatment. Currently, BMD measurement is the standard method for osteoporosis diagnosis, but it is costly and requires special equipment. While a plain radiograph can be obtained more simply and inexpensively, it is not used for diagnosis. Deep learning technologies had been applied to various medical contexts, yet few to osteoporosis unless they were trained on the advanced investigative images, such as computed tomography. The purpose of this study was to develop a deep learning model using the anteroposterior hip radiograph images and measure its diagnostic accuracy for osteoporosis. METHODS: We retrospectively collected all anteroposterior hip radiograph images of patients from 2013 to 2021 at a tertiary care hospital. The BMD measurements of the included patients were reviewed, and the radiograph images that had a time interval of more than two years from the measurements were excluded. All images were randomized using a computer-generated unequal allocation into two datasets, i.e., 80% of images were used for the training dataset and the remaining 20% for the test dataset. The T score of BMD obtained from the ipsilateral femoral neck of the same patient closest to the date of the performed radiograph was chosen. The T score cutoff value of - 2.5 was used to diagnose osteoporosis. Five deep learning models were trained on the training dataset, and their diagnostic performances were evaluated using the test dataset. Finally, the best model was determined by the area under the curves (AUC). RESULTS: A total of 363 anteroposterior hip radiograph images were identified. The average time interval between the performed radiograph and the BMD measurement was 6.6 months. Two-hundred-thirteen images were labeled as non-osteoporosis (T score > - 2.5), and the other 150 images as osteoporosis (T score ≤ - 2.5). The best-selected deep learning model achieved an AUC of 0.91 and accuracy of 0.82. CONCLUSIONS: This study demonstrates the potential of deep learning for osteoporosis screening using anteroposterior hip radiographs. The results suggest that the deep learning model might potentially be used as a screening tool to find patients at risk for osteoporosis to perform further BMD measurement.

Leukoaraiosis and stroke severity scores in post-rtPA intracerebral haemorrhage.

Introduction: Post thrombolytic symptomatic intracerebral haemorrhage (sICH) is a major concern in patients who had acute ischaemic stroke. Leukoaraiosis (LA) is reported to be related with sICH after intravenous thrombolytic treatment. However, the influence of LA and stroke neurological and imaging severity scores is still debated. Objective: To evaluate if LA or severity scores are related with sICH in patients who had acute ischaemic stroke who received thrombolytic therapy. And, predictors for sICH were also studied with adjustment of baseline severity scores. Methods: This was a retrospective, analytical study. The inclusion criteria were adult patients diagnosed as acute ischaemic stroke who received the recombinant tissue plasminogen activator (rtPA) treatment within 4.5 hours. The study period was between May 2007 and November 2016. Predictors for sICH were determined using logistic regression analysis. Results: During the study period, there were 504 eligible patients. Of those, 45 patients (8.92%) had sICH. Among nine factors in the final model for predicting sICH, there were four independent factors including previous antiplatelet therapy, previous anticoagulant therapy, presence of LA and hyperdense artery sign. The highest adjusted OR was previous anticoagulant therapy (5.08 with 95% CI of 1.18 to 11.83), while the LA factor had adjusted OR (95% CI) of 2.52 (1.01 to 6.30). Conclusions: LA, hyperdense artery sign, previous antiplatelet therapy and previous anticoagulant therapy were associated with post-rtPA sICH. Further studies are required to confirm the results of this study.

The diagnostic accuracy of mammography and ultrasonography for recurrent breast cancer after breast conserving treatment.

Objective: To evaluate the performance of mammography and breast ultrasonography to diagnose tumor recurrence in patients after breast conserving therapy. Material and Methods: Imaging findings of 130 breast cancer patients treated by breast conserving therapy (BCT) who have followed up with mammography and ultrasonography at our center between 1 st January 2010 and 1st January 2016 were interpreted by two radiologists. The information of recurrent tumor and baseline data were blinded. Imaging interpretation followed the ACR Breast imaging-reporting and data system (BI-RADS) 5th edition guideline. Findings of mammography, breast ultrasonography, demographic data and histological data were recorded and analyzed. Results: The presence of mass in mammography (P-value=0.025) and internal vascularity in mass in ultrasonography (P-value<0.001) were associated with recurrent tumor at the surgical bed. All the recurrent tumors were interpreted as BI-RADS 4 (71 patients) with sensitivity= 100%, specificity= 89.5%. BIRADS4 is significant in the diagnosis of recurrent breast cancer in BCT patients (AUC of the ROC curve = 0.742 and 95% CI=(0.7-0.79)). Conclusion: The presence of mass in mammography and internal vascularity in the mass in ultrasonography are the imaging findings which were significantly related to recurrent tumor at surgical bed in patient with breast conserving treatment.

BiTNet: Hybrid deep convolutional model for ultrasound image analysis of human biliary tract and its applications.

Certain life-threatening abnormalities, such as cholangiocarcinoma, in the human biliary tract are curable if detected at an early stage, and ultrasonography has been proven to be an effective tool for identifying them. However, the diagnosis often requires a second opinion from experienced radiologists, who are usually overwhelmed by many cases. Therefore, we propose a deep convolutional neural network model, named biliary tract network (BiTNet), developed to solve problems in the current screening system and to avoid overconfidence issues of traditional deep convolutional neural networks. Additionally, we present an ultrasound image dataset for the human biliary tract and demonstrate two artificial intelligence (AI) applications: auto-prescreening and assisting tools. The proposed model is the first AI model to automatically screen and diagnose upper-abdominal abnormalities from ultrasound images in real-world healthcare scenarios. Our experiments suggest that prediction probability has an impact on both applications, and our modifications to EfficientNet solve the overconfidence problem, thereby improving the performance of both applications and of healthcare professionals. The proposed BiTNet can reduce the workload of radiologists by 35% while keeping the false negatives to as low as 1 out of every 455 images. Our experiments involving 11 healthcare professionals with four different levels of experience reveal that BiTNet improves the diagnostic performance of participants of all levels. The mean accuracy and precision of the participants with BiTNet as an assisting tool (0.74 and 0.61, respectively) are statistically higher than those of participants without the assisting tool (0.50 and 0.46, respectively (p<0.001)). These experimental results demonstrate the high potential of BiTNet for use in clinical settings.

Artificial intelligence (AI) vs. human in hip fracture detection.

Objective: This study aimed to assess the diagnostic accuracy and sensitivity of a YOLOv4-tiny AI model for detecting and classifying hip fractures types. Materials and methods: In this retrospective study, a dataset of 1000 hip and pelvic radiographs was divided into a training set consisting of 450 fracture and 450 normal images (900 images total) and a testing set consisting of 50 fracture and 50 normal images (100 images total). The training set images were each manually augmented with a bounding box drawn around each hip, and each bounding box was manually labeled either (1) normal, (2) femoral neck fracture, (3) intertrochanteric fracture, or (4) subtrochanteric fracture. Next, a deep convolutional neural network YOLOv4-tiny AI model was trained using the augmented training set images, and then model performance was evaluated with the testing set images. Human doctors then evaluated the same testing set images, and the performances of the model and doctors were compared. The testing set contained no crossover data. Results: The resulting output images revealed that the AI model produced bounding boxes around each hip region and classified the fracture and normal hip regions with a sensitivity of 96.2%, specificity of 94.6%, and an accuracy of 95%. The human doctors performed with a sensitivity ranging from 69.2 to 96.2%. Compared with human doctors, the detection rate sensitivity of the model was significantly better than a general practitioner and first-year residents and equivalent to specialist doctors. Conclusions: This model showed hip fracture detection sensitivity comparable to well-trained radiologists and orthopedists and classified hip fractures highly accurately.

Diagnostic accuracy of deep learning for evaluation of C-spine injury from lateral neck radiographs.

Background: Traumatic spinal cord injury (TSI) is a leading cause of morbidity and mortality worldwide, with the cervical spine being the most affected. Delayed diagnosis carries a risk of morbidity and mortality. However, cervical spine CT scans are time-consuming, costly, and not always available in general care. In this study, deep learning was used to assess and improve the detection of cervical spine injuries on lateral radiographs, the most widely used screening method to help physicians triage patients quickly and avoid unnecessary CT scans. Materials and methods: Lateral neck or lateral cervical spine radiographs were obtained for patients who underwent CT scan of cervical spine. Ground truth was determined based on CT reports. CiRA CORE, a codeless deep learning program, was used as a training and testing platform. YOLO network models, including V2, V3, and V4, were trained to detect cervical spine injury. The diagnostic accuracy, sensitivity, and specificity of the model were calculated. Results: A total of 229 radiographs (129 negative and 100 positive) were selected for inclusion in our study from a list of 625 patients with cervical spine CT scans, 181 (28.9%) of whom had cervical spine injury. The YOLO V4 model performed better than the V2 or V3 (AUC = 0.743), with sensitivity, specificity, and accuracy of 80%, 72% and 75% respectively. Conclusion: Deep learning can improve the accuracy of lateral c-spine or neck radiographs. We anticipate that this will assist clinicians in quickly triaging patients and help to minimize the number of unnecessary CT scans.

Prevalence of Sclerotic Pterygoid Plate in Pretreatment Nasopharyngeal Carcinoma.

OBJECTIVES: The objective of our study was to determine the prevalence of sclerotic pterygoid plate in pretreatment CT of nasopharyngeal carcinoma compared with the control group. MATERIALS AND METHODS: A total of 51 nasopharyngeal carcinoma patients (37 men, 14 women) with a mean age of 51.94±13 years, and 51 controls (30 men, 21 women) with a mean age, 49.31±15 years were included in this study in a retrospective fashion. All computed tomographic (CT) images were evaluated by two neuroradiologists. Sclerosis of pterygoid plate and other findings included pterygoid plate erosion, adjacent tumor enhancement, and parapharyngeal extension which were assessed. MRI findings were also recorded. The prevalence of pterygoid plate sclerosis was compared using Chi-square statistical tests. Imaging findings were analyzed by binary logistic regression analyses. RESULTS: The prevalence of pterygoid plate sclerosis in nasopharyngeal carcinoma was 53.9% compared to the control group (16.7%) and the difference was statistically significant (P-value< 0.001). In nasopharyngeal carcinoma, the prevalence of tumor adjacent to the pterygoid plate, parapharyngeal extension and pterygoid plate erosion were 69.6%, 81.4%, 38.2%, respectively. No erosion of pterygoid plate was detected in the control group. The odds of adjacent tumor enhancement and pterygoid plate erosion was 7.29 and 20.56 times higher in the sclerotic pterygoid plate (p-values of 0.019 and 0.000, respectively). MRI was available for four nasopharyngeal carcinoma cases with five sclerotic pterygoid plates, where two showed enhancements. All non-sclerotic pterygoid plates showed no enhancement on MRI. CONCLUSION: The prevalence of sclerotic pterygoid plate is significantly higher in patients with nasopharyngeal carcinoma with a considerably higher chance of adjacent tumor enhancement and pterygoid plate erosion.

The Turtle Neck Sign: Identification of Severe Retracted Distal Biceps Tendon Rupture.

BACKGROUND: Chronic tendon retraction subsequent to distal biceps tendon rupture significantly increases repair difficulty and potential for tendon grafting. Biceps tendons that appear short or absent with magnetic resonance imaging (MRI) or that cannot be readily identified at surgery may erroneously be classified as irreparable. These apparent "absent" biceps tendons may actually be retracted and curled up inside the muscle, visually resembling the head-neck of a turtle retracted inside its shell (the "turtle neck sign"). When located, these tendons could be unfolded and repaired primarily. This type of tendon retraction seems to be associated with high-degree ruptures and larcertus fibrosus tears. PURPOSE: To test the hypothesis that tendon retractions with a turtle neck sign on MRI are more associated with high-degree ruptures and larcertus fibrosus tears versus tendon tears with simple linear retraction. STUDY DESIGN: Cross-sectional study; Level of evidence, 3. METHODS: Retracted distal biceps tendon ruptures on sagittal MRI were categorized as linear retraction or curled-up (turtle neck) retraction. Retraction length, injury severity, and lacertus fibrosus tears were analyzed. RESULTS: The authors retrospectively analyzed the patient records of 85 consecutive traumatic distal biceps tendon ruptures from 2003 to 2019; the final study cohort was 37 patients. Injury-to-surgery timing was as follows: <3 weeks, 43% (16 cases); 3 weeks to 3 months, 32% (12 cases); and >3 months, 24% (9 cases). Overall, 19 patients had linear retraction <7 cm (mean, 3.3 ± 1.9 cm) and 18 patients had a turtle neck retraction ≥7 cm (mean, 9.1 ± 1.6 cm). The injury-to-surgery time (median [± interquartile range]) was 27 days (±90 days) in the linear retraction group and 23 days (±65 days) in the turtle neck retraction group. The turtle neck retraction group had a significantly higher occurrence of abnormal hook test findings, complete distal biceps tendon rupture, and lacertus fibrosus tears compared with the linear retraction group (100% vs 58%, 100% vs 68%, and 100% vs 37%, respectively; P ≤ .02). However, significant repairability differences were not found. CONCLUSION: Highly retracted distal biceps turtle neck sign tendon ruptures occur frequently in association with high-degree ruptures and lacertus fibrosus tears. The presence of a turtle neck retraction did not affect reparability. Surgeons should be aware of this curled-up retraction to avoid mistaking it for an absent tendon or a muscle-tendon disruption.

Thyroid Nodule Size as a Predictor of Malignancy in Follicular and Hurthle Neoplasms.

INTRODUCTION: The management of follicular (FN) and Hurthle cell neoplasms (HCN) is often difficult because of the uncertainty of malignancy risk. We aimed to assess characteristics of benign and malignant follicular and Hurthle neoplasms based on their shape and size. MATERIALS AND METHODS: Patients with Follicular adenoma (FA) or carcinoma (FC) and Hurthle Cell adenoma (HCA) or carcinoma (HCC) who had preoperative ultrasonography were included. Demographic data were retrieved. Size and shape of the nodules were measured. Logistic regression analyses and odds ratios were performed. RESULTS: A total of 115 nodules with 57 carcinomas and 58 adenomas were included. Logistic regression analysis shows that the nodule height and the patient age are predictors of malignancy (p-values = 0.001 and 0.042). A cutoff value of nodule height ≥ 4 cm. produces an odds ratio of 4.5 (p-value = 0.006). An age ≥ 55 year-old demonstrates an odds ratio of 2.4-3.6 (p-value = 0.03). Taller-than-wide shape was not statistically significant (p-value = 0.613). CONCLUSION: FC and HCC are larger than FA and HCA in size, with a cutoff at 4 cm. Increasing age increases the odds of malignancy with a cutoff at 55 year-old. Taller-than-wide shape is not a predictor of malignancy.

Evaluation of Perienhancing Area in Differentiation between Glioblastoma and Solitary Brain Metastasis.

PURPOSE: Accurate differential diagnosis between glioblastoma and brain metastasis is important. We aimed to differentiate these tumors by evaluation of the perienhancing area. MATERIALS AND METHODS: Thirty patients with glioblastoma and solitary brain metastasis were included. The diameters of perienhancing and enhancing areas were measured, and the percentage of enhancing area was calculated. We measured Apparent diffusion coefficient (ADC) of perienhancing and enhancing areas. Intratumoral necrotic areas were measured. RESULTS: The enhancing area of glioblastoma was 56.61% and metastasis was 42.55% (p = 0.08). The ADC values of the perienhancing part of GBM was 0.7 and metastasis was 0.79 (p = 0.052). The ADC value of the enhancing part of the GBM was 0.82 and metastasis was 0.8 (p-value = 0.72).  The intratumoral necrotic area of glioblastoma (152.25 mm3) was higher than in metastasis (0 mm3) (p-value = 0.003) with a cutoff area of 11.8 mm2. CONCLUSION: The ADC values of the perienhancing area were lower in glioblastoma with a near-significant p-value. Other perienhancing parameters demonstrated no significant difference between both tumors. The intratumoral necrotic area of glioblastoma is larger than metastasis.
.