Advancing Diagnostics and Patient Care: The Role of Biomarkers in Radiology.

The integration of biomarkers into medical practice has revolutionized the field of radiology, allowing for enhanced diagnostic accuracy, personalized treatment strategies, and improved patient care outcomes. This review offers radiologists a comprehensive understanding of the diverse applications of biomarkers in medicine. By elucidating the fundamental concepts, challenges, and recent advancements in biomarker utilization, it will serve as a bridge between the disciplines of radiology and epidemiology. Through an exploration of various biomarker types, such as imaging biomarkers, molecular biomarkers, and genetic markers, I outline their roles in disease detection, prognosis prediction, and therapeutic monitoring. I also discuss the significance of robust study designs, blinding, power and sample size calculations, performance metrics, and statistical methodologies in biomarker research. By fostering collaboration between radiologists, statisticians, and epidemiologists, I hope to accelerate the translation of biomarker discoveries into clinical practice, ultimately leading to improved patient care.

Including diverse and admixed populations in genetic epidemiology research.

The inclusion of ancestrally diverse participants in genetic studies can lead to new discoveries and is important to ensure equitable health care benefit from research advances. Here, members of the Ethical, Legal, Social, Implications (ELSI) committee of the International Genetic Epidemiology Society (IGES) offer perspectives on methods and analysis tools for the conduct of inclusive genetic epidemiology research, with a focus on admixed and ancestrally diverse populations in support of reproducible research practices. We emphasize the importance of distinguishing socially defined population categorizations from genetic ancestry in the design, analysis, reporting, and interpretation of genetic epidemiology research findings. Finally, we discuss the current state of genomic resources used in genetic association studies, functional interpretation, and clinical and public health translation of genomic findings with respect to diverse populations.

Opportunistic screening for osteoporosis and osteopenia from CT scans of the abdomen and pelvis using machine learning.

OBJECTIVES: To use multivariable machine learning using the computed tomography (CT) attenuation of each of the bones in the lumbar spine, pelvis, and sacrum, to predict osteoporosis/osteopenia. METHODS: This was a retrospective study of 394 patients aged 50 years or older with CT scans of the abdomen and pelvis and dual-energy x-ray absorptiometry (DXA) scans obtained within 6 months of each other. Volumetric segmentations were performed for each of the bones from L1-L4 vertebrae, pelvis, and sacrum to obtain the mean CT attenuation of each bone. The data was randomly split into training/validation (n = 274, 70%) and test (n = 120, 30%) datasets. The CT attenuation of the L1 vertebrae, univariate logistic regression, least absolute shrinkage and selection operator (LASSO), and support vector machines (SVM) with radial basis function (RBF) were used to predict osteoporosis/osteopenia. The performance of using the CT attenuation at L1 to the univariate logistic regression, LASSO, and SVM models were compared using DeLong's test in the test dataset. RESULTS: All CT attenuation measurements were predictive of osteoporosis/osteopenia (p < 0.001 for all). The SVM model (accuracy = 0.892, AUC = 0.886) outperformed the models using the CT attenuation of threshold of 173.9 Hounsfield units (HU) at L1 (accuracy = 0.725, AUC = 0.739, p = 0.010), the univariate logistic regression model (accuracy = 0.767, AUC = 0.533, p < 0.001) and the LASSO model (accuracy = 0.817, AUC = 0.711, p = 0.007) to predict osteoporosis/osteopenia. CONCLUSION: A SVM model using the CT attenuations of multiple bones within the lumbar spine and pelvis and clinical data has a better ability to predict osteoporosis/osteopenia than using the CT attenuation of L1 or a LASSO model. KEY POINTS: • Multivariable SVM model using the CT attenuation of multiple bones and clinical/demographic data was more predictive than using the CT attenuation at L1 only.

Intravenous tranexamic acid is associated with an increased risk of pulmonary embolism following sarcoma resection.

INTRODUCTION: Tranexamic acid (TXA) is an antifibrinolytic drug that has been shown to reduce blood loss following surgery. The use of TXA during orthopedic procedures has gained widespread acceptance, with multiple clinical studies demonstrating no increase in thrombotic complications. While TXA has been shown to be safe and effective for several orthopedic procedures, its use in orthopedic sarcoma surgery is not well established. Cancer-associated thrombosis remains a significant cause of morbidity and mortality in patients with sarcoma. It is unknown if intraoperative TXA use will increase the risk of developing a postoperative thrombotic complication in this population. This study aimed to compare the risk of postoperative thrombotic complications in patients who received TXA during sarcoma resection to patients who did not receive TXA. METHODS: A retrospective review was performed of 1099 patients who underwent resection of a soft tissue or bone sarcoma at our institution between 2010 and 2021. Baseline demographics and postoperative outcomes were compared between patients who did and did not receive intraoperative TXA. We evaluated 90-day complication rates, including: deep venous thrombosis (DVT), pulmonary embolism (PE), myocardial infarction (MI), cerebrovascular accident (CVA), and mortality. RESULTS: TXA was used more commonly for bone tumors (p < 0.001), tumors located in the pelvis (p = 0.004), and larger tumors (p < 0.001). Patients who received intraoperative TXA were associated with a significant increase in developing a postoperative DVT (odds ratio [OR]: 2.22, p = 0.036) and PE (OR: 4.62, p < 0.001), but had no increase in CVA, MI, or mortality (all p > 0.05) within 90 days of surgery, following univariate analysis. Multivariable analysis confirmed that TXA was independently associated with developing a postoperative PE (OR: 10.64, 95% confidence interval: 2.23-50.86, p = 0.003). We found no association with DVT, MI, CVA, or mortality within 90 days postoperatively, following intraoperative TXA use. CONCLUSION: Our results demonstrate a higher associated risk of PE following TXA use in sarcoma surgery and caution is warranted with TXA use in this patient population.

Radiologists should use the hip-knee-ankle angle rather than the mechanical axis deviation to describe knee alignment.

OBJECTIVE: Preoperative and postoperative coronal knee alignment is an important predictor of total knee arthroplasty (TKA) failure. Radiologists often report the mechanical axis deviation (MAD) rather than hip-knee-ankle angle (HKAA) to describe coronal knee alignment. The aim of this study is to evaluate (i) how well the MAD predicts the HKAA; (ii) if patient height and sex affect the performance of the MAD; and (iii) if the MAD could be measured faster than the HKAA. MATERIALS AND METHODS: Two hundred patients undergoing hip-to-ankle radiographs for TKA planning were retrospectively reviewed. The MAD and HKAA were measured using previously published methods by the Visage picture archiving and communication systems (PACS) tools. Receiver operator characteristic (ROC) curves were used to evaluate the performance of the MAD to predict HKAA by gender and height. The performance of a linear model was used to predict HKAA from MAD in a prospectively collected cohort of 40 patients. Paired t tests were used for the comparison of time measurement in MAD and HKAA in this cohort. RESULTS: MAD strongly correlated with HKAA (r = 0.99, p < 0.001); however, the performance of MAD differed by height (p = 0.005) and sex (p < 0.001). There was no significant difference in the time taken to measure HKAA versus MAD (p > 0.05). CONCLUSION: HKAA should be used instead of the MAD because it is more clinically relevant and takes the same amount of time to be measured.

Utilizing machine learning for opportunistic screening for low BMD using CT scans of the cervical spine.

BACKGROUND: Computed Tomography (CT) scans of the cervical spine are often performed to evaluate patients for trauma and degenerative changes of the cervical spine. We hypothesized that the CT attenuation of the cervical vertebrae can be used to identify patients who should be screened for osteoporosis. METHODS: A retrospective study of 253 patients (177 training/validation and 76 test) with unenhanced CT scans of the cervical spine and Dual-energy x-ray Absorbtiometry (DXA) studies within 12 months of each other was performed. Volumetric segmentation of C1-T1, clivus, and first ribs was performed to obtain the CT attenuation of each bone. The correlations of the CT attenuations between the bones and with DXA measurements were evaluated. Univariate receiver operator characteristic (ROC) analyses, and multivariate classifiers (Random Forest (RF), XGBoost, Naïve Bayes (NB), and Support Vector Machines (SVM)) analyzing the CT attenuation of all bones, were utilized to predict patients with osteopenia/osteoporosis and femoral neck bone mineral density (BMD) T-scores <-1. RESULTS: There were positive correlations between the CT attenuation of each bone, and with the DXA measurements. A CT attenuation threshold of 305.2 Hounsfield Units (HU) at C3 had the highest accuracy (0.763, AUC=0.814) to detect femoral neck BMD T-scores ≤-1 and a CT attenuation threshold of 323.6 HU at C3 had the highest accuracy (0.774, AUC=0.843) to detect osteopenia/osteoporosis. The SVM classifier (AUC=0.756) had higher AUC than the RF (AUC=0.692, P=0.224), XGBoost (AUC=0.736; P=0.814), NB (AUC=0.622, P=0.133) and CT threshold of 305.2 HU at C3 (AUC=0.704, P=0.531) classifiers to identify patients with femoral neck BMD T-scores <-1. The SVM classifier (accuracy=0.816) was more accurate than using the CT threshold of 305.2 HU at C3 (accuracy=0.671) (McNemar's χ12=7.55, P=0.006). CONCLUSION: Opportunistic screening for low BMD can be done using cervical spine CT scans. A SVM classifier was more accurate than using the CT threshold of 305.2 HU at C3.

Machine Learning for Opportunistic Screening for Osteoporosis and Osteopenia Using Knee CT Scans.

PURPOSE: To predict whether a patient has osteoporosis/osteopenia using the attenuation of trabecular bone obtained from knee computed tomography (CT) scans. METHODS: Retrospective analysis of 273 patients who underwent contemporaneous knee CT scans and dual-energy X-ray absorptiometry (DXA) within 1 year. Volumetric segmentation of the trabecular bone of the distal femur, proximal tibia, patella, and proximal fibula was performed to obtain the bone CT attenuation. The data was randomly split into training/validation (78%) and test (22%) datasets and the performance in the test dataset were evaluated. The predictive properties of the CT attenuation of each bone to predict osteoporosis/osteopenia were assessed. Multivariable support vector machines (SVM) and random forest classifiers (RF) were used to predict osteoporosis/osteopenia. RESULTS: Patients with a mean age (range) of 67.9 (50-87) years, 85% female were evaluated. Seventy-seven (28.2%) of patients had normal bone mineral density (BMD), 140 (51.3%) had osteopenia, and 56 (20.5%) had osteoporosis. The proximal tibia had the best predictive ability of all bones and a CT attenuation threshold of 96.0 Hounsfield Units (HU) had a sensitivity of .791, specificity of .706, and area under the curve (AUC) of .748. The AUC for the SVM with cubic kernel classifier (AUC = .912) was better than the RF classifier (AUC = .683, P < .001) and better than using the CT attenuation threshold of 96.0 HU at the proximal tibia (AUC = .748, P = .025). CONCLUSIONS: Opportunistic screening for osteoporosis/osteopenia can be performed using knee CT scans. Multivariable machine learning models are more predictive than the CT attenuation of a single bone.

Artificial intelligence for prediction of COVID-19 progression using CT imaging and clinical data.

OBJECTIVES: Early recognition of coronavirus disease 2019 (COVID-19) severity can guide patient management. However, it is challenging to predict when COVID-19 patients will progress to critical illness. This study aimed to develop an artificial intelligence system to predict future deterioration to critical illness in COVID-19 patients. METHODS: An artificial intelligence (AI) system in a time-to-event analysis framework was developed to integrate chest CT and clinical data for risk prediction of future deterioration to critical illness in patients with COVID-19. RESULTS: A multi-institutional international cohort of 1,051 patients with RT-PCR confirmed COVID-19 and chest CT was included in this study. Of them, 282 patients developed critical illness, which was defined as requiring ICU admission and/or mechanical ventilation and/or reaching death during their hospital stay. The AI system achieved a C-index of 0.80 for predicting individual COVID-19 patients' to critical illness. The AI system successfully stratified the patients into high-risk and low-risk groups with distinct progression risks (p < 0.0001). CONCLUSIONS: Using CT imaging and clinical data, the AI system successfully predicted time to critical illness for individual patients and identified patients with high risk. AI has the potential to accurately triage patients and facilitate personalized treatment. KEY POINT: • AI system can predict time to critical illness for patients with COVID-19 by using CT imaging and clinical data.

Impact of Tumor-Infiltrating Lymphocytes on Overall Survival in Merkel Cell Carcinoma.

BACKGROUND: Merkel cell carcinoma (MCC) is a rare and aggressive neuroendocrine carcinoma of the skin. As the clinical course can be variable, prognostic markers are needed to better stratify patients. Prior literature, composed of small series with limited sample size, has demonstrated that tumor-infiltrating lymphocytes (TILs) are an important prognostic marker in MCC. To validate these findings on a population level, we sought to analyze and report the prognostic value of TILs in a large national data set. MATERIALS AND METHODS: A retrospective observational cohort study was conducted of patients with nonmetastatic MCC from 2010 to 2015 using the National Cancer Database. Individual variables trending toward significance using a univariable analysis were included in a multivariable Cox proportional hazards model to assess their independent effect on overall survival (OS). TILs were subclassified into none, nonbrisk, and brisk and the survival analysis was performed. Propensity score-weighted multivariable analysis (PS MVA) was performed to adjust for additional confounding. RESULTS: A total of 2,182 patients met inclusion criteria: 611 (28.0%) were identified as having TILs present, and 1,571 (72.0%) had TILs absent in the tumor. On MVA, subdivision of TIL status into nonbrisk (hazard ratio [HR], 0.750; 95% confidence interval [CI], 0.602-0.933) and brisk (HR, 0.499; 95% CI, 0.338-0.735) was associated with incrementally improved OS compared with no TILs. The association of nonbrisk and brisk TILs with improved OS was retained on PS MVA (Nonbrisk: HR, 0.720; 95% CI, 0.550-0.944; Brisk: HR, 0.483; 95% CI, 0.286-0.814). CONCLUSION: The presence of nonbrisk and brisk TILs is associated with incrementally improved OS in patients with nonmetastatic MCC in a large national data set. This pathologic feature can aid with risk stratification, estimation of prognosis, and, importantly, decision-making with respect to treatment intensification in high-risk patients. IMPLICATIONS FOR PRACTICE: Merkel cell carcinoma (MCC) is an aggressive neuroendocrine cutaneous malignancy with variable clinical course. Prognostic markers are needed to better risk stratify patients. We present the largest retrospective observational cohort study of patients with nonmetastatic MCC using the National Cancer Database. Our analysis demonstrates an association between increasing degrees of tumor-infiltrating lymphocytes and incrementally improved survival. These conclusions improve pathologic risk stratification, and decision-making with respect to treatment intensification. Intensification may include adjuvant radiation therapy to the primary site after wide excision despite small tumor size, to the nodal basin in sentinel lymph node-negative patients, or offering closer follow-up.

Radiotherapy Remains Underused in the Treatment of Soft-Tissue Sarcomas: Disparities in Practice Patterns in the United States.

BACKGROUND: Practice patterns of radiation therapy (RT) use for soft-tissue sarcoma (STS) remain quite variable, despite clinical practice guidelines recommending the addition of RT to surgery for patients with high-grade STS, particularly for larger tumors. Using the National Cancer Database (NCDB), we assessed patterns of overall RT use, neoadjuvant versus adjuvant treatment, and specific RT modalities in this population. PATIENTS AND METHODS: Patients aged ≥18 years with stage II/III STS in 2004 through 2015 were identified from the NCDB. Patterns of care were assessed using multivariable logistic regression analysis. RESULTS: Of 27,426 total patients, 11,654 (42%) were treated with surgery alone versus 15,772 (58%) with RT in addition to surgery, with no overall increase in RT use over the study period. Notable clinical predictors of receipt of RT included tumor size (>5 cm), grade III, and tumors arising in the extremities. Conversely, female sex, older age (≥70 years), Black race, noncommercial insurance coverage, farther distance to treatment, and poor performance status were negative predictors of RT use. Of those receiving RT, 27% were treated with neoadjuvant RT and 73% with adjuvant RT. The proportion of those receiving neoadjuvant RT increased over time. Relevant factors associated with neoadjuvant RT included treatment at academic centers, larger tumor size, and extremity tumors. Of those who received RT with a modality specified as either intensity-modulated RT (IMRT) or 3D conformal RT (3DCRT), 61% were treated with IMRT and 39% with 3DCRT. The proportion of patients treated with IMRT increased over time. Relevant factors associated with IMRT use included treatment at academic centers, commercial insurance coverage, and larger and nonextremity tumors. CONCLUSIONS: Although use of neoadjuvant RT and IMRT has increased over time, a significant number of patients with STS are not receiving adjuvant or neoadjuvant RT. Our findings also note potential sociodemographic disparities and highlight the concern that not all patients with STS are being equally considered for RT.

The utility of 2-[18F]FDG-PET/CT maximum SUV versus CT attenuation for directing bone biopsies.

OBJECTIVES: This study aimed to compare the accuracy of PET/CT parameters with CT parameters for directing bone biopsies. METHODS: The study was an IRB-approved retrospective study of 388 patients who underwent both 2-[18F] FDG PET/CT and CT within 6 weeks before a bone biopsy. Age, sex, cancer type, lesion length, SUVmax, tumor to liver (T/L) ratio, CT attenuation, difference in CT attenuation between the lesion and normal bone (delta CT attenuation), and the absolute delta CT attenuation were used as predictors. T tests and chi-squared tests were used to compare variables. DeLong's test was used to compare receiver operator characteristic (ROC) curves. RESULTS: We reviewed the data from 388 patients. Of these, 295 patients had bone lesion biopsies, and 93 patients had bone marrow aspirations/biopsies. Biopsies of larger bone lesions (p = 0.033) and bone lesions with higher SUVmax (p = 0.005) were more likely to show malignancy. For bone lesions, the ROC curve for the SUVmax (AUC = 0.6827) was better than the ROC curves for delta CT attenuation (AUC = 0.5766, p = 0.032) and absolute delta CT attenuation (AUC = 0.5491, p = 0.006), but not significantly better than the ROC curves for CT attenuation (AUC = 0.5894, p = 0.061) and T/L ratio (AUC = 0.6778, p = 0.774). A threshold SUVmax of 5.25 had an accuracy of 0.713, sensitivity of 0.766, and specificity of 0.549 to predict malignancy in bone lesion biopsies. None of these variables predicted malignancy in bone marrow biopsies (p > 0.05 for all). CONCLUSIONS: Metabolic 2-[18F]FDG PET/CT parameters have more clinical impact for planning bone biopsies as compared to CT parameters. KEY POINTS: • The 2-[18F]FDG PET/CT measurement (SUVmax) has more clinical impact for planning bone biopsies as compared to CT measurements. • Neither the change in CT attenuation of the lesion relative to normal bone nor the absolute value of this change was a significant predictor of malignancy. • 2-[18F]FDG PET/CT may have clinical benefit and an additional role in directing bone biopsies.

Artificial Intelligence Augmentation of Radiologist Performance in Distinguishing COVID-19 from Pneumonia of Other Origin at Chest CT.

Background Coronavirus disease 2019 (COVID-19) and pneumonia of other diseases share similar CT characteristics, which contributes to the challenges in differentiating them with high accuracy. Purpose To establish and evaluate an artificial intelligence (AI) system for differentiating COVID-19 and other pneumonia at chest CT and assessing radiologist performance without and with AI assistance. Materials and Methods A total of 521 patients with positive reverse transcription polymerase chain reaction results for COVID-19 and abnormal chest CT findings were retrospectively identified from 10 hospitals from January 2020 to April 2020. A total of 665 patients with non-COVID-19 pneumonia and definite evidence of pneumonia at chest CT were retrospectively selected from three hospitals between 2017 and 2019. To classify COVID-19 versus other pneumonia for each patient, abnormal CT slices were input into the EfficientNet B4 deep neural network architecture after lung segmentation, followed by a two-layer fully connected neural network to pool slices together. The final cohort of 1186 patients (132 583 CT slices) was divided into training, validation, and test sets in a 7:2:1 and equal ratio. Independent testing was performed by evaluating model performance in separate hospitals. Studies were blindly reviewed by six radiologists without and then with AI assistance. Results The final model achieved a test accuracy of 96% (95% confidence interval [CI]: 90%, 98%), a sensitivity of 95% (95% CI: 83%, 100%), and a specificity of 96% (95% CI: 88%, 99%) with area under the receiver operating characteristic curve of 0.95 and area under the precision-recall curve of 0.90. On independent testing, this model achieved an accuracy of 87% (95% CI: 82%, 90%), a sensitivity of 89% (95% CI: 81%, 94%), and a specificity of 86% (95% CI: 80%, 90%) with area under the receiver operating characteristic curve of 0.90 and area under the precision-recall curve of 0.87. Assisted by the probabilities of the model, the radiologists achieved a higher average test accuracy (90% vs 85%, Δ = 5, P < .001), sensitivity (88% vs 79%, Δ = 9, P < .001), and specificity (91% vs 88%, Δ = 3, P = .001). Conclusion Artificial intelligence assistance improved radiologists' performance in distinguishing coronavirus disease 2019 pneumonia from non-coronavirus disease 2019 pneumonia at chest CT. © RSNA, 2020 Online supplemental material is available for this article.

Structured mating: Patterns and implications.

Genetic similarity of spouses can reflect factors influencing mate choice, such as physical/behavioral characteristics, and patterns of social endogamy. Spouse correlations for both genetic ancestry and measured traits may impact genotype distributions (Hardy Weinberg and linkage equilibrium), and therefore genetic association studies. Here we evaluate white spouse-pairs from the Framingham Heart Study (FHS) original and offspring cohorts (N = 124 and 755, respectively) to explore spousal genetic similarity and its consequences. Two principal components (PCs) of the genome-wide association (GWA) data were identified, with the first (PC1) delineating clines of Northern/Western to Southern European ancestry and the second (PC2) delineating clines of Ashkenazi Jewish ancestry. In the original (older) cohort, there was a striking positive correlation between the spouses in PC1 (r = 0.73, P = 3x10(-22)) and also for PC2 (r = 0.80, P = 7x10(-29)). In the offspring cohort, the spouse correlations were lower but still highly significant for PC1 (r = 0.38, P = 7x10(-28)) and for PC2 (r = 0.45, P = 2x10(-39)). We observed significant Hardy-Weinberg disequilibrium for single nucleotide polymorphisms (SNPs) loading heavily on PC1 and PC2 across 3 generations, and also significant linkage disequilibrium between unlinked SNPs; both decreased with time, consistent with reduced ancestral endogamy over generations and congruent with theoretical calculations. Ignoring ancestry, estimates of spouse kinship have a mean significantly greater than 0, and more so in the earlier generations. Adjusting kinship estimates for genetic ancestry through the use of PCs led to a mean spouse kinship not different from 0, demonstrating that spouse genetic similarity could be fully attributed to ancestral assortative mating. These findings also have significance for studies of heritability that are based on distantly related individuals (kinship less than 0.05), as we also demonstrate the poor correlation of kinship estimates in that range when ancestry is or is not taken into account.

Computer-Aided Detection of Incidental Lumbar Spine Fractures from Routine Dual-Energy X-Ray Absorptiometry (DEXA) Studies Using a Support Vector Machine (SVM) Classifier.

To assess whether application of a support vector machine learning algorithm to ancillary data obtained from posterior-anterior dual-energy X-ray absorptiometry (DEXA) studies could identify patients with lumbar spine (L1-L4) vertebral body fractures without additional DEXA imaging or radiation. Three hundred seven patients (199 without any fractures of the spine, and 108 patients with at least one fracture of the L1, L2, L3, or L4 vertebral bodies) who had DEXA studies were evaluated. Ancillary data from DEXA output was analyzed. The dataset was split into training (80%) and test (20%) datasets. Support vector machines (SVMs) with 10-fold cross-validation and different kernels were used to identify the best kernel based on the greatest area under the curve (AUC) and the best training vectors in the training dataset. The SVM with the best kernel was then applied to the test dataset to assess the accuracy of the SVM. Receiver operating characteristic (ROC) curves of the SVMs using different kernels in the test dataset were compared using DeLong's test. The SVM classifier with the linear kernel had the greatest AUC in the training dataset (AUC = 0.9258). The AUC of the SVM classifier with the linear kernel in the test dataset was 0.8963. The SVM classifier with the linear kernel had an overall average accuracy of 91.8% in the test dataset. The sensitivity, specificity, positive predictive value, and negative predictive of the SVM classifier with the linear kernel to detect lumbar spine fractures were 81.8%, 97.4%, 94.7%, and 90.5%, respectively. The SVM classifier with the linear kernel ROC curve had a significantly better AUC than the SVM classifier with the cubic polynomial kernel (P = 0.034) for discriminating between patients with lumbar spine fractures and control patients, but not significantly different from the SVM classifier with a radial basis function (RBF) kernel (P = 0.317) or the SVM classifier with a sigmoid kernel (P = 0.729). All fractures identified by the SVM classifiers were not prospectively identified by the radiologist. SVM analysis of ancillary data obtained from routine DEXA studies can identify lumbar spine fractures without the use of vertebral fracture assessment (VFA) DEXA imaging or radiation, and identify fractures missed by radiologists.

Radiomic features from MRI distinguish myxomas from myxofibrosarcomas.

BACKGROUND: Myxoid tumors pose diagnostic challenges for radiologists and pathologists. All myxoid tumors can be differentiated from each other using fluorescent in-situ hybridization (FISH) or immunohistochemical markers, except for myxomas and myxofibrosarcomas. Myxomas and myxofibrosarcomas are rare tumors. Myxomas are benign and histologically bland, whereas myxofibrosarcomas are malignant and histologically heterogenous. Because of the histological heterogeneity, low grade myxofibrosarcomas may be mistaken for myxomas on core needle biopsies. We evaluated the performance of T1-weighted signal intensity (T1SI), tumor volume, and radiomic features extracted from magnetic resonance imaging (MRI) to differentiate myxomas from myxofibrosarcomas. METHODS: The MRIs of 56 patients (29 with myxomas, 27 with myxofibrosarcomas) were analyzed. We extracted 89 radiomic features. Random forests based classifiers using the T1SI, volume features, and radiomic features were used to differentiate myxomas from myxofibrosarcomas. The classifiers were validated using a leave-one-out cross-validation. The performances of the classifiers were then compared. RESULTS: Myxomas had lower normalized T1SI than myxofibrosaromas (p = 0.006) and the AUC using the T1SI was 0.713. However, the classification model using radiomic features had an AUC of 0.885 (accuracy = 0.839, sensitivity = 0.852, specificity = 0.828), and outperformed the classification models using T1SI (AUC = 0.713) and tumor volume (AUC = 0.838). The classification model using radiomic features was significantly better than the classifier using T1SI values (p = 0.039). CONCLUSIONS: Myxofibrosarcomas are on average higher in T1-weighted signal intensity than myxomas. Myxofibrosarcomas are larger and have shape differences compared to myxomas. Radiomic features performed best for differentiating myxomas from myxofibrosarcomas compared to T1-weighted signal intensity and tumor volume features.

Accuracy of CT Attenuation Measurement for Differentiating Treated Osteoblastic Metastases From Enostoses.

OBJECTIVE: The objective of our study was to assess whether the maximum and mean CT attenuations are accurate for differentiating between enostoses and treated sclerotic metastases. MATERIALS AND METHODS: We retrospectively reviewed CT studies of 165 patients (167 lesions) that included 49 patients with 49 benign lesions, 69 patients with 71 sclerotic treated lesions, and 47 patients with 47 untreated lesions, and calculated the mean and maximum CT attenuations of each lesion. ROC curves were used to identify thresholds for differentiating enostoses from treated sclerotic metastases and from untreated sclerotic metastases. RESULTS: The maximum CT attenuation of enostoses (1212.0 HU) was higher from that of untreated (754.7 HU) (p = 9.7 × 10-16) and that of treated (891.7 HU) (p = 9.9 × 10-10) sclerotic metastases. The maximum CT attenuation of treated sclerotic metastases (891.7 HU) was higher than that of untreated sclerotic metastases (754.7 HU) (p = 0.003). Enostoses had higher mean CT attenuation (1123.0 HU) than untreated (602.0 HU) (p < 2.2 × 10-16) and treated (731.7 HU) (p = 9.6 × 10-15) sclerotic metastases. A threshold mean CT attenuation of 885 HU had an accuracy of 91.7% and 81.7% to differentiate enostoses from untreated and treated metastases, respectively, whereas a threshold maximum CT attenuation of 1060.0 HU had an accuracy of 81.3% and 72.5% to differentiate enostoses from untreated and treated metastases. CONCLUSION: The mean and maximum CT attenuations can differentiate between enostoses and sclerotic metastases; however, the accuracy of both metrics decreases after treatment.

Intraosseous pseudomeningocele of the mobile spine: a case report and review of the literature.

BACKGROUND: Pseudomeningoceles most commonly occur due to prior trauma or surgery and are often located in the posterior paraspinous tissues. Here, we report a case of an intraosseous pseudomeningocele that mimicked an intra-osseous T2 hyperintense lesion in the L1 vertebral body. CASE DESCRIPTION: A 64-year-old male presented with back, left lateral thigh and left knee pain lasting several months. He had no prior history of trauma or surgery. Radiographs of the lumbar spine showed mild levoscoliotic curvature of the lumbar spine, Baastrup's changes between the spinous processes, multilevel degenerative disc disease and facet arthropathy. Magnetic resonance imaging (MRI) of the lumbar spine performed without intravenous contrast showed severe spinal canal stenosis from L1-L2 to L3-L4 and moderate spinal canal stenosis at L4-L5. MRI also showed a 2.5-cm T2 hyperintense lesion involving the posterior aspect of the L1 vertebral body, with questionable contiguity with cerebrospinal fluid. Computed tomography (CT) myelogram was performed instead of biopsy. CT myelogram showed contiguity of the lesion with the intrathecal contrast and a rent in the posterior longitudinal ligament and anterior dura consistent with an intraosseous pseudomeningocele. The patient opted for non-operative management of the pseudomeningocele and his lumbar stenosis due to medical comorbidities. CONCLUSIONS: This case illustrates a rare case of an intra-osseous pseudomeningocele and highlights the importance of CT myelogram for diagnosis.