Point-of-Care Ultrasound for a Deep Venous Thrombosis

Evaluation of Machine Learning Algorithms for Early Diagnosis of Deep Venous Thrombosis

Deep venous thrombosis (DVT) is a disease that must be diagnosed quickly, as it can trigger the death of patients. Nowadays, one can find different ways to determine it, including clinical scoring, D-dimer, ultrasonography, etc. Recently, scientists have focused efforts on using machine learning (ML) and neural networks for disease diagnosis, progressively increasing the accuracy and efficacy. Patients with suspected DVT have no apparent symptoms. Using pattern recognition techniques, aiding good timely diagnosis, as well as well-trained ML models help to make good decisions and validation. The aim of this paper is to propose several ML models for a more efficient and reliable DVT diagnosis through its implementation on an edge device for the development of instruments that are smart, portable, reliable, and cost-effective. The dataset was obtained from a state-of-the-art article. It is divided into 85% for training and cross-validation and 15% for testing. The input data in this study are the Wells criteria, the patient’s age, and the patient’s gender. The output data correspond to the patient’s diagnosis. This study includes the evaluation of several classifiers such as Decision Trees (DT), Extra Trees (ET), K-Nearest Neighbor (KNN), Multi-Layer Perceptron Neural Network (MLP-NN), Random Forest (RF), and Support Vector Machine (SVM). Finally, the implementation of these ML models on a high-performance embedded system is proposed to develop an intelligent system for early DVT diagnosis. It is reliable, portable, open source, and low cost. The performance of different ML algorithms was evaluated, where KNN achieved the highest accuracy of 90.4% and specificity of 80.66% implemented on personal computer (PC) and Raspberry Pi 4 (RPi4). The accuracy of all trained models on PC and Raspberry Pi 4 is greater than 85%, while the area under the curve (AUC) values are between 0.81 and 0.86. In conclusion, as compared to traditional methods, the best ML classifiers are effective at predicting DVT in an early and efficient manner.

Incremental value of compression ultrasound sonography in the emergency department

The quick evaluation of venous thromboembolism is a key point of modern medicine since the delayed diagnosis is associated with a worse prognosis. Venous ultrasound (VU) is a sensitive and rapidly performed test in cases of suspected deep venous thrombosis. Various protocols have been proposed for its execution, such as the study of the whole deep venous circulation of the lower limb or the analysis of the femoral-popliteal area. The aim is to detect a vessel thrombus and the most sensitive element is the non-compressibility with the probe. Initially, the thrombus is hypoechogenic and adherent to the vessel; later, it tends to organize and recanalize. Usually, in the early stages, the risk of embolism is higher. The role of studying the iliac axis and calf veins is still uncertain. VU is not useful for assessing response to anticoagulation therapy and it is unclear whether the persistence of thrombotic abnormalities can guide on a possible prolongation of therapy.

Diagnostic Value of Neutrophil Lymphocyte Ratio and D-Dimer as Biological Markers of Deep Vein Thrombosis in Patients Presenting with Unilateral Limb Edema

Introduction

Patients with deep vein thrombosis (DVT) pose high morbidity and mortality risk thus needing fast and accurate diagnosis. Wells clinical prediction scores with D-dimer testing are traditionally used to rule out patients with low probability of DVT. However, D-dimer testing has a few limitations regarding its relatively low specificity. Neutrophil-lymphocyte ratio (NLR), a marker of inflammation, was found to increase in DVT. Hence, we aimed to evaluate the role of NLR for DVT diagnosis.

Methods

Data were collected from medical records of patients with suspected DVT at Cipto Mangunkusumo National General Hospital during January–December 2014. Diagnosis of DVT was conducted using lower limb ultrasonography. Diagnostic values for NLR, D-dimer, and NLR + D-dimer were determined by receiver operating characteristic (ROC) analysis to obtain area under the curve (AUC), sensitivity, specificity, negative predictive value, and positive predictive values. Sensitivity and specificity analyses of NLR and D-dimer were also conducted based on Wells score and divided into groups of low and high probability of DVT.

Results

The AUC values for NLR, D-dimer, and NLR + D-dimer were 72.6%, 70.4%, and 76.1%, respectively. The optimal cut-off value determined for NLR was 5.12 with sensitivity of 67.7%, specificity of 67.9%, PPV of 68.85%, and NPV of 64.91% in differentiating subjects with and without DVT. This study also found that D-dimer had sensitivity of 69.4%, specificity of 71.4%, PPV of 72.88%, and NPV of 67.8%. Meanwhile, the NLR + D-dimer combination had sensitivity of 66.1% and specificity of 72.6%. Multivariate analysis showed that NLR (OR: 2.636; 95% CI: 1.144–6.076; p: 0.023) and D-dimer (OR: 4.175; 95% CI: 1.810–9.633; p: 0.001) were associated with DVT.

Conclusion

NLR value has wider AUC than D-Dimer and is relatively easier to obtain and does not require specific assay, thus enabling rapid evaluation of symptomatic patients suspected of having DVT. Adding NLR to D-dimer increased AUC to detect DVT. Therefore, NLR could serve as a complementary diagnostic tool for D-dimer to exclude DVT, especially in low clinical probability patients.

An Artificial Neural Network Approach and a Data Augmentation Algorithm to Systematize the Diagnosis of Deep-Vein Thrombosis by Using Wells’ Criteria

The use of a back-propagation artificial neural network (ANN) to systematize the reliability of a Deep Vein Thrombosis (DVT) diagnostic by using Wells’ criteria is introduced herein. In this paper, a new ANN model is proposed to improve the Accuracy when dealing with a highly unbalanced dataset. To create the training dataset, a new data augmentation algorithm based on statistical data known as the prevalence of DVT of real cases reported in literature and from the public hospital is proposed. The above is used to generate one dataset of 10,000 synthetic cases. Each synthetic case has nine risk factors according to Wells’ criteria and also the use of two additional factors, such as gender and age, is proposed. According to interviews with medical specialists, a training scheme was established. In addition, a new algorithm is presented to improve the Accuracy and Sensitivity/Recall. According to the proposed algorithm, two thresholds of decision were found, the first one is 0.484, which is to improve Accuracy. The other one is 0.138 to improve Sensitivity/Recall. The Accuracy achieved is 90.99%, which is greater than that obtained with other related machine learning methods. The proposed ANN model was validated performing the k-fold cross validation technique using a dataset with 10,000 synthetic cases. The test was performed by using 59 real cases obtained from a regional hospital, achieving an Accuracy of 98.30%.

Point-of-Care Ultrasound: A Useful Diagnostic Tool in Africa

In resource limited settings with limited tests and diagnostic tools, most of diagnoses are based on clinical findings, and patients are managed empirically, e.g. with anti-tuberculosis drugs. This article aims at describing the use of point-of-care ultrasound in diagnosing the most important conditions in Africa, in addition to clinical work-up. Different protocols exist for the diagnosis of trauma-related disorders, tuberculosis, schistosomiasis, thromboembolism, causes of dyspnea, and non- traumatic shock. Point-of-care ultrasound might be a beneficial tool in Africa, aiding diagnostics and management of patients with these conditions. However, studies must be done to assess the impact of point-of-care ultrasound on mortality.

Ultrasound for Lower Extremity Deep Venous Thrombosis

Venous ultrasound is the standard imaging test for patients suspected of having acute deep venous thrombosis (DVT). There is variability and disagreement among authoritative groups regarding the necessary components of the test. Some protocols include scanning the entire lower extremity, whereas others recommend scans limited to the thigh and knee supplemented with serial testing. Some protocols use gray-scale ultrasound alone, whereas others include Doppler interrogation. Point-of-care ultrasound is recommended in some settings, and there is heterogeneity of these protocols as well. Heterogeneity of recommendations can lead to errors including incorrect application of guidelines, confusion among requesting physicians, and incorrect follow-up. In October 2016, the Society of Radiologists in Ultrasound convened a multidisciplinary panel of experts to evaluate the current evidence to develop recommendations regarding ultrasound protocols for DVT and the terminology used to communicate results to clinicians. Recommendations were made after open discussion and by unanimous consensus.

The panel recommends a comprehensive duplex ultrasound protocol from thigh to ankle with Doppler at selected sites rather than a limited or complete compression-only examination. This protocol is currently performed in many facilities and is achievable with standard ultrasound equipment and personnel. The use of these recommendations will increase the diagnosis of calf DVT and provide better data to explain the presenting symptoms. The panel recommends a single point-of-care protocol that minimizes underdiagnoses of proximal DVT.

The panel recommends the term chronic postthrombotic change to describe the residual material that persists after the acute presentation of DVT to avoid potential overtreatment of prior thrombus.

Adoption of a single standardized comprehensive duplex ultrasound and a single point-of-care examination will enhance patient safety and clinicians’ confidence.

The utility of on-site ultrasound screening in population at high risk for deep venous thrombosis in temporary housing after the great East Japan Earthquake

BACKGROUND

To evaluate the diagnostic performance of ultrasonography for screening of a population at risk for deep vein thrombosis (DVT) in a post-disaster setting.

METHODS

Ultrasonography was applied as a screening technique to the residents of a temporary housing facility who were displaced following the Great East Japan Earthquake. Thirty DVT screening sessions were held from April 2013 to June 2015. Individuals were invited to participate if they were identified as "high risk" for DVT, defined as those with low activity levels, a history of lower limb trauma, baseline lower limb pain or swelling, a cancer-bearing status, or a history of venous thromboembolic event. Ultrasonographic examinations were performed from the calf to the popliteal veins using portable devices.

RESULTS

Of the 3,316 subjects screened (682 men and 2,634 women) with a mean age of 71 ± 9.7 years, DVT was diagnosed in 382 (11.5%). DVT was more likely in older, symptomatic, or female patients with a history of venous thromboembolism and attempting to perform daily exercise. The rate of DVT diagnosis increased from 9.9% in 2013 to 13.5% in 2015.

CONCLUSIONS

Ultrasonography is efficient for screening at-risk populations in challenging settings. © 2017 Wiley Periodicals, Inc. J Clin Ultrasound 45:566-574, 2017.

Emergency Department Management of Suspected Calf-Vein Deep Venous Thrombosis: A Diagnostic Algorithm

Introduction

Unilateral leg swelling with suspicion of deep venous thrombosis (DVT) is a common emergency department (ED) presentation. Proximal DVT (thrombus in the popliteal or femoral veins) can usually be diagnosed and treated at the initial ED encounter. When proximal DVT has been ruled out, isolated calf-vein deep venous thrombosis (IC-DVT) often remains a consideration. The current standard for the diagnosis of IC-DVT is whole-leg vascular duplex ultrasonography (WLUS), a test that is unavailable in many hospitals outside normal business hours. When WLUS is not available from the ED, recommendations for managing suspected IC-DVT vary. The objectives of the study is to use current evidence and recommendations to (1) propose a diagnostic algorithm for IC-DVT when definitive testing (WLUS) is unavailable; and (2) summarize the controversy surrounding IC-DVT treatment.

Discussion

The Figure combines D-dimer testing with serial CUS or a single deferred FLUS for the diagnosis of IC-DVT. Such an algorithm has the potential to safely direct the management of suspected IC-DVT when definitive testing is unavailable. Whether or not to treat diagnosed IC-DVT remains widely debated and awaiting further evidence.

Conclusion

When IC-DVT is not ruled out in the ED, the suggested algorithm, although not prospectively validated by a controlled study, offers an approach to diagnosis that is consistent with current data and recommendations. When IC-DVT is diagnosed, current references suggest that a decision between anticoagulation and continued follow-up outpatient testing can be based on shared decision-making. The risks of proximal progression and life-threatening embolization should be balanced against the generally more benign natural history of such thrombi, and an individual patient’s risk factors for both thrombus propagation and complications of anticoagulation.