COVID-19 classification of X-ray images using deep neural networks.

OBJECTIVES: In the midst of the coronavirus disease 2019 (COVID-19) outbreak, chest X-ray (CXR) imaging is playing an important role in diagnosis and monitoring of patients with COVID-19. We propose a deep learning model for detection of COVID-19 from CXRs, as well as a tool for retrieving similar patients according to the model's results on their CXRs. For training and evaluating our model, we collected CXRs from inpatients hospitalized in four different hospitals. METHODS: In this retrospective study, 1384 frontal CXRs, of COVID-19 confirmed patients imaged between March and August 2020, and 1024 matching CXRs of non-COVID patients imaged before the pandemic, were collected and used to build a deep learning classifier for detecting patients positive for COVID-19. The classifier consists of an ensemble of pre-trained deep neural networks (DNNS), specifically, ReNet34, ReNet50¸ ReNet152, and vgg16, and is enhanced by data augmentation and lung segmentation. We further implemented a nearest-neighbors algorithm that uses DNN-based image embeddings to retrieve the images most similar to a given image. RESULTS: Our model achieved accuracy of 90.3%, (95% CI: 86.3-93.7%) specificity of 90% (95% CI: 84.3-94%), and sensitivity of 90.5% (95% CI: 85-94%) on a test dataset comprising 15% (350/2326) of the original images. The AUC of the ROC curve is 0.96 (95% CI: 0.93-0.97). CONCLUSION: We provide deep learning models, trained and evaluated on CXRs that can assist medical efforts and reduce medical staff workload in handling COVID-19. KEY POINTS: • A machine learning model was able to detect chest X-ray (CXR) images of patients tested positive for COVID-19 with accuracy and detection rate above 90%. • A tool was created for finding existing CXR images with imaging characteristics most similar to a given CXR, according to the model's image embeddings.

The management of cardiac implantable electronic device lead perforations: a multicentre study.

AIMS: Lead perforation is a rare, well-known complication of cardiac implantable electronic device (CIED) implants, whose management is mostly not evidence-based. Main management strategies include conservative approach based on clinical and lead function follow-up vs. routine invasive lead revision approach. This study compared the complications of both strategies by composite endpoint, including recurrent perforation-related symptoms, recurrent pericardial effusion (PEf), lead dysfunction, and device infection during 12 month follow-up. METHODS AND RESULTS: Multicentre retrospective analysis, inquiring data from imaging studies, device interrogation, pericardiocentesis, and clinical charts of patients with suspected perforating leads between 2007 and 2014 in five hospitals. All cases were reviewed by electrophysiologist and defined as definite perforations by suggestive symptoms along with lead perforation on imaging, bloody PEf on pericardiocentesis shortly after implant, or right ventricular (RV) lead non-capture along with diaphragmatic stimulation upon bipolar pacing. Clinical outcomes associated with both management approaches were compared, with respect to the composite endpoint. The study included 48 definitive perforation cases: 22 managed conservatively and 26 via lead revision. Conservative management was associated with an increased composite endpoint compared with lead revision (8/22 vs. 1/26; P = 0.007). The dominant complication among the conservative cohort was appearance of cardiac tamponade during follow-up; 5/6 occurring in cases which presented with no or only mild PEf and were treated by antiplatelets/coagulants during or shortly after CIED implantation. CONCLUSION: A conservative management of CIED lead perforation is associated with increased complications compared with early lead revision. Lead revision may be the preferred management particularly in patients receiving antiplatelets/coagulants.

Ibrutinib-associated invasive fungal diseases in patients with chronic lymphocytic leukaemia and non-Hodgkin lymphoma: An observational study.

BACKGROUND: Invasive fungal diseases (IFD) are life-threatening infections most commonly diagnosed in acute leukaemia patients with prolonged neutropenia and are uncommonly diagnosed in patients with lymphoproliferative diseases. OBJECTIVES: Following the initial report of aspergillosis diagnosed shortly after beginning ibrutinib for chronic lymphocytic leukaemia, a survey was developed to seek additional cases of IFD during ibrutinib treatment. METHODS: Local and international physicians and groups were approached for relevant cases. Patients were included if they met the following criteria: diagnosis of chronic lymphocytic leukaemia/non-Hodgkin lymphoma; proven or probable IFD; and ibrutinib treatment on the date IFD were diagnosed. Clinical and laboratory data were captured using REDCap software. RESULT: Thirty-five patients with IFD were reported from 22 centres in eight countries: 26 (74%) had chronic lymphocytic leukaemia. The median duration of ibrutinib treatment before the onset of IFD was 45 days (range 1-540). Aspergillus species were identified in 22 (63%) of the patients and Cryptococcus species in 9 (26%). Pulmonary involvement occurred in 69% of patients, cranial in 60% and disseminated disease in 60%. A definite diagnosis was made in 21 patients (69%), and the mortality rate was 69%. Data from Israel regarding ibrutinib treated patients were used to evaluate a prevalence of 2.4% IFD. CONCLUSIONS: The prevalence of IFD among chronic lymphocytic leukaemia/non-Hodgkin lymphoma patients treated with ibrutinib appears to be higher than expected. These patients often present with unusual clinical features. Mortality from IFD in this study was high, indicating that additional studies are urgently needed to identify patients at risk for ibrutinib-associated IFD.

MEK inhibitor treatment is effective in a patient with metastatic carcinoma of the ampulla of Vater with BRAF and NRAS mutations shown by next-generation sequencing.

Here, we present a case of an 84-year-old woman who developed obstructive jaundice and was diagnosed with nonoperable adenocarcinoma originating from the ampulla of Vater, a lethal disease with a median overall survival of less than a year. Her tumor was examined by next-generation sequencing, which showed BRAF and NRAS mutations. To target these mutations, a MEK inhibitor was chosen for treatment. The patient has been treated with a MEK inhibitor for the last 12 months since diagnosis, with clinical and laboratory improvement and manageable side effects. PET-computed tomography imaging has shown stable disease or improvement in the primary and metastatic lesions. This is the first case report of an ampulla of a Vater cancer patient with NRAS and BRAF mutations, identified in next-generation sequencing, and treated successfully with a MEK inhibitor.

Radiological artificial intelligence - predicting personalized immunotherapy outcomes in lung cancer.

Personalized medicine has revolutionized approaches to treatment in the field of lung cancer by enabling therapies to be specific to each patient. However, physicians encounter an immense number of challenges in providing the optimal treatment regimen for the individual given the sheer complexity of clinical aspects such as tumor molecular profile, tumor microenvironment, expected adverse events, acquired or inherent resistance mechanisms, the development of brain metastases, the limited availability of biomarkers and the choice of combination therapy. The integration of innovative next-generation technologies such as deep learning-a subset of machine learning-and radiomics has the potential to transform the field by supporting clinical decision making in cancer treatment and the delivery of precision therapies while integrating numerous clinical considerations. In this review, we present a brief explanation of the available technologies, the benefits of using these technologies in predicting immunotherapy response in lung cancer, and the expected future challenges in the context of precision medicine.

Distribution of Aeration and Pulmonary Blood Volume in Healthy, ARDS and COVID-19 Lungs: A Dual-Energy Computed Tomography Retrospective Cohort Study.

RATIONALE AND OBJECTIVES: Few reports have studied lung aeration and perfusion in normal lungs, COVID-19, and ARDS from other causes (NC-ARDS) using dual-energy computed tomography pulmonary angiograms (DE-CTPA). To describe lung aeration and blood-volume distribution using DE-CTPAs of patients with NC-ARDS, COVID-19, and controls with a normal DE-CTPA ("healthy lungs"). We hypothesized that each of these conditions has unique ranges of aeration and pulmonary blood volumes. MATERIALS AND METHODS: This retrospective, single-center study of DE-CTPAs included patients with COVID-19, NC-ARDS (Berlin criteria), and controls. Patients with macroscopic pulmonary embolisms were excluded. The outcomes studied were the (1) lung blood-volume in areas with different aeration levels (normal, ground glass opacities [GGO], consolidated lung) and (2) aeration/blood-volume ratios. RESULTS: Included were 20 patients with COVID-19 (10 milds, 10 moderate-severe), six with NC-ARDS, and 12 healthy-controls. Lung aeration was lowest in patients with severe COVID-19 24% (IQR13%-31%) followed by those with NC-ARDS 40%(IQR21%-46%). Blood-volume in GGO was lowest in patients with COVID-19 [moderate-severe:-28.6 (IQR-33.1-23.2); mild: -30.1 (IQR-33.3-23.4)] and highest in normally aerated areas in NC-ARDS -37.4 (IQR-52.5-30.2-) and moderate-severe COVID-19 -33.5(IQR-44.2-28.5). The median aeration/blood-volume ratio was lowest in severe COVID-19 but some values overlapped with those observed among patients with NC-ARDS. CONCLUSION: Severe COVID-19 disease is associated with low total aerated lung volume and blood-volume in areas with GGO and overall aeration/blood volume ratios, and with high blood volume in normal lung areas. In this hypothesis-generating study, these findings were most pronounced in severe COVID disease. Larger studies are needed to confirm these preliminary findings.

The Added Value of a High CT Coronary Artery Calcium Score in the Management of Patients Presenting with Acute Chest Pain vs. Stable Chest Pain.

Background: Contrast computerized tomography (CT) scan is occasionally aborted due to a high coronary artery calcium score (CACS). For the same CACS in our clinical practice, we observed a higher occurrence of severe coronary artery disease (CAD) in patients with acute chest pain (ACP) compared to patients with stable chest pain (SCP). Since it is known that ACP differs in many ways from SCP, the aim of this study was to compare the predictive value of a high CACS for the diagnosis of severe CAD between ACP and SCP patients. Methods: This single center observational retrospective study included consecutive patients who underwent cardiac CT for chest pain and were found to have a CACS of >200 Agatston units. Patients were divided into two groups, ACP and SCP. Severe CAD was defined as ≥70% stenosis on coronary CT angiography or invasive coronary angiography. Baseline characteristics and final diagnosis of severe CAD were compared. Results: The cohort included 220 patients, 106 with ACP and 114 with SCP. ACP patients had higher severe CAD rates (60.4% vs. 36.8%; p < 0.001). On multivariate analysis including cardiac risk factors, CACS > 400 au (OR = 2.34 95% CI [1.32−4.15]; p = 0.004) and ACP (OR = 2.54 95% CI [1.45−4.45]; p = 0.001) were independent predictors of severe CAD. The addition of the clinical setting of ACP added significant incremental predictive value for severe stenosis. Conclusion: A high CACS is more associated with severe CAD in patients presenting with ACP than SCP. The findings suggest that the CACS could impact the management of patients during the scan.

Radiologic imaging and intervention for gastrointestinal and hepatic complications of hematopoietic stem cell transplantation.

Hematopoietic stem cell transplantation (HSCT) is an increasingly available treatment option for patients with various oncologic, hematologic, and immunologic diseases. Although HSCT can be curative for some diseases, complications associated with this treatment limit its success and applicability. Gastrointestinal graft-versus-host disease (GVHD) and hepatic veno-occlusive disease are unique and deadly complications of HSCT. These diseases can mimic other HSCT complications, such as infection, hemorrhage, and hepatotoxicity with cholestasis, but GVHD and veno-occlusive disease require specific treatment. Early treatment improves the probability of treatment success. For these reasons, timely and accurate diagnosis is essential. Abdominal imaging and intervention play an important role in the early, minimally invasive diagnosis and treatment of GVHD and veno-occlusive disease. Imaging findings tend to be nonspecific, but common findings that may guide further management or establish a diagnosis in the clinical setting have been defined. In cases where the diagnosis is unclear and liver biopsy is required, image-guided transvenous liver biopsy may be a safer and more practical option than the transcutaneous approach. Image-guided interventions, including intraarterial steroid-injection therapy in severe, systemic steroid-refractory GVHD and transjugular intrahepatic portosystemic shunt placement in veno-occlusive disease with portal hypertension, have shown some promise in small, uncontrolled series. Larger, controlled studies are needed to define the role of these invasive procedures in this patient population.

Image quality of low-energy pulmonary CT angiography: comparison with standard CT.

OBJECTIVE: The purpose of this article is to prospectively compare visualization of central and peripheral pulmonary arteries on simultaneously acquired low-energy and standard pulmonary CT angiography. SUBJECTS AND METHODS: Thirty-three consecutive patients (20 women and 13 men; mean age, 55.6 years; range, 21-92 years) with suspected pulmonary embolism (PE) were scanned (140 kVp; 250-300 mA) on a single-source dual-layer dual-energy MDCT scanner. Attenuation and image noise were measured at the main and segmental pulmonary arteries. Signal-to-noise ratios were calculated. Two blinded experienced radiologists assessed segmental and subsegmental artery visibility in consensus, using slab maximum-intensity-projection (MIP) reconstructions. Nonparametric sign test and kappa statistic were used for statistical analysis. RESULTS: PE was detected in three patients (9.1%); two segmental vessel and subsegmental emboli were seen in the low-energy images only. Higher attenuation was noted in low-energy versus standard images for all arteries evaluated, with a mean (± SD) increase of 66.6 ± 4.4 HU (p < 0.0001). Low-energy images improved visualization of segmental and subsegmental arteries from 97.0% to 99.2% and from 88.0% to 93.9%, respectively. A larger number of subsegmental vessels was seen on low-energy MIP reconstructions in 69.7% (95% CI, 36.5-71.89%) of studies compared with 9.1% on the standard images. Visualization of subsegmental vessels was superior in 55.5% of cases using low-energy imaging. The mean image noise increased by 9.7 ± 0.6 HU (p < 0.0001). The mean signal-to-noise ratio showed no significant difference in the low-energy (8.2) versus standard (8.1) CT images (p = 0.7759). CONCLUSION: Improved visualization of central and peripheral arteries can be obtained with low-energy pulmonary CT angiography, without a substantial decrease in image quality.

Severe influenza A (H1N1): the course of imaging findings.

BACKGROUND: An outbreak of respiratory illness caused by a novel swine-origin influenza virus (influenza A/H1N1 2009) that began in Mexico was declared a global pandemic by the World Health Organization in June 2009. The pandemic affected many countries, including Israel. OBJECTIVES: To compare the course of chest radiographic and computed tomography findings in patients who survived and those who died following admission to the intensive care unit (ICU) or intubation due to severe laboratory-confirmed swine-origin influenza A/H1N1 2009. METHODS: We retrospectively reviewed the patient records (267 radiographs, 8 CTs) of 22 patients (10 males, 12 females) aged 3.5-66 years (median 34) with confirmed influenza A/ H1N1 2009, admitted to the ICU and/or intubated in five major Israeli medical centers during the period July-November 2009. We recorded demographic, clinical, and imaging findings--including pattern of opacification, extent, laterality, distribution, zone of findings, and presence/absence of nodular opacities--at initial radiography and during the course of disease, and compared the findings of survivors and non-survivors. Statistical significance was calculated using the Wilcoxon (continuous variables) and Fisher exact tests. RESULTS: The most common findings on the initial chest radiography were airspace opacities, which were multifocal in 17 patients (77%) and bilateral in 16 (73%), and located in the lower or lower and middle lung zones in 19 patients (86%). Large airspace nodules with indistinct margins were seen in 8 patients (36%). Twelve patients survived, 10 died. Patients who died had multiple background illnesses and were significantly older than survivors (P = 0.006). Radiologic findings for the two groups were not significantly different. CONCLUSION: Airspace opacities, often with nodular appearance, were the most common findings among patients with severe influenza A/H1N1 2009. The course of radiologic findings was similar in patients with severe influenza A/ H1N1 2009 who survived and those who died.

Is There a Need for a Pulmonary Artery Catheter in Cardiac Surgery Today?

Pulmonary artery catheters are a useful tool for hemodynamic monitoring in high-risk patients during surgery and while in intensive care. However, there are major risks inherent to the device, and with modern day technology, their routine use has decreased. We discuss the need for routine insertion of pulmonary artery catheters in cardiac surgery. We also present a case of a left ventricular assist device implantation complicated by serious pulmonary hemorrhage due to pulmonary artery catheter insertion, highlighting the potentially life-threatening risks involved.

Loss of left ventricular rotation is a significant determinant of functional mitral regurgitation.

AIM: To evaluate insufficient rotational movement of the left ventricle (LV) as a potential novel mechanism for functional regurgitation of the mitral valve (FMR). METHODS AND RESULTS: We compared reference subjects and patients with LV dysfunction (LVD, ejection fraction EF < 50%) with and without FMR (regurgitant volume RVol>10 ml). Subjects without structural mitral valve pathology undergoing cardiac MRI were evaluated. Delayed enhancement, global LV remodeling parameters, systolic twist and torsion were measured (using manual and novel automated cardiac MRI tissue-tracking). The study included 117 subjects with mean ± SD age 50.4 ± 17.8 years, of which 30.8% were female. Compared to subjects with LVD without FMR (n = 31), those with FMR (n = 37) had similar clinical characteristics, diagnoses, delayed enhancement, EF, and longitudinal strain. Subjects with FMR had significantly larger left ventricles (EDVi:136.6 ± 41.8 vs 97.5 ± 26.2 ml/m, p < 0.0001) with wider separation between papillary muscles (21.1 ± 7.6 vs 17.2 ± 5.7 mm, p = 0.023). Notably, they had lower apical (p < 0.0001) but not basal rotation and lower peak systolic twist (3.1 ± 2.4° vs 5.5 ± 2.5°, p < 0.0001) and torsion (0.56 ± 0.38°/cm vs 0.88 ± 0.52°/cm, p = 0.004). In a multivariate model for RVol including age, gender, twist, LV end-diastolic volume, sphericity index and separation between papillary muscles, only gender, volume and twist were significant. Twist was the most powerful correlate (beta -2.23, CI -3.26 to -1.23 p < 0.001). In patients with FMR, peak systolic twist negatively correlates with RVol (r = -0.73, p < 0.0001). CONCLUSION: Reduced rotational systolic LV motion is significantly and independently associated with RVol among patients with FMR, suggesting a novel pathophysiological mechanism and a potential therapeutic target.

Diffuse pulmonary calcifications: A case series and review of literature.

Pulmonary calcifications are usually incidental asymptomatic findings discovered on x-rays or computed tomography scans that can be easily overlooked, and their significance undermined, especially in a seemingly asymptomatic person. Calcifications can be a marker of chronicity or disease severity, and thus have diagnostic value. Rarely, calcification can be the direct cause of morbidity. Calcifications can be either localized or diffuse. Many diseases, in particular infectious diseases, can cause localized calcifications. Diffuse calcifications are less common and usually secondary to a handful of conditions such as dystrophic pulmonary calcifications, metastatic pulmonary calcifications, disseminated pulmonary ossifications and pulmonary alveolar microlithiasis. We describe three cases of diffuse pulmonary calcifications, review the different causes of diffuse pulmonary calcifications and provide some indicators on how to differentiate between them. Differentiating between the different types of pulmonary calcifications has significant implications on the management and prognosis of the patients, and thus it is important to distinguish between them.

Bilateral Adrenal Hemorrhage in Coronavirus Disease 2019 Patient: A Case Report.

CONTEXT: Bilateral adrenal hemorrhage is a rare condition with potentially life-threatening consequences such as acute adrenal insufficiency. Early adrenal axis testing, as well as directed imaging, is crucial for immediate diagnosis and treatment. Coronavirus disease 2019 (COVID-19) has been associated with coagulopathy and thromboembolic events. CASE DESCRIPTION: A 66-year-old woman presented with acute COVID-19 infection and primary adrenal insufficiency due to bilateral adrenal hemorrhage (BAH). She also had a renal vein thrombosis. Her past medical history revealed primary antiphospholipid syndrome (APLS). Four weeks after discharge she had no signs of COVID-19 infection and her polymerase chain reaction test for COVID-19 was negative, but she still needed glucocorticoid and mineralocorticoid replacement therapy. The combination of APLS and COVID-19 was probably responsible of the adrenal event as a "two-hit" mechanism. CONCLUSIONS: COVID-19 infection is associated with coagulopathy and thromboembolic events, including BAH. Adrenal insufficiency is life threatening; therefore, we suggest that early adrenal axis testing for COVID-19 patients with clinical suspicion of adrenal insufficiency should be carried out.

Computer-aided diagnosis of pulmonary nodules on CT scans: improvement of classification performance with nodule surface features.

The purpose of this work is to develop a computer-aided diagnosis (CAD) system to differentiate malignant and benign lung nodules on CT scans. A fully automated system was designed to segment the nodule from its surrounding structured background in a local volume of interest (VOI) and to extract image features for classification. Image segmentation was performed with a 3D active contour method. The initial contour was obtained as the boundary of a binary object generated by k-means clustering within the VOI and smoothed by morphological opening. A data set of 256 lung nodules (124 malignant and 132 benign) from 152 patients was used in this study. In addition to morphological and texture features, the authors designed new nodule surface features to characterize the lung nodule surface smoothness and shape irregularity. The effects of two demographic features, age and gender, as adjunct to the image features were also investigated. A linear discriminant analysis (LDA) classifier built with features from stepwise feature selection was trained using simplex optimization to select the most effective features. A two-loop leave-one-out resampling scheme was developed to reduce the optimistic bias in estimating the test performance of the CAD system. The area under the receiver operating characteristic curve, A(z), for the test cases improved significantly (p < 0.05) from 0.821 +/- 0.026 to 0.857 +/- 0.023 when the newly developed image features were included with the original morphological and texture features. A similar experiment performed on the data set restricted to primary cancers and benign nodules, excluding the metastatic cancers, also resulted in an improved test A(z), though the improvement did not reach statistical significance (p = 0.07). The two demographic features did not significantly affect the performance of the CAD system (p > 0.05) when they were added to the feature space containing the morphological, texture, and new gradient field and radius features. To investigate if a support vector machine (SVM) classifier can achieve improved performance over the LDA classifier, we compared the performance of the LDA and SVMs with various kernels and parameters. Principal component analysis was used to reduce the dimensionality of the feature space for both the LDA and the SVM classifiers. When the number of selected principal components was varied, the highest test A(z) among the SVMs of various kernels and parameters was slightly higher than that of the LDA in one-loop leave-one-case-out resampling. However, no SVM with fixed architecture consistently performed better than the LDA in the range of principal components selected. This study demonstrated that the authors' proposed segmentation and feature extraction techniques are promising for classifying lung nodules on CT images.