Development and external validation of a mixed-effects deep learning model to diagnose COVID-19 from CT imaging.

Background: The automatic analysis of medical images has the potential improve diagnostic accuracy while reducing the strain on clinicians. Current methods analyzing 3D-like imaging data, such as computerized tomography imaging, often treat each image slice as individual slices. This may not be able to appropriately model the relationship between slices. Methods: Our proposed method utilizes a mixed-effects model within the deep learning framework to model the relationship between slices. We externally validated this method on a data set taken from a different country and compared our results against other proposed methods. We evaluated the discrimination, calibration, and clinical usefulness of our model using a range of measures. Finally, we carried out a sensitivity analysis to demonstrate our methods robustness to noise and missing data. Results: In the external geographic validation set our model showed excellent performance with an AUROC of 0.930 (95%CI: 0.914, 0.947), with a sensitivity and specificity, PPV, and NPV of 0.778 (0.720, 0.828), 0.882 (0.853, 0.908), 0.744 (0.686, 0.797), and 0.900 (0.872, 0.924) at the 0.5 probability cut-off point. Our model also maintained good calibration in the external validation dataset, while other methods showed poor calibration. Conclusion: Deep learning can reduce stress on healthcare systems by automatically screening CT imaging for COVID-19. Our method showed improved generalizability in external validation compared to previous published methods. However, deep learning models must be robustly assessed using various performance measures and externally validated in each setting. In addition, best practice guidelines for developing and reporting predictive models are vital for the safe adoption of such models.

Feasibility of dynamic chest radiography to calculate lung volumes in adult people with cystic fibrosis: a pilot study.

INTRODUCTION: Dynamic chest radiography (DCR) is a novel, low-dose, real-time digital imaging system where software identifies moving thoracic structures and can automatically calculate lung areas. In an observational, prospective, non-controlled, single-centre pilot study, we compared it with whole-body plethysmography (WBP) in the measurement of lung volume subdivisions in people with cystic fibrosis (pwCF). METHODS: Lung volume subdivisions were estimated by DCR using projected lung area (PLA) during deep inspiration, tidal breathing and full expiration, and compared with same-day WBP in 20 adult pwCF attending routine review. Linear regression models to predict lung volumes from PLA were developed. RESULTS: Total lung area (PLA at maximum inspiration) correlated with total lung capacity (TLC) (r=0.78, p<0.001), functional residual lung area with functional residual capacity (FRC) (r=0.91, p<0.001), residual lung area with residual volume (RV) (r=0.82, p=0.001) and inspiratory lung area with inspiratory capacity (r=0.72, p=0.001). Despite the small sample size, accurate models were developed for predicting TLC, RV and FRC. CONCLUSION: DCR is a promising new technology that can be used to estimate lung volume subdivisions. Plausible correlations between plethysmographic lung volumes and DCR lung areas were identified. Further studies are needed to build on this exploratory work in both pwCF and individuals without CF. TRIAL REGISTRATION NUMBER: ISRCTN64994816.

Bilateral adaptive graph convolutional network on CT based Covid-19 diagnosis with uncertainty-aware consensus-assisted multiple instance learning.

Coronavirus disease (COVID-19) has caused a worldwide pandemic, putting millions of people's health and lives in jeopardy. Detecting infected patients early on chest computed tomography (CT) is critical in combating COVID-19. Harnessing uncertainty-aware consensus-assisted multiple instance learning (UC-MIL), we propose to diagnose COVID-19 using a new bilateral adaptive graph-based (BA-GCN) model that can use both 2D and 3D discriminative information in 3D CT volumes with arbitrary number of slices. Given the importance of lung segmentation for this task, we have created the largest manual annotation dataset so far with 7,768 slices from COVID-19 patients, and have used it to train a 2D segmentation model to segment the lungs from individual slices and mask the lungs as the regions of interest for the subsequent analyses. We then used the UC-MIL model to estimate the uncertainty of each prediction and the consensus between multiple predictions on each CT slice to automatically select a fixed number of CT slices with reliable predictions for the subsequent model reasoning. Finally, we adaptively constructed a BA-GCN with vertices from different granularity levels (2D and 3D) to aggregate multi-level features for the final diagnosis with the benefits of the graph convolution network's superiority to tackle cross-granularity relationships. Experimental results on three largest COVID-19 CT datasets demonstrated that our model can produce reliable and accurate COVID-19 predictions using CT volumes with any number of slices, which outperforms existing approaches in terms of learning and generalisation ability. To promote reproducible research, we have made the datasets, including the manual annotations and cleaned CT dataset, as well as the implementation code, available at https://doi.org/10.5281/zenodo.6361963.

Use of Dynamic Chest Radiography to Assess Treatment of Pulmonary Exacerbations in Cystic Fibrosis.

Background Although spirometry is an important marker in the management of pulmonary exacerbations in cystic fibrosis (CF), it is a forced maneuver and can generate aerosol. Therefore, it may be difficult to perform in some individuals. Dynamic chest radiography (DCR) provides real-time information regarding pulmonary dynamics alongside fluoroscopic-style thoracic imaging. Purpose To assess the effect of pulmonary exacerbation treatment by using both spirometry and DCR and assess the clinical utility of DCR in participants with CF experiencing pulmonary exacerbations. Materials and Methods In this prospective, observational, single-center pilot study, spirometry and DCR were performed before and after treatment of pulmonary exacerbations in participants with CF between December 2019 and August 2020. Spirometry measured forced expiratory volume in 1 second (FEV1) and forced vital capacity. DCR helped to measure projected lung area (PLA), hemidiaphragm midpoint position, and speed during tidal and deep breathing. Data were analyzed by using the paired t test or Wilcoxon signed-rank test. Correlation was assessed by using the Spearman rank correlation coefficient. Results Twenty participants with CF (mean age, 25 years ± 7 [standard deviation]; 14 women) were evaluated. Spirometry showed that percentage predicted FEV1 improved from a median of 44% (interquartile range [IQR], 17%) before treatment to 55% (IQR, 16%) after treatment (P = .004). DCR showed improvement in median deep breathing excursion for left and right hemidiaphragms (from 18 [IQR, 11] to 25 [IQR, 16] mm [P = .03] and from 13 [IQR, 6] to 19 [IQR, 14] mm [P = .03], respectively) and in median expiratory speed following deep breathing for left and right hemidiaphragms (from 7 [IQR, 2] to 11 [IQR, 5] mm/sec [P = .004] and 6 [IQR, 3] to 9 [IQR, 6] mm/sec [P = .004], respectively). PLA rate of change during full expiration and change in PLA during tidal breathing improved (from a mean of 42 cm2/sec ± 16 to 56 cm2/sec ± 24 [P = .03] and from a mean of 29 cm2 ± 14 to 35 cm2 ± 10 [P = .03], respectively). Conclusion Dynamic chest radiography demonstrated improvement in diaphragm speed and range of chest wall movement during respiration aftere treatment for pulmonary exacerbations in cystic fibrosis and showed potential as a tool to investigate the effect of pulmonary exacerbations on lung mechanics. Clinical trials registration no. NCT01234567 Published under a CC BY 4.0 license. Online supplemental material is available for this article.

Characterisation of hemidiaphragm dysfunction using dynamic chest radiography: a pilot study.

OBJECTIVES: Dynamic chest radiography (DCR) is a novel real-time digital fluoroscopic imaging system that produces clear, wide field-of-view diagnostic images of the thorax and diaphragm in motion, alongside novel metrics on moving structures within the thoracic cavity. We describe the use of DCR in the measurement of diaphragm motion in a pilot series of cases of suspected diaphragm dysfunction. METHODS: We studied 21 patients referred for assessment of diaphragm function due to suspicious clinical symptoms or imaging (breathlessness, orthopnoea, reduced exercise tolerance and/or an elevated hemidiaphragm on plain chest radiograph). All underwent DCR with voluntary sniff manoeuvres. RESULTS: Paradoxical motion on sniffing was observed in 14 patients, and confirmed in six who also underwent fluoroscopy or ultrasound. In four patients, DCR showed reduced hemidiaphragm excursion, but no paradoxical motion; in three, normal bilateral diaphragm motion was demonstrated. DCR was quick to perform, and well tolerated in all cases and with no adverse events reported. DCR was achieved in ∼5 min per patient, with images available to view by the clinician immediately within the clinical setting. CONCLUSION: DCR is a rapid, well-tolerated and straightforward chest radiography technique that warrants further investigation in the assessment of diaphragm dysfunction.

Measuring the effect of elexacaftor/tezacaftor/ivacaftor combination therapy on the respiratory pump in people with CF using dynamic chest radiography.

BACKGROUND: The CFTR modulator elexacaftor/tezacaftor/ivacaftor (ELX/TEZ/IVA) leads to significant improvement in the symptoms and spirometry of people with cystic fibrosis (pwCF), but little evidence exists to understand its effect on respiratory pump function. Dynamic chest radiography (DCR) is a novel cineradiographic tool that identifies and tracks the chest wall and diaphragm throughout the breathing cycle, alongside fluoroscopic images of the chest of diagnostic quality. METHODS: In this observational work, we examined the spirometry and DCR of 24 pwCF before and after starting ELX/TEZ/IVA. DCR automatically tracked the hemidiaphragm midpoints and projected lung area (PLA) during tidal and deep breathing manoeuvres. RESULTS: ppFEV1 (61±18 to 73±22, P<0.001) and ppFVC (77±16 to 88±15, P<0.001) improved significantly. DCR demonstrated a significant increase in hemidiaphragm excursion on both the right (18±11 to 26±9 mm, P<0.001) and left (21±11 to 31±11 mm, P<0.001) sides, as well as maximum hemidiaphragm speed during inspiration (right 22±14 to 31±11 mm/s, P=0.03; left 28±11 to 37±16 mm/s, P=0.02). PLA at end-expiration was significantly reduced (334±71 to 290±72cm2, P<0.001), with a significant increase in ΔPLA (83±40 to 117±36cm2, P<0.001). CONCLUSIONS: DCR demonstrated significant improvements in hemidiaphragm excursion and ΔPLA in pwCF started on ELX/TEZ/IVA. These changes likely reflect a reduction in air trapping and improved elastic recoil of the chest, and are consistent with improvements seen in spirometry. The changes seen with DCR are physiologically plausible and correlate well with spirometry. DCR warrants further investigation as a tool for assessing the impact of CFTR-modulating therapies.

Unilateral diaphragm paralysis with COVID-19 infection.

Neurological complications are well described in SARS-CoV-2, but for the first time we report a case of unilateral diaphragm paralysis occurring early in mechanical ventilation for respiratory failure due to such an infection. The patient subsequently required tracheostomy and ventilator support for 37 days, and had increased breathlessness and an elevated diaphragm at clinic review 9 months later. Dynamic chest radiography demonstrated persistent diaphragm paralysis with an accompanying postural change in lung volumes, and he subsequently underwent surgical plication. This case demonstrates that although persistent dyspnoea is a common feature following SARS-CoV-2 infection and is usually due to deconditioning or persistent parenchymal involvement, it can be due to other causes and needs to be investigated appropriately.

Mycobacterium Simiae Infection in a Person With Cystic Fibrosis.

Although non-tuberculous mycobacteria can be found in the airways of people with cystic fibrosis (pwCF), their role as pathogens may be uncertain, and treatment is problematic. We report the first case of Mycobacterium simiae (M. simiae), often associated with asymptomatic disease or colonisation, which caused pulmonary infection requiring treatment in a pwCF. This report shows that M. simiae, rare in pwCF in the United Kingdom, can cause significant illness and highlights the diagnostic difficulty in individuals with positive smear mycobacteria that can be mistaken initially for pulmonary tuberculosis (TB).

Utility and validity of dynamic chest radiography in cystic fibrosis (dynamic CF): an observational, non-controlled, non-randomised, single-centre, prospective study.

INTRODUCTION: Dynamic chest radiography (DCR) uses novel, low-dose radiographic technology to capture images of the thoracic cavity while in motion. Pulmonary function testing is important in cystic fibrosis (CF). The tolerability, rapid acquisition and lower radiation and cost compared with CT imaging may make DCR a useful adjunct to current standards of care. METHODS AND ANALYSIS: This is an observational, non-controlled, non-randomised, single-centre, prospective study. This study is conducted at the Liverpool Heart and Chest Hospital (LHCH) adult CF unit. Participants are adults with CF. This study reviews DCR taken during routine CF Annual Review (n=150), validates DCR-derived lung volumes against whole body plethysmography (n=20) and examines DCR at the start and end of pulmonary exacerbations of CF (n=20). The primary objectives of this study are to examine if DCR provides lung function information that correlates with PFT, and lung volumes that correlate whole body plethysmography. ETHICS AND DISSEMINATION: This study has received the following approvals: HRA REC (11 December 2019) and LHCH R&I (11 October 2019). Results are made available to people with CF, the funders and other researchers. Processed, anonymised data are available from the research team on request. TRIAL REGISTRATION NUMBER: ISRCTN 64994816.

Double jeopardy: multi-modality imaging of monozygotic "twin cap" atherosclerosis.

The investigation of asymptomatic but potentially vulnerable atherosclerosis is not yet a major focus for clinical Cardiologists. We have illustrated the contemporary investigation and treatment of such disease using a clinical case that involved monozygotic twins. One twin (T1) had unfortunately suffered a cardiac arrest whilst jogging and survived only due to bystander CPR and prompt defibrillation. His identical twin brother (T2), on subsequent investigation, harbours a compositionally identical lesion in a proximal coronary vessel that has not yet ruptured or provoked a clinical event. Following the presentation of both non-invasive and invasive images, we discuss the need for active suspicion and intensive treatment for those people with a 'genetic' risk of future myocardial infarction.

Imaging in pulmonary hypertension, part 2: Large vessel diseases.

Pulmonary hypertension is defined by physiological parameters but there are numerous causes that differ in their pathogenesis, management and prognosis. Causes include chronic cardiac or pulmonary diseases and diffuse small vessel disease but also a range of large vessel obstructive diseases. The physiological manifestation of all these diseases is increased pulmonary vascular resistance and pulmonary arterial hypertension, and while clinical features may provide a clue to diagnosis, imaging plays a fundamental role in establishing a precise diagnosis and therefore guiding therapy. Chronic thromboembolic pulmonary hypertension (CTEPH) is the most common large vessel cause of pulmonary hypertension. It is increasingly recognised as a major cause of morbidity and mortality which is underdiagnosed and often diagnosed late. The importance of CTEPH is that for patients in whom the distribution of disease lies predominantly in the proximal vasculature there is potential for symptomatic and physiological cure by surgical pulmonary endarterectomy. More distal disease may be suitable for medical management. Increased awareness on behalf of both clinicians and imagers is therefore paramount. However, there are other rare causes or large vessel obstruction/stenosis such as large vessel vasculitis, pulmonary artery tumour, fibrosing mediastinitis, congenital stenosis or extrinsic compression of the pulmonary arteries/veins. Atypical imaging appearance such as unilateral central pulmonary artery obstruction should lead to consideration of a diagnosis other than CTEPH.

Imaging in pulmonary hypertension, part 3: small vessel diseases.

Pulmonary hypertension is a significant cause of morbidity and mortality. Unfortunately, non-specific presentation and lack of awareness of the disease frequently lead to significant delay in diagnosis, often with the onset of right heart failure, when prognosis is poor and therapy is of limited effectiveness. The classification of pulmonary hypertension is a clinical one grouping diseases into categories with similar patho-physiological mechanism and therapeutic options. Pulmonary biopsy can provide a definitive diagnosis but is hazardous in patients with pulmonary hypertension. Imaging has emerged as an invaluable tool in differentiating the aetiology, assessing disease severity and directing further management. One of the most important roles of imaging is to differentiate diseases resulting from obstruction of the large pulmonary arteries from those secondary to diffuse small vessel disease, as these have very different prognosis and are also treated differently. Small vessel diseases causing pulmonary arterial hypertension most commonly result from diffuse remodelling of the pulmonary arterioles. There are multiple causes of arteriolar remodelling which share similar histopathological, clinical and imaging features. In a subgroup of small vessel diseases causing pulmonary hypertension the predominant site of increased vascular resistance is at the level of the capillaries or venules. Correct diagnosis of pulmonary veno-occlusive disease and pulmonary capillary haemangiomatosis is essential since poor prognosis and inadvertent administration of vasodilators (conventional therapy for arteriolar predominant disease) can result in fatal pulmonary oedema. Multimodality imaging plays an important role in suggesting a diagnosis, guiding further investigation and directing treatment.

Imaging in pulmonary hypertension, part 1: clinical perspectives, classification, imaging techniques and imaging algorithm.

Pulmonary arterial hypertension (PAH) is an uncommon condition associated with significant morbidity and mortality. It has diverse aetiology with differing clinical presentations, imaging features and treatments that range from surgical treatment of proximal chronic thromboembolic disease to targeted medical therapies in small vessel disease. Current classification of pulmonary hypertension (PH) is clinically based and groups diseases with similar pathophysiological mechanisms and therapeutic approaches. Groupings include conditions characterised by diffuse small vessel diseases such as idiopathic PAH, PH secondary to chronic hypoxic lung disease, left sided cardiac disease, chronic large vessel obstruction such as chronic thromboembolic disease and a miscellaneous group of diseases. The physiological manifestation of all of these diseases is increased pulmonary vascular resistance and PAH and while clinical features may provide a clue to diagnosis imaging plays a fundamental role in establishing a precise diagnosis and therefore guides therapy. A broad range of imaging modalities is available for the patient with suspected PH including chest radiograph, echocardiography, ventilation/perfusion scintigraphy, catheter pulmonary angiography as well as cross-sectional CT and MRI. Each modality has its strengths and limitations and different techniques may be used at different stages of diagnostic investigation and frequently complement each other. For example, while MRI and echocardiography permit cardiac structural and functional assessment, CT pulmonary angiography provides exquisite morphological information about the proximal pulmonary vasculature and lung parenchyma but little functional information. Modern cross-sectional imaging techniques (CT and MRI) hold the promise of a comprehensive evaluation of the heart, circulation and lung parenchyma in PH. The authors present a multimodality-imaging algorithm for the investigation of patients with suspected PH though it is acknowledged that there is some variation in practice depending on availability of resources and expertise.

Cerebral venous thrombosis-A pictorial review.

Cerebral venous thrombosis is a relatively uncommon but serious neurological condition. It can have disastrous consequences if not diagnosed and treated promptly. It can be missed on initial presentation clinically as well as radiologically unless there is high level of suspicion. This pictorial review is intended to bring an increased awareness among radiology and medical trainees who are involved in the management of these patients. In this review we also discuss the various radiological investigations that can be performed for making the diagnosis and emphasise ways to avoid pitfalls.

Risk assessment of pneumothorax and pulmonary haemorrhage complicating percutaneous co-axial cutting needle lung biopsy.

INTRODUCTION: The primary aim of this study was to evaluate the ability of radiologists to accurately estimate pneumothorax and pulmonary haemorrhage during percutaneous co-axial cutting needle CT-guided lung biopsy. METHODOLOGY: Patients undergoing cutting needle lung biopsy during the study period were identified; the path taken by the cutting needle marked on each pre-biopsy staging CT scan. Each scan was then reviewed independently by two thoracic radiologists blinded to clinical details and complications; pneumothorax and pulmonary haemorrhage risk estimated with a percentage Visual Analogue Scale. RESULTS: In 134 patients, pneumothorax occurred in 24%. The radiologists differed in the estimation of pneumothorax risk in 55% (74 episodes). When pneumothorax risk was estimated <20% by radiologists 1 and 2, 16% and 14% of biopsies resulted in pneumothorax; where risk was estimated at 20-49%, pneumothorax incidence rose to 33% and 31%; where risk was deemed > or =50%, pneumothorax rate was 87% and 100%. Pulmonary haemorrhage occurred in 4%; estimated haemorrhage risk for biopsies complicated by haemorrhage did not differ significantly from where haemorrhage did not occur. CONCLUSION: Radiologists differ markedly in the estimation of pneumothorax risk for a patient undergoing co-axial lung biopsy. Identifying individual patients developing pneumothorax was only possible when risk was estimated at > or =50%. Pulmonary haemorrhage was uncommon and difficult to predict accurately.

Stimulation of the plasma membrane Na+/H+ exchanger NHE1 by sustained intracellular acidosis. Evidence for a novel mechanism mediated by the ERK pathway.

Activity of the Na+/H+ exchanger (NHE) isoform 1 (NHE1) is increased by intracellular acidosis through the interaction of intracellular H+ with an allosteric modifier site in the transport domain. Additional regulation is achieved via kinase-mediated modulation of the NHE1 regulatory domain. To determine if intracellular acidosis stimulates NHE1 activity solely by the allosteric mechanism, we subjected cultured neonatal rat ventricular myocytes (NRVM) with native NHE1 expression to intracellular acidosis (pHi approximately 6.6) for up to 6 min by transient exposure to NH4Cl and its washout in the presence of NHE inhibition (by zero [Na+]o or the NHE1 inhibitor cariporide) in HCO3- -free medium. After the desired duration of acidosis, NHE was reactivated (by reintroduction of [Na+]o or removal of cariporide), and the rate of recovery of pHi (dpHi/dt) was measured as the index of NHE activity. Regardless of the method used when intracellular acidosis was sustained for > or =3 min, subsequent NHE activity was significantly increased (>4-fold). Similar NHE stimulatory effects of sustained acidosis were observed in adult rat ventricular myocytes and COS-7 cells. Sustained (3 min) intracellular acidosis activated several NHE1 kinases in NRVM, in an in-gel kinase assay using as substrate a glutathione S-transferase fusion protein of the NHE1 regulatory domain. Detailed investigation of ERK and its downstream effector p90RSK, two putative NHE1 kinases, revealed time-dependent activation of both by intracellular acidosis in NRVM. Furthermore, inhibition of MEK1/2 by pretreatment of NRVM with two structurally distinct inhibitors, PD98059 (30 microM) or UO126 (3 microM), inhibited the activation of ERK and p90RSK and abolished the stimulation of NHE activity by sustained (3 min) intracellular acidosis. Our data show that not only the extent but also the duration of intracellular acidosis regulates NHE1 activity and suggest that the stimulatory effect of sustained intracellular acidosis occurs through a novel mechanism mediated by activation of the ERK pathway.