A History and Modern Framework for Quality Improvement in Interventional Radiology.

Quality improvement (QI) initiatives have benefited patients as well as the broader practice of medicine. Large-scale QI has been facilitated by multi-institutional data registries, many of which were formed out of national or international medical society initiatives. With broad participation, QI registries have provided benefits that include but are not limited to establishing treatment guidelines, facilitating research related to uncommon procedures and conditions, and demonstrating the fiscal and clinical value of procedures for both medical providers and health systems. Because of the benefits offered by these databases, Society of Interventional Radiology (SIR) and SIR Foundation have committed to the development of an interventional radiology (IR) clinical data registry known as VIRTEX. A large IR database with participation from a multitude of practice environments has the potential to have a significant positive impact on the specialty through data-driven advances in patient safety and outcomes, clinical research, and reimbursement. This article reviews the current landscape of societal QI programs, presents a vision for a large-scale IR clinical data registry supported by SIR, and discusses the anticipated results that such a framework can produce.

Erratum: Author Correction: Advancing COVID-19 diagnosis with privacy-preserving collaboration in artificial intelligence.

[This corrects the article DOI: 10.1038/s42256-021-00421-z.].

Advancing COVID-19 Diagnosis with Privacy-Preserving Collaboration in Artificial Intelligence.

Artificial intelligence (AI) provides a promising substitution for streamlining COVID-19 diagnoses. However, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalised model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the AI model can be distributedly trained and independently executed at each host institution under a federated learning framework (FL) without data sharing. Here we show that our FL model outperformed all the local models by a large yield (test sensitivity /specificity in China: 0.973/0.951, in the UK: 0.730/0.942), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals leaving out the FL) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans (CTs) from 3,336 patients collected from 23 hospitals located in China and the UK. Collectively, our work advanced the prospects of utilising federated learning for privacy-preserving AI in digital health.

Advancing COVID-19 diagnosis with privacy-preserving collaboration in artificial intelligence.

Artificial intelligence provides a promising solution for streamlining COVID-19 diagnoses; however, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalized model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the artificial intelligence (AI) model can be distributedly trained and independently executed at each host institution under a federated learning framework without data sharing. Here we show that our federated learning framework model considerably outperformed all of the local models (with a test sensitivity/specificity of 0.973/0.951 in China and 0.730/0.942 in the United Kingdom), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals without the federated learning framework) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans from 3,336 patients collected from 23 hospitals located in China and the United Kingdom. Collectively, our work advanced the prospects of utilizing federated learning for privacy-preserving AI in digital health.

A collaborative online AI engine for CT-based COVID-19 diagnosis.

Artificial intelligence can potentially provide a substantial role in streamlining chest computed tomography (CT) diagnosis of COVID-19 patients. However, several critical hurdles have impeded the development of robust AI model, which include deficiency, isolation, and heterogeneity of CT data generated from diverse institutions. These bring about lack of generalization of AI model and therefore prevent it from applications in clinical practices. To overcome this, we proposed a federated learning-based Unified CT-COVID AI Diagnostic Initiative (UCADI, http://www.ai-ct-covid.team/), a decentralized architecture where the AI model is distributed to and executed at each host institution with the data sources or client ends for training and inferencing without sharing individual patient data. Specifically, we firstly developed an initial AI CT model based on data collected from three Tongji hospitals in Wuhan. After model evaluation, we found that the initial model can identify COVID from Tongji CT test data at near radiologist-level (97.5% sensitivity) but performed worse when it was tested on COVID cases from Wuhan Union Hospital (72% sensitivity), indicating a lack of model generalization. Next, we used the publicly available UCADI framework to build a federated model which integrated COVID CT cases from the Tongji hospitals and Wuhan Union hospital (WU) without transferring the WU data. The federated model not only performed similarly on Tongji test data but improved the detection sensitivity (98%) on WU test cases. The UCADI framework will allow participants worldwide to use and contribute to the model, to deliver a real-world, globally built and validated clinic CT-COVID AI tool. This effort directly supports the United Nations Sustainable Development Goals' number 3, Good Health and Well-Being, and allows sharing and transferring of knowledge to fight this devastating disease around the world.

Safety and efficacy of radiofrequency wire recanalization of chronic central venous occlusions.

PURPOSE: The purpose of this study is to evaluate the safety and efficacy of using a radiofrequency wire (RF) for central venous occlusion (CVO) recanalization after failure using conventional techniques. MATERIALS AND METHODS: A retrospective analysis of all central venous recanalization procedures using an RF wire from January 2007 to December 2012 was performed. This comprised 13 consecutive procedures in 12 patients. The electronic medical record and radiologic imaging studies were reviewed to obtain information regarding patient demographics, indication for revascularization, duration of vascular occlusion, procedure outcome, and complications. RESULTS: Technical success was achieved in nine of 13 (69%) occluded vessels in nine of 12 (75%) patients. Lengths of successfully crossed lesions (mean 29.8 ± 29.3 mm) were significantly less than those of unsuccessfully crossed lesions (mean 90 ± 73.7 mm), p = 0.039. Of the nine patients with technically successful procedures, three died with patent stents of causes unrelated to the procedure, one had a patent stent at 547 days, two had stent patency of 94 days and 345 days, and three were lost to follow-up. One patient (8.3%) experienced a major complication wherein tracheal perforation by the RF wire contributed to the patient's death. CONCLUSIONS: RF recanalization is a viable option in patients with central vein occlusion refractory to traditional procedures. However, it does not guarantee successful revascularization and is not without the potential for harm.

Value of complete full-length lower extremity sonography in patients undergoing computed tomographic pulmonary angiography.

OBJECTIVES: To evaluate the diagnostic yield of complete lower extremity venous sonography for diagnosing deep venous thrombosis (DVT) in patients undergoing computed tomographic (CT) pulmonary angiography for suspected pulmonary embolism (PE). METHODS: We retrospectively reviewed all cases of lower extremity venous sonography and CT pulmonary angiography performed within 1 day of each other (n = 147) in a tertiary care center. Indications for the studies performed, angiographic findings, sonographic findings, age, sex, inpatient/outpatient status, lower extremity symptoms, and treatment status were recorded. Prevalence rates and patient characteristics were compared by χ(2) and Fisher exact probability tests where appropriate. Multivariable logistic regression with acute PE as the outcome was performed for age, sex, interval between angiography and sonography, indication for angiography, inpatient/outpatient status, and venous sonographic findings. RESULTS: The prevalence of PE (23.8%) was similar to the prevalence of DVT (27.9%). Angiographic findings were not associated with the interval between angiography and sonography or inpatient/outpatient status. Acute DVT was more likely (P = .0009) when angiographic findings were positive (51.4%), but DVT prevalence was still substantial (20.5%) in patients with negative angiographic findings. Lower extremity symptoms were not associated with DVT in cases with negative angiographic findings (P = .48). Eighteen of the 23 patients with acute DVT and negative angiographic findings were treated. CONCLUSIONS: There is a high rate of DVT in a population undergoing CT pulmonary angiography for suspected PE even when PE is not diagnosed. Our data apply to a tertiary care institution, suggest a surveillance bias, and favor the utility of venous sonography in this population.

Cephalic arch stenosis in dialysis patients: review of clinical relevance, anatomy, current theories on etiology and management.

Arteriovenous hemodialysis fistulas (AVFs) serve as a lifeline for many individuals with end-stage renal failure. A common cause of AVF failure is cephalic arch stenosis. Its high prevalence compounded with its resistance to treatment makes cephalic arch stenosis important to understand. Proposed etiologies include altered flow in a fistulized cephalic vein, external compression by fascia, the unique morphology of the cephalic arch, large number of valves in the cephalic outflow tract and biochemical changes that accompany renal failure. Management options are also in debate and include angioplasty, cutting balloon angioplasty, bare metal stents, stent grafts and surgical techniques including flow reduction with minimally invasive banding as well as more invasive venovenostomy with transposition surgeries for refractory cases. In this review, the evidence for the clinical relevance of cephalic arch stenosis, its etiology and management are summarized.

Analysis of baseline parameters in the HALT polycystic kidney disease trials.

HALT PKD consists of two ongoing randomized trials with the largest cohort of systematically studied patients with autosomal dominant polycystic kidney disease to date. Study A will compare combined treatment with an angiotensin-converting inhibitor and receptor blocker to inhibitor alone and standard compared with low blood pressure targets in 558 early-stage disease patients with an eGFR over 60 ml/min per 1.73 m(2). Study B will compare inhibitor-blocker treatment to the inhibitor alone in 486 late-stage patients with eGFR 25-60 ml/min per 1.73 m(2). We used correlation and multiple regression cross-sectional analyses to determine associations of baseline parameters with total kidney, liver, or liver cyst volumes measured by MRI in Study A and eGFR in both studies. Lower eGFR and higher natural log-transformed urine albumin excretion were independently associated with a larger natural log-transformed total kidney volume adjusted for height (ln(HtTKV)). Higher body surface area was independently associated with a higher ln(HtTKV) and lower eGFR. Men had larger height-adjusted total kidney volume and smaller liver cyst volumes than women. A weak correlation was found between the ln(HtTKV) and natural log-transformed total liver volume adjusted for height or natural log liver cyst volume in women only. Women had higher urine aldosterone excretion and lower plasma potassium. Thus, our analysis (1) confirms a strong association between renal volume and functional parameters, (2) shows that gender and other factors differentially affect the development of polycystic disease in the kidney and liver, and (3) suggests an association between anthropomorphic measures reflecting prenatal and/or postnatal growth and disease severity.

Cardiac magnetic resonance assessment of left ventricular mass in autosomal dominant polycystic kidney disease.

BACKGROUND AND OBJECTIVES: Autosomal dominant polycystic kidney disease (ADPKD) is associated with a substantial cardiovascular disease burden including early onset hypertension, intracranial aneurysms, and left ventricular hypertrophy (LVH). A 41% prevalence of LVH has been reported in ADPKD, using echocardiographic assessment of LV mass (LVM). The HALT PKD study was designed to assess the effect of intensive angiotensin blockade on progression of total kidney volume and LVM. Measurements of LVM were performed using cardiac magnetic resonance (MR). DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: Five hundred forty-three hypertensive patients with GFR >60 ml/min per 1.73 m(2) underwent MR assessment of LVM at baseline. LVM was adjusted for body surface area and expressed as LVM index (LVMI; g/m(2)). RESULTS: Baseline BP was 125.1 ± 14.5/79.3 ± 11.6 mmHg. Average duration of hypertension was 5.79 years. Prior use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers was present in 59.5% of patients. The prevalence of LVH assessed using nonindexed LVM (g) was 3.9% (n = 21, eight men and 13 women) and 0.93% (n = 5, one man and four women) using LVMI (g/m(2)). In exploratory analyses, the prevalence of LVH using LVM indexed to H(2.7), and the allometric index ppLVmass(HW), ranged from 0.74% to 2.23% (n = 4 to 12). Multivariate regression showed significant direct associations of LVMI with systolic BP, serum creatinine, and albuminuria; significant inverse associations with LVMI were found with age and female gender. CONCLUSIONS: The prevalence of LVH in hypertensive ADPKD patients <50 years of age with short duration of hypertension, and prior use of angiotensin-converting enzyme inhibitors/angiotensin receptor blockers is low. Early BP intervention in ADPKD may have decreased LVH and may potentially decrease cardiovascular mortality.