A rare etiology of frequent ventricular ectopy: Embolic complication of a venous iliac stent to the right ventricle

Peripheral venous stent migration is an exceedingly rare complication of endovascular stenting. In this clinical vignette, we present a case of a 74-year-old male with a history of endo-venous laser ablation therapy of the right greater saphenous vein complicated with an occlusion requiring a left iliac vein stent. The patient presented to the clinic months after the procedure with complaints of palpitations. Multimodality imaging revealed a stent that had become dislodged and was now located in the right ventricle, trapped within the tricuspid valve apparatus.

Detecting insertion, substitution, and deletion errors in radiology reports using neural sequence-to-sequence models

Background

Errors in grammar, spelling, and usage in radiology reports are common. To automatically detect inappropriate insertions, deletions, and substitutions of words in radiology reports, we proposed using a neural sequence-to-sequence (seq2seq) model.

Methods

Head CT and chest radiograph reports from Mount Sinai Hospital (MSH) (n=61,722 and 818,978, respectively), Mount Sinai Queens (MSQ) (n=30,145 and 194,309, respectively) and MIMIC-III (n=32,259 and 54,685) were converted into sentences. Insertions, substitutions, and deletions of words were randomly introduced. Seq2seq models were trained using corrupted sentences as input to predict original uncorrupted sentences. Three models were trained using head CTs from MSH, chest radiographs from MSH, and head CTs from all three collections. Model performance was assessed across different sites and modalities. A sample of original, uncorrupted sentences were manually reviewed for any error in syntax, usage, or spelling to estimate real-world proofreading performance of the algorithm.

Results

Seq2seq detected 90.3% and 88.2% of corrupted sentences with 97.7% and 98.8% specificity in same-site, same-modality test sets for head CTs and chest radiographs, respectively. Manual review of original, uncorrupted same-site same-modality head CT sentences demonstrated seq2seq positive predictive value (PPV) 0.393 (157/400; 95% CI, 0.346-0.441) and negative predictive value (NPV) 0.986 (789/800; 95% CI, 0.976-0.992) for detecting sentences containing real-world errors, with estimated sensitivity of 0.389 (95% CI, 0.267-0.542) and specificity 0.986 (95% CI, 0.985-0.987) over n=86,211 uncorrupted training examples.

Conclusions

Seq2seq models can be highly effective at detecting erroneous insertions, deletions, and substitutions of words in radiology reports. To achieve high performance, these models require site- and modality-specific training examples. Incorporating additional targeted training data could further improve performance in detecting real-world errors in reports.

Machine learning for semi-automated classification of glioblastoma, brain metastasis and central nervous system lymphoma using magnetic resonance advanced imaging

Background

Differentiating glioblastoma, brain metastasis, and central nervous system lymphoma (CNSL) on conventional magnetic resonance imaging (MRI) can present a diagnostic dilemma due to the potential for overlapping imaging features. We investigate whether machine learning evaluation of multimodal MRI can reliably differentiate these entities.

Methods

Preoperative brain MRI including diffusion weighted imaging (DWI), dynamic contrast enhanced (DCE), and dynamic susceptibility contrast (DSC) perfusion in patients with glioblastoma, lymphoma, or metastasis were retrospectively reviewed. Perfusion maps (rCBV, rCBF), permeability maps (K-trans, Kep, Vp, Ve), ADC, T1C+ and T2/FLAIR images were coregistered and two separate volumes of interest (VOIs) were obtained from the enhancing tumor and non-enhancing T2 hyperintense (NET2) regions. The tumor volumes obtained from these VOIs were utilized for supervised training of support vector classifier (SVC) and multilayer perceptron (MLP) models. Validation of the trained models was performed on unlabeled cases using the leave-one-subject-out method. Head-to-head and multiclass models were created. Accuracies of the multiclass models were compared against two human interpreters reviewing conventional and diffusion-weighted MR images.

Results

Twenty-six patients enrolled with histopathologically-proven glioblastoma (n=9), metastasis (n=9), and CNS lymphoma (n=8) were included. The trained multiclass ML models discriminated the three pathologic classes with a maximum accuracy of 69.2% accuracy (18 out of 26; kappa 0.540, P=0.01) using an MLP trained with the VpNET2 tumor volumes. Human readers achieved 65.4% (17 out of 26) and 80.8% (21 out of 26) accuracies, respectively. Using the MLP VpNET2 model as a computer-aided diagnosis (CADx) for cases in which the human reviewers disagreed with each other on the diagnosis resulted in correct diagnoses in 5 (19.2%) additional cases.

Conclusions

Our trained multiclass MLP using VpNET2 can differentiate glioblastoma, brain metastasis, and CNS lymphoma with modest diagnostic accuracy and provides approximately 19% increase in diagnostic yield when added to routine human interpretation.

Yttrium-90 Complications: Prevention and Management

Radioembolization has become a more prevalent treatment for both primary and secondary liver of the liver. Radioembolization is a relatively safe procedure with major complications being rare. Understanding how to identify the potential complications and their treatment can help make the procedure even safer and mitigate the risk of severe life threatening complications. In this article, we will review the most common complications, how to identify them, and how manage them.

End-hole Versus Microvalve Infusion Catheters in Patients Undergoing Drug-Eluting Microspheres-TACE for Solitary Hepatocellular Carcinoma Tumors: A Retrospective Analysis

INTRODUCTION

Pre-transplant locoregional therapy for hepatocellular carcinoma (HCC) during bridge-to-transplant impacts recurrence and survival rates following liver transplantation. Optimizing the effectiveness of transarterial chemoembolization (TACE) in this population is imperative, and microvalve infusion catheters offer a means of such improvement.

METHODS

All treatment-naive patients with solitary HCC tumors < 6.5 cm who underwent drug-eluting microspheres (DEM) TACE between 04/2015 and 08/2017 were retrospectively reviewed. Eighty-eight included patients underwent DEM-TACE with either standard end-hole catheters (EH) or microvalve infusion catheters (MVI). The EH (n = 70) and MVI (n = 18) cohorts had similar baseline tumor size, laboratory values, and tumor etiologies.

RESULTS

Initial objective response rates were significantly higher in MVI vs. EH (100% vs. 76.5%, p = 0.019). There was no difference in adverse events between groups (p = 0.265). MVI patients exhibited lower AST (p = 0.003) and ALT (p = 0.044) at 6 months. Blinded pathological analysis of explanted livers showed greater concentrations of microspheres within the tumor relative to the surrounding tissue in MVI explants (88.7 ± 10.6%) versus the EH explants (55.3 ± 32.7%) (p = 0.002). There was significantly higher percentage tumor necrosis in the MVI group (89.0 ± 2.2%) compared with the EH group (56.1 ± 44.5%) (p = 0.006).

CONCLUSION

In this retrospective study of a single-center cohort, DEM-TACE procedures with MVI were associated with improved tumor response, increased deposition of microspheres within tumor tissue, and higher percentage tumor necrosis at explant relative to those performed using EH catheters.

Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study

BACKGROUND

There is interest in using convolutional neural networks (CNNs) to analyze medical imaging to provide computer-aided diagnosis (CAD). Recent work has suggested that image classification CNNs may not generalize to new data as well as previously believed. We assessed how well CNNs generalized across three hospital systems for a simulated pneumonia screening task.

METHODS AND FINDINGS

A cross-sectional design with multiple model training cohorts was used to evaluate model generalizability to external sites using split-sample validation. A total of 158,323 chest radiographs were drawn from three institutions: National Institutes of Health Clinical Center (NIH; 112,120 from 30,805 patients), Mount Sinai Hospital (MSH; 42,396 from 12,904 patients), and Indiana University Network for Patient Care (IU; 3,807 from 3,683 patients). These patient populations had an age mean (SD) of 46.9 years (16.6), 63.2 years (16.5), and 49.6 years (17) with a female percentage of 43.5%, 44.8%, and 57.3%, respectively. We assessed individual models using the area under the receiver operating characteristic curve (AUC) for radiographic findings consistent with pneumonia and compared performance on different test sets with DeLong's test. The prevalence of pneumonia was high enough at MSH (34.2%) relative to NIH and IU (1.2% and 1.0%) that merely sorting by hospital system achieved an AUC of 0.861 (95% CI 0.855-0.866) on the joint MSH-NIH dataset. Models trained on data from either NIH or MSH had equivalent performance on IU (P values 0.580 and 0.273, respectively) and inferior performance on data from each other relative to an internal test set (i.e., new data from within the hospital system used for training data; P values both <0.001). The highest internal performance was achieved by combining training and test data from MSH and NIH (AUC 0.931, 95% CI 0.927-0.936), but this model demonstrated significantly lower external performance at IU (AUC 0.815, 95% CI 0.745-0.885, P = 0.001). To test the effect of pooling data from sites with disparate pneumonia prevalence, we used stratified subsampling to generate MSH-NIH cohorts that only differed in disease prevalence between training data sites. When both training data sites had the same pneumonia prevalence, the model performed consistently on external IU data (P = 0.88). When a 10-fold difference in pneumonia rate was introduced between sites, internal test performance improved compared to the balanced model (10× MSH risk P < 0.001; 10× NIH P = 0.002), but this outperformance failed to generalize to IU (MSH 10× P < 0.001; NIH 10× P = 0.027). CNNs were able to directly detect hospital system of a radiograph for 99.95% NIH (22,050/22,062) and 99.98% MSH (8,386/8,388) radiographs. The primary limitation of our approach and the available public data is that we cannot fully assess what other factors might be contributing to hospital system-specific biases.

CONCLUSION

Pneumonia-screening CNNs achieved better internal than external performance in 3 out of 5 natural comparisons. When models were trained on pooled data from sites with different pneumonia prevalence, they performed better on new pooled data from these sites but not on external data. CNNs robustly identified hospital system and department within a hospital, which can have large differences in disease burden and may confound predictions.

Treatment of Primary Liver Tumors and Liver Metastases, Part 1: Nuclear Medicine Techniques

⁹⁰Y radioembolization is an increasingly used treatment for both primary and metastatic malignancy in the liver. Understanding the biophysical properties, dosing concerns, and imaging appearance of this treatment is important for interventional radiologists and nuclear medicine physicians to provide important therapy. ⁹⁰Y radioembolization is efficacious and safe, although the possibility of complications does exist. This article provides a comprehensive in-depth discussion about the indications for ⁹⁰Y radioembolization, reviews the role of preprocedural angiography and ^99mTc-macroaggregated albumin scans, illustrates different dosing techniques, compares and contrasts resin and glass microspheres, and describes potential complications.

Outcomes of radioembolization for unresectable hepatocellular carcinoma in patients with marginal functional hepatic reserve

PURPOSE

To evaluate the outcomes of radioembolization (RE) as a therapy for unresectable hepatocellular carcinoma (HCC) in patients with marginal functional hepatic reserve.

METHODS

A retrospective review of 471 patients (1/2010-7/2015) treated with RE (Therasphere, BTG, UK) was performed. A total of 36 patients (mean age: 66.1±9.3, male: 86.1%) underwent therapy for HCC with a MELD≥15 (median: 16, range: 15-22). Baseline demographics of the study cohort were as follows: etiology (HCV: 26, 72.2%), cirrhosis (n=32, 88.9%), ECOG 0 (n=16, 44.4%), Child-Pugh class (A=15, B=19, C=2), unilobar distribution (n=27, 75%), AFP>200 (n=11, 30.6%), portal vein thrombosis (PVT, n=7, 19.4%), metastasis (n=3, 8.3%). Outcomes analyzed included CTCAEv4.03 laboratory toxicities (120-day), imaging response (mRECIST), progression-free survival (PFS), and overall survival (OS).

RESULTS

A total of 42 treatments were performed with mean dose of 2.02±1.23GBq. The cumulative grade 3/4 toxicity was 28% overall and 21% for bilirubin at 120-days. The objective response and disease control rates were 48.3% (14/29) and 69% (20/29) respectively. The median (95% CI) PFS was 5.9 (4.4-7.7) months. Ten (27.8%) patients received additional locoregional therapy at a median (IQR) of 138 (102-243) days post RE. The mean (95% CI) OS was 21.9 (14.8-29.0) months. The absence of PVT was associated with improved OS (p=0.005) Disease control at 90-days was also associated with an OS benefit (p=0.037).

CONCLUSIONS

Patients with unresectable HCC and marginal functional hepatic reserve treated with RE had favorable objective response and disease control rates, both predictive of overall survival.

Yttrium-90 Glass-Based Microsphere Radioembolization in the Treatment of Hepatocellular Carcinoma Secondary to the Hepatitis B Virus: Safety, Efficacy, and Survival

PURPOSE

To evaluate outcomes of yttrium-90 radioembolization performed with glass-based microspheres in the treatment of hepatocellular carcinoma (HCC) secondary to the hepatitis B virus (HBV).

MATERIALS AND METHODS

A total of 675 patients treated between January 2006 and July 2014 were reviewed, of which 45 (age 62 y ± 10; 91% male) received glass-based radioembolization for HCC secondary to HBV. All patients were stratified according to previous therapy (naive, n = 14; 31.1%), Child-Pugh class (class A, n = 41; 91%), Eastern Cooperative Oncology Group (ECOG) performance status (PS; < 1, n = 21; 47%), solitary (n = 26; 58%) and unilobar (n = 37; 82%) tumor distribution, tumor size < 5 cm (n = 29; 64%), portal vein thrombosis (n = 14; 31%), α-fetoprotein level > 400 ng/mL (n = 17; 38%), and Barcelona Clinic Liver Cancer stage (A, n = 8; B, n = 9; C, n = 28).

RESULTS

A total of 50 radioembolization treatments were performed, with a 100% technical success rate (median target dose, 120 Gy). Clinical toxicities included pain (16%), fatigue (12%), and nausea (4%). Grade 3/4 laboratory toxicities included bilirubin (8%) and aspartate aminotransferase (4%) toxicities. Observed toxicities were independent of treatment dose. The objective response rates were 55% per modified Response Evaluation Criteria In Solid Tumors and 21% per World Health Organization criteria, and the disease control rate was 63%. Disease progression was secondary to new, nontarget HCC in 45% of cases. Median time to progression and overall survival were 6.0 mo (95% confidence interval [CI], 4.4-8.0 mo) and 19.3 mo (95% CI, 11.2-22.7 mo), respectively. Multivariate analysis demonstrated ECOG PS ≥ 1 and AFP level > 400 ng/mL to be independent predictors of inferior overall survival.

CONCLUSIONS

Glass-based radioembolization for HCC secondary to HBV can be safely performed, with favorable target lesion response and overall survival.

Tailoring the endovascular management of transplant renal artery stenosis

In this study we analyze the different types of endovascular interventions (EVIs) in de novo transplant renal artery stenosis (TRAS) and its anatomical subtypes to examine any variation in recovery of allograft function, blood pressure control, EVI patency and allograft survival with respect to EVI type (DES: drug-eluting stent, BMS: bare-metal stent, PTA: percutaneous transluminal angioplasty). Forty-five patients underwent a total of 50 primary EVIs (DES: 18, BMS: 26, PTA: 6). Patients were stratified according to medical co-morbidities, graft characteristics, biopsy results, clinical presentation and TRAS anatomic subtypes (anastomotic: 26, postanastomotic: 17, bend-kink: 2). There was significant improvement in allograft function and mean arterial blood pressure (MAP) control across all interventions (pre-EVI-creatinine [CR]: 2.8 ± 1.4, post-EVI-Cr: 2.1 ± 0.7, p < 0.001; pre-EVI-MAP: 117 ± 16, post-EVI-MAP: 112 ± 17, p = 0.03) with no significant difference among EVI types. There was no significant difference in allograft survival with respect to EVI type. Patency was significantly higher in EVIs performed with DES and BMS compared to PTA (p = 0.001). In the postanastomotic TRAS subtype, patency rates were significantly higher in DES compared to BMS (p = 0.012) in vessels of comparable reference diameter (≤5 mm).