Emergency triage of brain computed tomography via anomaly detection with a deep generative model.

Triage is essential for the early diagnosis and reporting of neurologic emergencies. Herein, we report the development of an anomaly detection algorithm (ADA) with a deep generative model trained on brain computed tomography (CT) images of healthy individuals that reprioritizes radiology worklists and provides lesion attention maps for brain CT images with critical findings. In the internal and external validation datasets, the ADA achieved area under the curve values (95% confidence interval) of 0.85 (0.81-0.89) and 0.87 (0.85-0.89), respectively, for detecting emergency cases. In a clinical simulation test of an emergency cohort, the median wait time was significantly shorter post-ADA triage than pre-ADA triage by 294 s (422.5 s [interquartile range, IQR 299] to 70.5 s [IQR 168]), and the median radiology report turnaround time was significantly faster post-ADA triage than pre-ADA triage by 297.5 s (445.0 s [IQR 298] to 88.5 s [IQR 179]) (all p < 0.001).

Patient recalls associated with resident-to-attending radiology report discrepancies: predictive factors for risky discrepancies.

BACKGROUND: This study aimed to identify predictive factors for risky discrepancies in the emergency department (ED) by analyzing patient recalls associated with resident-to-attending radiology report discrepancies (RRDs). RESULTS: This retrospective study analyzed 759 RRDs in computed tomography (CT) and magnetic resonance imaging and their outcomes from 2013 to 2021. After excluding 73 patients lost to follow-up, we included 686 records in the final analysis. Risky discrepancies were defined as RRDs resulting in (1) inpatient management (hospitalization) and (2) adverse outcomes (delayed operations, 30-day in-hospital mortality, or intensive care unit admission). Predictors of risky discrepancies were assessed using multivariable logistic regression analysis. The overall RRD rate was 0.4% (759 of 171,419). Of 686 eligible patients, 21.4% (147 of 686) received inpatient management, and 6.0% (41 of 686) experienced adverse outcomes. RRDs with neurological diseases were associated with the highest ED revisit rate (79.4%, 81 of 102) but not with risky RRDs. Predictive factors of inpatient management were critical finding (odds ratio [OR], 5.60; p < 0.001), CT examination (OR, 3.93; p = 0.01), digestive diseases (OR, 2.54; p < 0.001), and late finalized report (OR, 1.65; p = 0.02). Digestive diseases (OR, 6.14; p = 0.006) were identified as the only significant predictor of adverse outcomes. CONCLUSIONS: Risky RRDs were associated with several factors, including CT examination, digestive diseases, and late finalized reports, as well as critical image findings. This knowledge could aid in determining the priority of discrepancies for the appropriate management of RRDs.

Realistic High-Resolution Body Computed Tomography Image Synthesis by Using Progressive Growing Generative Adversarial Network: Visual Turing Test.

BACKGROUND: Generative adversarial network (GAN)-based synthetic images can be viable solutions to current supervised deep learning challenges. However, generating highly realistic images is a prerequisite for these approaches. OBJECTIVE: The aim of this study was to investigate and validate the unsupervised synthesis of highly realistic body computed tomography (CT) images by using a progressive growing GAN (PGGAN) trained to learn the probability distribution of normal data. METHODS: We trained the PGGAN by using 11,755 body CT scans. Ten radiologists (4 radiologists with <5 years of experience [Group I], 4 radiologists with 5-10 years of experience [Group II], and 2 radiologists with >10 years of experience [Group III]) evaluated the results in a binary approach by using an independent validation set of 300 images (150 real and 150 synthetic) to judge the authenticity of each image. RESULTS: The mean accuracy of the 10 readers in the entire image set was higher than random guessing (1781/3000, 59.4% vs 1500/3000, 50.0%, respectively; P<.001). However, in terms of identifying synthetic images as fake, there was no significant difference in the specificity between the visual Turing test and random guessing (779/1500, 51.9% vs 750/1500, 50.0%, respectively; P=.29). The accuracy between the 3 reader groups with different experience levels was not significantly different (Group I, 696/1200, 58.0%; Group II, 726/1200, 60.5%; and Group III, 359/600, 59.8%; P=.36). Interreader agreements were poor (κ=0.11) for the entire image set. In subgroup analysis, the discrepancies between real and synthetic CT images occurred mainly in the thoracoabdominal junction and in the anatomical details. CONCLUSIONS: The GAN can synthesize highly realistic high-resolution body CT images that are indistinguishable from real images; however, it has limitations in generating body images of the thoracoabdominal junction and lacks accuracy in the anatomical details.

Recurrent glioblastoma: optimum area under the curve method derived from dynamic contrast-enhanced T1-weighted perfusion MR imaging.

PURPOSE: To determine whether the ratio of the initial area under the time-signal intensity curve (AUC) (IAUC) to the final AUC--or AUCR--derived from dynamic contrast material-enhanced magnetic resonance (MR) imaging can be an imaging biomarker for distinguishing recurrent glioblastoma multiforme (GBM) from radiation necrosis and to compare the diagnostic accuracy of the AUCR with commonly used model-free dynamic contrast-enhanced MR imaging parameters. MATERIALS AND METHODS: The institutional review board approved this retrospective study and waived the informed consent requirement. Fifty-seven consecutive patients with pathologically confirmed recurrent GBM (n = 32) or radiation necrosis (n = 25) underwent dynamic contrast-enhanced MR imaging. Histogram parameters of the IAUC at 30, 60, and 120 seconds and the AUCR, which included the mean value at the higher curve of the bimodal histogram (mAUCR(H)), as well as 90th percentile cumulative histogram cutoffs, were calculated and were correlated with final pathologic findings. The best predictor for differentiating recurrent GBM from radiation necrosis was determined by means of receiver operating characteristic (ROC) curve analysis. RESULTS: The demographic data were not significantly different between the two patient groups. There were statistically significant differences in all of the IAUC and AUCR parameters between the recurrent GBM and the radiation necrosis patient groups (P < .05 for each). ROC curve analyses showed mAUCR(H) to be the best single predictor of recurrent GBM (mAUCR(H) for recurrent GBM = 0.35 ± 0.11 [standard deviation], vs 0.19 ± 0.17 for radiation necrosis; P < .0001; optimum cutoff, 0.23), with a sensitivity of 93.8% and a specificity of 88.0%. CONCLUSION: A bimodal histogram analysis of AUCR derived from dynamic contrast-enhanced MR imaging can be a potential noninvasive imaging biomarker for differentiating recurrent GBM from radiation necrosis. SUPPLEMENTAL MATERIAL: http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.13130016/-/DC1.

Long-term Outcomes of Drug-eluting Stents in Symptomatic Intracranial Stenosis.

PURPOSE: The use of drug-eluting stent (DES) to treat intracranial stenosis has shown short-term success. However, there are no reports regarding the long-term results of DES. We present the long-term clinical outcome after DES stenting for symptomatic severe intracranial stenosis. MATERIALS AND METHODS: Our study included a consecutive series of 11 patients who underwent intracranial stenting using DES between March and July, 2006, during the time when bare metal stents were not available at our medical institution. The mean patient age was 59 years. Lesion location was the middle cerebral artery in five patients, the intradural vertebral artery in three, the basilar artery in one, the vertebrobasilar junction in one, and the cavernous internal cerebral artery in one patient. We evaluated the technical success, defined as reduction of residual stenosis ≤30% in the target lesion) as well as the clinical and imaging outcomes as long as 75 months following the procedure. In addition to a cerebral angiogram (n = 2), follow-up study was obtained by CT angiography (n = 6) or intracranial Doppler imaging (n = 2) during a mean time of 55 months after the procedure (range, 24 to 73 months). Three patients refused imaging follow-up and accepted only clinical follow-up. The mean clinical follow-up period was 67 months (range, 47-75 months). RESULTS: Stenting in all patients was technically successful and without periprocedural complications. There was thrombus formation during the procedure in one patient who experienced no further complications. There were no new neurological events during the mean follow-up period of 5.6 years. No patients were found to have restenosis ≥50% at during the mean follow-up period of 55 months. One patient died of a sudden heart attack 59 months following the procedure which was regarded as unrelated to the cerebral lesion. CONCLUSION: Our study demonstrates that DES shows long-term stability and safety, and results in good clinical outcomes with a low rate of restenosis.

Eosinophilic otitis media: CT and MRI findings and literature review.

Eosinophilic otitis media (EOM) is a relatively rare, intractable, middle ear disease with extremely viscous mucoid effusion containing eosinophils. EOM is associated with adult bronchial asthma and nasal allergies. Conventional treatments for otitis media with effusion (OME) or for chronic otitis media (COM), like tympanoplasty or mastoidectomy, when performed for the treatment of EOM, can induce severe complications such as deafness. Therefore, it should be differentiated from the usual type of OME or COM. To our knowledge, the clinical and imaging findings of EOM of temporal bone are not well-known to radiologists. We report here the CT and MRI findings of two EOM cases and review the clinical and histopathologic findings of this recently described disease entity.

Imaging findings in a case of mixed acinar-endocrine carcinoma of the pancreas.

Mixed acinar-endocrine carcinoma (MAEC) of the pancreas is extremely uncommon. We report here a rare case of MAEC of the pancreas presenting as watery diarrhea. This is the first report in the English-language literature that describes the imaging findings of MAEC of the pancreas, including computed tomography (CT), magnetic resonance (MR) imaging, and MR cholangiopancreatography features.