Cardiothoracic ratio measurement using artificial intelligence: observer and method validation studies.

BACKGROUND: Artificial Intelligence (AI) is a promising tool for cardiothoracic ratio (CTR) measurement that has been technically validated but not clinically evaluated on a large dataset. We observed and validated AI and manual methods for CTR measurement using a large dataset and investigated the clinical utility of the AI method. METHODS: Five thousand normal chest x-rays and 2,517 images with cardiomegaly and CTR values, were analyzed using manual, AI-assisted, and AI-only methods. AI-only methods obtained CTR values from a VGG-16 U-Net model. An in-house software was used to aid the manual and AI-assisted measurements and to record operating time. Intra and inter-observer experiments were performed on manual and AI-assisted methods and the averages were used in a method variation study. AI outcomes were graded in the AI-assisted method as excellent (accepted by both users independently), good (required adjustment), and poor (failed outcome). Bland-Altman plot with coefficient of variation (CV), and coefficient of determination (R-squared) were used to evaluate agreement and correlation between measurements. Finally, the performance of a cardiomegaly classification test was evaluated using a CTR cutoff at the standard (0.5), optimum, and maximum sensitivity. RESULTS: Manual CTR measurements on cardiomegaly data were comparable to previous radiologist reports (CV of 2.13% vs 2.04%). The observer and method variations from the AI-only method were about three times higher than from the manual method (CV of 5.78% vs 2.13%). AI assistance resulted in 40% excellent, 56% good, and 4% poor grading. AI assistance significantly improved agreement on inter-observer measurement compared to manual methods (CV; bias: 1.72%; - 0.61% vs 2.13%; - 1.62%) and was faster to perform (2.2 ± 2.4 secs vs 10.6 ± 1.5 secs). The R-squared and classification-test were not reliable indicators to verify that the AI-only method could replace manual operation. CONCLUSIONS: AI alone is not yet suitable to replace manual operations due to its high variation, but it is useful to assist the radiologist because it can reduce observer variation and operation time. Agreement of measurement should be used to compare AI and manual methods, rather than R-square or classification performance tests.

Correlation Between Apparent Diffusion Coefficient Values of the Renal Parenchyma and Estimated Glomerular Filtration Rates on 3-T Diffusion-Weighted Echo-Planar Magnetic Resonance Imaging.

OBJECTIVE: To evaluate the relationship between the apparent diffusion coefficient (ADC) values of renal parenchyma and estimated glomerular filtration rates (eGFR). METHODS: Data on 216 patients examined by 3-T magnetic resonance imaging for various reasons were retrospectively collected. RESULTS: There was a significant linear correlation between the ADC values and eGFRs (r = 0.254, P < 0.001). The ADC values in patients with an eGFR of less than 60 mL/min per 1.73 m were significantly lower than those with an eGFR of 60 mL/min per 1.73 m or greater. The mean ADC value of patients with grouped stage 2 disease was significantly higher than those with grouped stage 3 of chronic kidney disease (P < 0.01). CONCLUSIONS: The ADC value of renal parenchyma may be a promising marker for the determination of patients with normal to mild reduction in renal function (eGFR ≥60 mL/min per 1.73 m) versus those with a moderate to severe reduction in renal function (eGFR <60 mL/min per 1.73 m).

The different faces of renal angiomyolipomas on radiologic imaging: a pictorial review.

Renal angiomyolipoma (AML) is an uncommon renal tumour, generally composed of mature adipose tissue, dysmorphic blood vessels and smooth muscle. Identification of intratumoral fat on unenhanced CT images is the most reliable finding for establishing the diagnosis of renal AML. However, AMLs sometimes exhibit atypical findings, including cystic as well as solid forms; some of these variants overlap with the appearance of other renal tumours. A rare type of AML, the epithelioid type, possesses malignant potential. The aim of this pictorial review is to gather the different imaging features of AMLs including the classic and fat-poor types, AMLs with epithelial cysts, epithelioid AML, AML associated with tuberous sclerosis, haemorrhagic AML and large AMLs mimicking retroperitoneal liposarcomas. The diagnostic clues that help to distinguish AMLs from other renal tumours are also described in the review.

High-Flow Nasal Cannula Versus Conventional Oxygen Therapy in Emergency Department Patients With Cardiogenic Pulmonary Edema: A Randomized Controlled Trial.

STUDY OBJECTIVE: High-flow nasal cannula is a new method for delivering high-flow supplemental oxygen for victims of respiratory failure. This randomized controlled trial compares high-flow nasal cannula with conventional oxygen therapy in emergency department (ED) patients with cardiogenic pulmonary edema. METHODS: We conducted an open-label randomized controlled trial in the ED of Siriraj Hospital, Bangkok, Thailand. Patients aged 18 years or older with cardiogenic pulmonary edema were randomly assigned to receive either conventional oxygen therapy or high-flow nasal cannula. The primary outcome was the respiratory rate 60 minutes postintervention. RESULTS: We enrolled 128 participants (65 in the conventional oxygen therapy and 63 in the high-flow nasal cannula groups). Baseline high-flow nasal cannula and conventional oxygen therapy mean respiratory rates were 28.7 breaths/min (SD 3.2) and 28.6 breaths/min (SD 3.5). Mean respiratory rates at 60 minutes postintervention were lower in the high-flow nasal cannula group (21.8 versus 25.1 breaths/min; difference 3.3; 95% confidence interval 1.9 to 4.6). No significant differences were found in the admission rate, ED and hospital lengths of stay, noninvasive ventilation, intubation, or mortality. CONCLUSION: In patients with cardiogenic pulmonary edema in the ED, high-flow nasal cannula therapy may decrease the severity of dyspnea during the first hour of treatment.

MRI of placenta percreta: differentiation from other entities of placental adhesive disorder.

OBJECTIVES: To retrospectively review the MRI findings of placenta percreta and identify those helpful for differentiation from non-placenta percreta. MATERIALS AND METHODS: The MRI images of 21 patients with a preliminary diagnosis of placental adhesive disorder scanned between 2005 and 2014 were evaluated. Radiologists blinded to the final diagnosis evaluated six previously described MRI findings of placenta adhesive disorder. The sensitivity, specificity, accuracy, negative predictive value (NPV), and positive predictive value (PPV) of MRI for the diagnosis of placenta percreta were also calculated. RESULTS: The study included 12 cases of placenta percreta and 9 cases of non-placenta percreta. Invasion of placental tissue outside the uterus was found only in placenta percreta (p = 0.045; sensitivity 41.7 %; specificity 100 %). All placenta percreta cases also had a moderate to marked degree of heterogeneous placental signal intensity (p = 0.063; sensitivity 100 %; specificity 33.3 %). The size of the dark bands on T2-weighted imaging, and the presence of disorganized intra-placental vessels, showed no statistically significant difference between placenta percreta and non-placenta percreta. The sensitivity, specificity, NPV, PPV, and accuracy of MRI for detection of placenta percreta were 91.7, 44, 80, 68, and 71.4 %, respectively. CONCLUSIONS: MRI is recommended for the evaluation of placenta percreta, with the most specific signs including the invasion of placental tissue outside the uterus on B-FFE sequences, and consideration of the degree of placental signal heterogeneity. The size of the T2 dark band alone, or bizarre disorganized intra-placental vessels, did not correlate with the severity of invasion.