Artificial Intelligence Trained by Deep Learning Can Improve Computed Tomography Diagnosis of Nontraumatic Subarachnoid Hemorrhage by Nonspecialists.

Subarachnoid hemorrhage (SAH) is a serious cerebrovascular disease with a high mortality rate and is known as a disease that is hard to diagnose because it may be overlooked by noncontrast computed tomography (NCCT) examinations that are most frequently used for diagnosis. To create a system preventing this oversight of SAH, we trained artificial intelligence (AI) with NCCT images obtained from 419 patients with nontraumatic SAH and 338 healthy subjects and created an AI system capable of diagnosing the presence and location of SAH. Then, we conducted experiments in which five neurosurgery specialists, five nonspecialists, and the AI system interpreted NCCT images obtained from 135 patients with SAH and 196 normal subjects. The AI system was capable of performing a diagnosis of SAH with equal accuracy to that of five neurosurgery specialists, and the accuracy was higher than that of nonspecialists. Furthermore, the diagnostic accuracy of four out of five nonspecialists improved by interpreting NCCT images using the diagnostic results of the AI system as a reference, and the number of oversight cases was significantly reduced by the support of the AI system. This is the first report demonstrating that an AI system improved the diagnostic accuracy of SAH by nonspecialists.

Alberta Stroke Program Early CT Score Calculation Using the Deep Learning-Based Brain Hemisphere Comparison Algorithm.

OBJECTIVES: The Alberta Stroke Program Early Computed Tomography Score (ASPECTS) is a promising tool for the evaluation of stroke expansion to determine suitability for reperfusion therapy. The aim of this study was to validate deep learning-based ASPECTS calculation software that utilizes a three-dimensional fully convolutional network-based brain hemisphere comparison algorithm (3D-BHCA). MATERIALS AND METHODS: We retrospectively collected head non-contrast computed tomography (CT) data from 71 patients with acute ischemic stroke and 80 non-stroke patients. The results for ASPECTS on CT assessed by 5 stroke neurologists and by the 3D-BHCA model were compared with the ground truth by means of region-based and score-based analyses. RESULTS: In total, 151 patients and 3020 (151 × 20) ASPECTS regions were investigated. Median time from onset to CT was 195 min in the stroke patients. In region-based analysis, the sensitivity (0.80), specificity (0.97), and accuracy (0.96) of the 3D-BHCA model were superior to those of stroke neurologists. The sensitivity (0.98), specificity (0.92), and accuracy (0.97) of dichotomized ASPECTS > 5 analysis and the intraclass correlation coefficient (0.90) in total score-based analysis of the 3D-BHCA model were superior to those of stroke neurologists overall. When patients with stroke were stratified by onset-to-CT time, the 3D-BHCA model exhibited the highest performance to calculate ASPECTS, even in the earliest time period. CONCLUSIONS: The automated ASPECTS calculation software we developed using a deep learning-based algorithm was superior or equal to stroke neurologists in performing ASPECTS calculation in patients with acute stroke and non-stroke patients.

Comparison of dosimetric properties among four commercial multi-detector computed tomography scanners.

This study compared dosimetric properties among four commercial multi-detector CT (MDCT) scanners. The X-ray beam characteristics were obtained from photon intensity attenuation curves of aluminum and off-center ratio (OCR) profiles in air, which were measured with four commercial MDCT scanners. The absorbed dose for MDCT scanners was evaluated with Farmer ionization chamber measurements at the center and four peripheral points in the body- and head-type cylindrical water phantoms. Measured collected charge was converted to absorbed dose using a 60Co absorbed dose-to-water calibration factor and Monte Carlo (MC)-calculated correction factors. Four MDCT scanners were modeled to correspond with measured X-ray beam characteristics using GMctdospp (IMPS, Germany) software. Al half-value layers (Al-HVLs) with a body-bowtie filter determined from measured Al-attenuation curves ranged 7.2‒9.1mm at 120kVp and 6.1‒8.0mm at 100kVp. MC-calculated Al-HVLs and OCRs in air were in acceptable agreement within 0.5mm and 5% of measured values, respectively. The percentage difference between nominal and actual beam width was greater with decreasing collimation width. The absorbed doses for MDCT scanners at 120kVp ranged 5.1‒7.1mGy and 10.8‒17.5mGy per 100mAs at the center in the body- and head-type water phantoms, respectively. Measured doses at four peripheral points were within 5% agreement of MC-calculated values. The absorbed dose at the center in both water phantoms increased with decreasing Al-HVL for the same charge on the focus. In this study the X-ray beam characteristics and the absorbed dose were measured and compared with calculated values for four MDCT scanners.

Incidence and severity of acute adverse reactions to four different gadolinium-based MR contrast agents.

PURPOSE: Differences in acute adverse reactions to different gadolinium (Gd)-based contrast agents have not been thoroughly evaluated. We investigated the relationships among the incidence and severity of acute adverse reactions, backgrounds of patients, and 4 types of different Gd-based contrast agents (gadopentetate dimeglumine, gadoteridol, gadoterate meglumine, and gadoxetate disodium). MATERIALS AND METHODS: We retrospectively reviewed the radiological records of 10,595 consecutive patients (4,343 female; 6,252 male; mean age, 63.8 ± 14.0 years) who underwent contrast-enhanced magnetic resonance imaging between August 2006 and March 2011. Adverse reactions were classified as mild, moderate, and severe according to the definition of the American College of Radiology. The incidence of adverse reactions were compared on the basis of clinical characteristics and type, dose, and delivery methods of contrast agents by univariate and multivariate logistic regression analyses. RESULTS: The incidence of overall reactions was 0.45% (48/10,595); 45 reactions were mild and three were moderate. No severe reactions were observed. Although the incidence of adverse reactions did not differ significantly between male and female patients, younger individuals were at higher risk for acute adverse reactions. The contrast injection rate and contrast dose were not significantly related to the incidence of adverse reactions. The incidence of adverse reactions was significantly higher for gadoxetate disodium (0.82%) than gadopentetate dimeglumine (0.43%). CONCLUSION: The incidence of acute adverse reactions elicited by Gd-based contrast agents injection was only 0.45%. Younger age was a risk factor for acute reactions. All 4 agents were found to be safe, although gadoxetate disodium showed a relatively higher incidence of adverse reactions.