To develop algorithms using convolutional neural networks (CNNs) for automatic segmentation of acute ischemic lesions on diffusion-weighted imaging (DWI) and compare them with conventional algorithms, including a thresholding-based segmentation. Between September 2005 and August 2015, 429 patients presenting with acute cerebral ischemia (training:validation:test set = 246:89:94) were retrospectively enrolled in this study, which was performed under Institutional Review Board approval. Ground truth segmentations for acute ischemic lesions on DWI were manually drawn under the consensus of two expert radiologists. CNN algorithms were developed using two-dimensional U-Net with squeeze-and-excitation blocks (U-Net) and a DenseNet with squeeze-and-excitation blocks (DenseNet) with squeeze-and-excitation operations for automatic segmentation of acute ischemic lesions on DWI. The CNN algorithms were compared with conventional algorithms based on DWI and the apparent diffusion coefficient (ADC) signal intensity. The performances of the algorithms were assessed using the Dice index with 5-fold cross-validation. The Dice indices were analyzed according to infarct volumes (< 10 mL, ≥ 10 mL), number of infarcts (≤ 5, 6-10, ≥ 11), and b-value of 1000 (b1000) signal intensities (< 50, 50-100, > 100), time intervals to DWI, and DWI protocols. The CNN algorithms were significantly superior to conventional algorithms (p < 0.001). Dice indices for the CNN algorithms were 0.85 for U-Net and DenseNet and 0.86 for an ensemble of U-Net and DenseNet, while the indices were 0.58 for ADC-b1000 and b1000-ADC and 0.52 for the commercial ADC algorithm. The Dice indices for small and large lesions, respectively, were 0.81 and 0.88 with U-Net, 0.80 and 0.88 with DenseNet, and 0.82 and 0.89 with the ensemble of U-Net and DenseNet. The CNN algorithms showed significant differences in Dice indices according to infarct volumes (p < 0.001). The CNN algorithm for automatic segmentation of acute ischemic lesions on DWI achieved Dice indices greater than or equal to 0.85 and showed superior performance to conventional algorithms.

Background The black-blood (BB) technique was developed to suppress the signal from blood and cerebrospinal fluid (CSF) to provide improved depiction of vessel walls. Purpose The aim was to compare three-dimensional turbo spin echo T1-weighted imaging (3D TSE T1WI) with or without two BB techniques (delay alternating with nutation for tailored excitation [DANTE], and improved motion-sensitized driven equilibrium [iMSDE]) for high-resolution magnetic resonance imaging (HR-MRI) of the vessel walls of intracranial arteries. Study type Prospective. Population Fourteen healthy volunteers who underwent 3D T1WI for examination of intracranial vessel walls. Field strength/Sequence 3 Tesla, 3D TSE T1WI (SPACE and BrainVIEW) and BB (DANTE and iMSDE). Assessment SPACE with or without DANTE, and BrainVIEW with or without iMSDE, were acquired in each subject. Two neuroradiologists independently assessed image quality, vessel wall delineation, BB effect, CSF, and acceptability using visual scoring systems, and measured signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) in vessel walls, lumen, and CSF, while blinded to the presence and type of BB technique used. Statistical tests Repeated measures ANOVA or Friedman tests were performed for the comparisons, followed by Bonferroni correction. Results The 3T T1WI sequences without BB are significantly superior in vessel wall delineation (P = 0.001). Black CSF scores were lower in SPACE with DANTE than SPACE without DANTE, and in BrainVIEW without iMSDE than SPACE without DANTE (P < 0.001). However, there were no significant differences in BB effect, image quality, and acceptability between the four 3D T1WI sequences (p >.05). The SNRVessel wall, CNRWall-Lumen, and CNRWall-CSF were higher (all p <.001) on SPACE with and without DANTE than on BrainVIEW with and without iMSDE. SNRLumen were higher (all p <.001) on BrainVIEW with and without iMSDE than on SPACE with and without DANTE. SNRCSF was higher (all p <.001) on BrainVIEW with iMSDE than on SPACE with DANTE. Data conclusion Both 3D TSE T1WI sequences were acceptable for intracranial vessel wall evaluation, with or without BB techniques. Therefore, BB techniques may not necessarily be required with 3D TSE T1WI with a long ETL and TR (below 1160 ms).

Introduction: Carbon monoxide (CO) poisoning can result in delayed neurological sequelae (DNS). Factors predicting DNS are still controversial. This study aims to determine whether acute brain lesions observed using diffusion-weighted magnetic resonance imaging (MRI) following acute CO poisoning are related to the subsequent development of DNS. Methods: This prospective study was conducted on patients with CO poisoning treated at a university hospital in Bucheon, Korea. From August 2016 to July 2019, a total of 283 patients visited the hospital because of CO poisoning. Exclusion criteria included age under 18 years, refusing hyperbaric oxygen therapy, refusing MRI, being discharged against medical advice, being lost to follow-up, having persistent neurological symptoms at discharge, and being transferred from another hospital 24 h after exposure. Results: Of the 154 patients included in the final study, acute brain lesions on MRI (ABLM) were observed in 49 patients (31.8%) and DNS occurred in 30 patients (19.5%). In a logistic regression analysis, lower Glasgow coma scale score and higher exposure time were associated with DNS, and the presence of ABLM in white matter was significantly associated with DNS (OR 6.741; 95% CI, 1.843–24.660; p = 0.004). Conclusion: The presence of ABLM in white matter was significantly related to the occurrence of DNS. Early prediction of the risk of developing DNS through MRI may be helpful in treating patients with CO poisoning.