Index by author

Forty-two patients with high-grade gliomas were enrolled in this study. The final diagnosis was determined by histopathologic analysis or clinical follow-up. PWI and PET parameters were recorded and compared between patients with recurrence and those with radiation injury using Student t tests. Receiver operating characteristic and logistic regression analyses were used to determine the diagnostic performance of each parameter. The final diagnosis was recurrence in 33 patients and radiation injury in 9. PET/CT showed a patient-based sensitivity and specificity of 0.909 and 0.556, respectively, while PWI showed values of 0.667 and 0.778, respectively. The maximum standardized uptake value, mean standardized uptake value, tumor-to-background maximum standardized uptake value, and mean relative CBV were significantly higher for patients with recurrence than for patients with radiation injury. All these parameters showed a significant discriminative power in receiver operating characteristic analysis. Both ¹¹C-methionine PET/CT and PWI are equally accurate in the differentiation of recurrence from radiation injury in patients with high-grade gliomas, and a combination of the 2 modalities could result in increased diagnostic accuracy.

Forty patients with MS were prospectively included and scanned (3T) to acquire synthetic MR imaging and conventional FLAIR images. Synthetic FLAIR images were created with the SyMRI software. Acquired data were divided into 30 training and 10 test datasets. A conditional generative adversarial network was trained to generate improved FLAIR images from raw synthetic MR imaging data using conventional FLAIR images as targets. The peak signal-to-noise ratio, normalized root mean square error, and the Dice index of MS lesion maps were calculated for synthetic and deep learning FLAIR images against conventional FLAIR images, respectively. Lesion conspicuity and the existence of artifacts were visually assessed. The peak signal-to-noise ratio and normalized root mean square error were significantly higher and lower, respectively, in generated-versus-synthetic FLAIR images in aggregate intracranial tissues and all tissue segments. The Dice index of lesion maps and visual lesion conspicuity were comparable between generated and synthetic FLAIR images. Using deep learning, the authors conclude that they improved the synthetic FLAIR image quality by generating FLAIR images that have contrast closer to that of conventional FLAIR images and fewer granular and swelling artifacts, while preserving the lesion contrast.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

Of 101 included patients, the 4 subgroups of >6 were the following: chemotherapy (n = 35), opiates (n = 19), acute hepatic encephalopathy (n = 14), and immunosuppressants (n = 11). Other causes (n = 22 total) notably included carbon monoxide (n = 3) metronidazole (n = 2), and uremia (n = 1). Acute hepatic/hyperammonemic encephalopathy clinically resolved in 36%, with severe outcomes in 23% (coma or death, 9/16 deaths from fludarabine). Notable laboratory results were elevated CSF myelin basic protein levels in 8/9 patients and serum blood urea nitrogen levels in 24/91. Acute toxic leukoencephalopathy is an imaging appearance that can arise from various etiologies, with potentially reversible reduced diffusion predominately affecting the periventricular WM. Given the shared DWI appearance among this heterogeneous array of etiologies, their outcomes may differ. Thus, the neurologic symptoms completely resolved in 36%, while severe outcomes occurred in 23%. The clinical outcome was most severe with chemotherapy-related ATL.

Forty patients with MS were prospectively included and scanned (3T) to acquire synthetic MR imaging and conventional FLAIR images. Synthetic FLAIR images were created with the SyMRI software. Acquired data were divided into 30 training and 10 test datasets. A conditional generative adversarial network was trained to generate improved FLAIR images from raw synthetic MR imaging data using conventional FLAIR images as targets. The peak signal-to-noise ratio, normalized root mean square error, and the Dice index of MS lesion maps were calculated for synthetic and deep learning FLAIR images against conventional FLAIR images, respectively. Lesion conspicuity and the existence of artifacts were visually assessed. The peak signal-to-noise ratio and normalized root mean square error were significantly higher and lower, respectively, in generated-versus-synthetic FLAIR images in aggregate intracranial tissues and all tissue segments. The Dice index of lesion maps and visual lesion conspicuity were comparable between generated and synthetic FLAIR images. Using deep learning, the authors conclude that they improved the synthetic FLAIR image quality by generating FLAIR images that have contrast closer to that of conventional FLAIR images and fewer granular and swelling artifacts, while preserving the lesion contrast.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

Of 101 included patients, the 4 subgroups of >6 were the following: chemotherapy (n = 35), opiates (n = 19), acute hepatic encephalopathy (n = 14), and immunosuppressants (n = 11). Other causes (n = 22 total) notably included carbon monoxide (n = 3) metronidazole (n = 2), and uremia (n = 1). Acute hepatic/hyperammonemic encephalopathy clinically resolved in 36%, with severe outcomes in 23% (coma or death, 9/16 deaths from fludarabine). Notable laboratory results were elevated CSF myelin basic protein levels in 8/9 patients and serum blood urea nitrogen levels in 24/91. Acute toxic leukoencephalopathy is an imaging appearance that can arise from various etiologies, with potentially reversible reduced diffusion predominately affecting the periventricular WM. Given the shared DWI appearance among this heterogeneous array of etiologies, their outcomes may differ. Thus, the neurologic symptoms completely resolved in 36%, while severe outcomes occurred in 23%. The clinical outcome was most severe with chemotherapy-related ATL.

The purpose of this study was to develop a fast, automated method for assessing rescan need in motion-corrupted brain series. A deep learning–based approach was developed, outputting a probability for a series to be clinically useful. Comparison of this per-series probability with a threshold, which can depend on scan indication and reading radiologist, determines whether a series needs to be rescanned. The deep learning classification performance was compared with that of 4 technologists and 5 radiologists in 49 test series with low and moderate motion artifacts. Fast, automated deep learning–based image-quality rating can decrease rescan and recall rates, while rendering them technologist-independent. It was estimated that decreasing rescans and recalls from the technologists' values to the values of deep learning could save hospitals $24,000/scanner/year.

Forty patients with MS were prospectively included and scanned (3T) to acquire synthetic MR imaging and conventional FLAIR images. Synthetic FLAIR images were created with the SyMRI software. Acquired data were divided into 30 training and 10 test datasets. A conditional generative adversarial network was trained to generate improved FLAIR images from raw synthetic MR imaging data using conventional FLAIR images as targets. The peak signal-to-noise ratio, normalized root mean square error, and the Dice index of MS lesion maps were calculated for synthetic and deep learning FLAIR images against conventional FLAIR images, respectively. Lesion conspicuity and the existence of artifacts were visually assessed. The peak signal-to-noise ratio and normalized root mean square error were significantly higher and lower, respectively, in generated-versus-synthetic FLAIR images in aggregate intracranial tissues and all tissue segments. The Dice index of lesion maps and visual lesion conspicuity were comparable between generated and synthetic FLAIR images. Using deep learning, the authors conclude that they improved the synthetic FLAIR image quality by generating FLAIR images that have contrast closer to that of conventional FLAIR images and fewer granular and swelling artifacts, while preserving the lesion contrast.