Philbrick, Kenneth

There is a growing demand for high-resolution (HR) medical images for both clinical and research applications. Image quality is inevitably traded off with acquisition time, which in turn impacts patient comfort, examination costs, dose, and motion-induced artifacts. For many image-based tasks, increasing the apparent spatial resolution in the perpendicular plane to produce multi-planar reformats or 3D images is commonly used. Single-image super-resolution (SR) is a promising technique to provide HR images based on deep learning to increase the resolution of a 2D image, but there are few reports on 3D SR. Further, perceptual loss is proposed in the literature to better capture the textural details and edges versus pixel-wise loss functions, by comparing the semantic distances in the high-dimensional feature space of a pre-trained 2D network (e.g., VGG). However, it is not clear how one should generalize it to 3D medical images, and the attendant implications are unclear. In this paper, we propose a framework called SOUP-GAN: Super-resolution Optimized Using Perceptual-tuned Generative Adversarial Network (GAN), in order to produce thinner slices (e.g., higher resolution in the ‘Z’ plane) with anti-aliasing and deblurring. The proposed method outperforms other conventional resolution-enhancement methods and previous SR work on medical images based on both qualitative and quantitative comparisons. Moreover, we examine the model in terms of its generalization for arbitrarily user-selected SR ratios and imaging modalities. Our model shows promise as a novel 3D SR interpolation technique, providing potential applications for both clinical and research applications.

(1) Purpose: To determine the association between visceral adipose tissue (VAT) and proton density fat fraction (PDFF) with magnetic resonance imaging (MRI), and hepatic steatosis (HS), non-alcoholic steatohepatitis (NASH) and hepatic fibrosis (HF) in patients with known or suspected non-alcoholic fatty liver disease (NAFLD). (2) Methods: 135 subjects that had a liver biopsy performed within 3 months (bariatric cohort) or 1 month (NAFLD cohort) of an MRI exam formed the study group. VAT volume was quantified at L2-L3 level on opposed-phase images with signal intensity-based painting using a semi-quantitative software. Liver PDFF and pancreas PDFF were calculated on fat fraction maps. Liver volume (Lvol) and spleen volume (Svol) were also calculated using a semi-automated 3D volume tool available on PACS. A histological analysis was performed by an expert hepatopathologist blinded to imaging findings. (3) Results: The mean Lvol, Svol, liver PDFF, pancreas PDFF and VAT of the study population were 2492.2 mL, 381.6 mL, 13.2%, 12.7% and 120.6 mL, respectively. VAT showed moderate correlation with liver PDFF (r = 0.41, p < 0.001) and weak correlation with Lvol (r = 0.38, p < 0.001), Svol (r = 0.20, p = 0.025) and pancreas PDFF (rs = 0.29, p = 0.001). VAT, Lvol and liver PDFF were significantly higher in patients with HS (p < 0.001), NASH (p < 0.05) and HF (p < 0.05). VAT was also significantly higher in the presence of lobular inflammation (p = 0.019) and hepatocyte ballooning (p = 0.001). The cut-off VAT volumes for predicting HS, NASH and HF were 101.8 mL (AUC, 0.7), 111.8 mL (AUC, 0.64) and 111.6 mL (AUC, 0.66), respectively. (4) Conclusion: The MRI determined VAT can be used for predicting the presence of HS, NASH and HF in patients with known or suspected NAFLD.