The Super-Resolution Generative Adversarial Network (SR-GAN) [1] is a seminal work that is capable of generating realistic textures during single image super-resolution. However, the hallucinated details are often accompanied with unpleasant artifacts. To further enhance the visual quality, we thoroughly study three key components of SRGANnetwork architecture, adversarial loss and perceptual loss, and improve each of them to derive an Enhanced SRGAN (ESRGAN). In particular, we introduce the Residual-in-Residual Dense Block (RRDB) without batch normalization as the basic network building unit. Moreover, we borrow the idea from relativistic GAN [2] to let the discriminator predict relative realness instead of the absolute value. Finally, we improve the perceptual loss by using the features before activation, which could provide stronger supervision for brightness consistency and texture recovery. Benefiting from these improvements, the proposed ESRGAN achieves consistently better visual quality with more realistic and natural textures than SRGAN and won the first place in the PIRM2018-SR Challenge 1 [3]. The code is available at https://github.com/xinntao/ESRGAN.

No abstract

Maturity-onset diabetes of the young (MODY) is a form of non-insulin-dependent diabetes mellitus characterized by an, early age of onset, usually before 25 years of age, and an autosomal dominant mode of inheritance. The largest and best-studied MODY pedigree is the RW family. The majority of the diabetic subjects in this pedigree has a reduced and delayed insulin-secretory response to glucose, and it has been proposed that this abnormal response is the manifestation of the basic genetic defect that leads to diabetes. Using DNA from members of the'RW family, we tested more than 75 DNA markers for linkage with MODY. A DNA polymorphism in the adenosine deam.ase gene (ADA) on the long arm of chromosome 20 was found to cosegregate with MODY. The maximum logarithm of odds (lod score) for linkage between MODY and ADA was 5.25 at a recombination fraction of0.00. These results indicate that the odds are >178,000:1 that the gene responsible for MODY in this family is tightly linked to the ADA gene on chromosome 20q.Non-insulin-dependent or type 2 diabetes mellitus (NIDDM) is a common disorder of glucose homeostasis affecting -5% of the general population. The causes of the fasting hyperglycemia and/or glucose intolerance associated with this form of diabetes are not well understood. The contribution of heredity to the development of NIDDM has been recognized for many years (1), and the high degree of concordance of NIDDM in monozygotic twin pairs (2) indicates that genetic factors play an important role in its development. Since an understanding of the molecular basis of NIDDM would elucidate the mechanisms controlling glucose homeostasis and facilitate the development of more rational therapeutic strategies, we have undertaken a linkage study of NIDDM to identify diabetes-susceptibility genes. The use of linkage strategies to identify DNA markers for NIDDM has been difficult because this disorder does not exhibit simple Mendelian recessive or dominant inheritance. In addition, because of its late age of onset, it is difficult to obtain large multigenerational families suitable for genetic studies.

This genome-wide search for susceptibility genes to type 2 diabetes/impaired glucose homeostasis (IGH) was performed on a relatively homogenous Chinese sample with a total number of 257 pedigrees and 385 affected sibpairs. Two regions showed significant linkage to type 2 diabetes/IGH in the Chinese. T ype 2 diabetes is a complex disease that develops in individuals with genetic susceptibility to impaired insulin secretion, as well as to impaired insulin sensitivity, in the presence of appropriate environmental factors, particularly those leading to obesity (1). Marked increases in the prevalence of type 2 diabetes occur in those societies or countries that have experienced tremendous economic development from a starting point of an impoverished economic base (2). Along with the economic development in China, nationwide surveys have revealed an increase in the prevalence of diabetes in the adult population from 0.9 to 2.4% over the years 1980 -1994 (3). In Shanghai, China, the prevalence of diabetes was only 1.0% in 1978 but had reached 9.8% by the turn of the last century, i.e., there has been a 10-fold increase within the last two decades in the Shanghai population alone (4,5).Genetic heterogeneity of type 2 diabetes has been suggested among ethnic groups (6). This may be one of the reasons for the different locations of susceptibility loci reported to be linked to type 2 diabetes among ethnic groups in genome-wide scans (7-22). In the large geographic area of China, 56 ethnic groups are officially recognized, the Han being the largest. The Chinese of Han ethnicity reside throughout China, mostly in the eastern and central regions. Studies of the origin of the East Asian population revealed that, even within the Han ethnic group, considerable genetic heterogeneity might exist according to geographical location (23)(24)(25). Because appropriate definition of a more homogenous sample set is one of the issues for a genome-wide screen for type 2 diabetes susceptibility genes, geographical genetic heterogeneity should be considered when conducting such a study on the Chinese population. Thus, the genome-wide screen reported in this study was performed with Chinese pedigrees recruited from a limited geographic area in China. RESEARCH DESIGN AND METHODSPedigrees for this study were selected from a sampling scheme for the collection of multiplex diabetic families aimed at the genetic studies of type 2 Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org. ASP, affected sibpair; IA-2, protein tyrosine phosphatase-like protein; IGH, impaired glucose homeostasis; LOD, logarithm of odds; MLS, maximum likelihood score; MODY, maturity-onset diabetes of the young; NPL, nonparametric linkage.