Contrastive self-supervised learning for diabetic retinopathy early detection.

Diabetic Retinopathy (DR) is the major cause of blindness, which seriously threatens the world's vision health. Limited medical resources make early diagnosis and a large-scale screening of DR difficult. Most of the current automatic diagnostic methods are mostly based on deep learning and large-scale labeled data. However, the insufficiency of manual annotations for medical images still is a great challenge of training deep neural networks. Self-supervised learning methods are proposed to learn general features from dataset without manual annotations. Inspired by this, we proposed a deep learning based DR classification model (SimCLR-DR). In this paper, we first use contrastive self-learning algorithm to pre-train the encoder based on convolution network with unlabeled retinal images, then retrain the encoder with classifier on a small annotated training data to detect referable DR. The experimental results on Kaggle dataset show that this proposed method can overcome the training data insufficiency problem and performs better than transfer learning. SimCLR-DR is a good beginning for other deep learning based medical image detection approaches facing the challenge of insufficient annotated data. Figure presents an overview of the proposed framework, which contains three main steps: (i) Data preprocessing; (ii) Pretext task of SimCLR-DR based on contrastive learning; (iii) Downstream Task of SimCLRDR based on CNN.

No Association of Homocysteine, Anticardiolipin Antibody, and Anti-ß2 Glycoprotein I Antibody With Portal Venous System Thrombosis in Liver Cirrhosis.

Portal venous system thrombosis (PVST), a common complication of liver cirrhosis, is closely associated with thrombophilia. To explore the association of homocysteine (Hcy), anticardiolipin antibody (aCL), and anti-ß2 glycoprotein I antibody (aß2GPI), which are possible thrombophilic factors, with PVST in liver cirrhosis. Overall, 654 non-malignant patients (219 with and 435 without liver cirrhosis) admitted between January 2016 and June 2020 were retrospectively evaluated. Presence of PVST, degree of main portal vein (MPV) thrombosis, and clinically significant PVST were identified. Hcy level, hyperhomocysteinemia (HHcy), aCL positivity, and aß2GPI positivity were compared according to the presence of liver cirrhosis and PVST. Positive aß2GPI was significantly more frequent in patients with liver cirrhosis than those without, but Hcy level and proportions of HHcy and positive aCL were not significantly different between them. PVST could be evaluated in 136 cirrhotic patients. Hcy level [10.57 µmol/L (2.71-56.82) versus 9.97 µmol/L (2.05-53.44); P = 0.796] and proportions of HHcy [4/44 (9.1%) versus 13/81 (16.0%); P = 0.413] and positive aCL [1/23 (4.3%) versus 10/52 (19.2%); P = 0.185] and aß2GPI [9/23 (39.1%) versus 21/52 (40.4%); P = 0.919] were not significantly different between cirrhotic patients with and without PVST. There was still no significant association of Hcy level, HHcy, aCL, or aß2GPI with PVST based on Child-Pugh classification, MPV thrombosis >50%, and clinically significant PVST. Hcy, aCL, and aß2GPI may not be associated with PVST in liver cirrhosis, suggesting that routine screening for Hcy, aCL, and aß2GPI should be unnecessary in such patients.

Correlation of Serum Cardiac Markers with Acute Decompensating Events in Liver Cirrhosis.

METHODS: Cirrhotic patients who were consecutively hospitalized between January 2016 and March 2019 were screened. Serum cardiac biomarkers at admission, including N-Terminal pro-B-type natriuretic peptide (NT-pro BNP), high-sensitivity cardiac troponin T (hs-cTnT), creatine kinase (CK), creatine kinase MB (CK-MB), and lactate dehydrogenase (LDH), were collected. Acute decompensating events at admission, primarily including ascites, acute gastrointestinal hemorrhage, and acute-on-chronic liver failure (ACLF), were recorded. RESULTS: The NT-pro BNP level was significantly higher in cirrhotic patients with acute decompensating events than in those without any decompensating events (median: 140.75 pg/mL versus 41.86 pg/mL, P < 0.001). The NT-pro BNP level significantly correlated with ascites, acute gastrointestinal hemorrhage, and ACLF. The hs-cTnT level was significantly higher in cirrhotic patients with acute decompensating events than in those without decompensating events (median: 0.008 ng/mL versus 0.006 ng/mL, P = 0.007). The hs-cTnT level significantly correlated with acute gastrointestinal hemorrhage, but not ascites or ACLF. LDH (185.0 U/L versus 173.5 U/L, P = 0.281), CK (71 U/L versus 84 U/L, P = 0.157), and CK-MB (29.5 U/L versus 33.0 U/L, P = 0.604) levels were not significantly different between cirrhotic patients with and without acute decompensating events. CONCLUSION: The elevated NT-pro BNP level seems to be closely related to the development of acute decompensating events in liver cirrhosis.

Association of non-alcoholic fatty liver disease with thyroid function: A systematic review and meta-analysis.

BACKGROUND: Nonalcoholic fatty liver disease (NAFLD) is one of the most common chronic liver diseases. The relationship of NAFLD with thyroid function parameters and hypothyroidism remains controversial. AIM: To clarify the effect of thyroid function parameters and hypothyroidism on the development of NAFLD and progression to nonalcoholic steatohepatitis (NASH). METHODS: PubMed, EMBASE, and Cochrane library databases were searched. Study quality was assessed. Weighted mean difference (WMD) and odds ratio (OR) with 95% confidence interval (CI) were calculated. RESULTS: Twenty six studies involving 61,548 participants were eligible, most of which were of high quality. NAFLD/NASH patients had significantly higher TSH levels than controls in adults (NAFLD versus health: WMD = 0.105, 95%CI = 0.012-0.197; NAFLD versus euthyroidism: WMD = 0.100, 95%CI = 0.005-0.194; NASH versus NAFLD: WMD = 0.540, 95%CI = 0.136-0.944) and children/adolescents (NAFLD versus lean controls: WMD = 1.039, 95%CI = 0.104-1.973; NAFLD versus overweight/obese controls: WMD = 0.485, 95%CI = 0.267-.703). Unclassified hypothyroidism was positively associated with the risk of NAFLD/NASH in adults (NAFLD versus health: OR = 1.605, 95%CI = 1.180-2.183; NASH versus NAFLD: OR = 2.317, 95%CI = 1.425-3.768) and children/adolescents (NAFLD versus overweight/obese controls: OR = 2.015, 95%CI = 1.246-3.258). However, the statistical results were inconsistent among the subgroup meta-analyses of subclinical and overt hypothyroidism. Association of NAFLD with FT3 and FT4 levels was heterogeneous among population. CONCLUSION: TSH level may be an important risk factor for the development and progression of NAFLD, independent of thyroid hormones.

Prevalence of CALR mutations in splanchnic vein thrombosis: A systematic review and meta-analysis.

BACKGROUND: The prevalence of calreticulin (CALR) mutations in splanchnic vein thrombosis (SVT) varies among studies. The role of routine screening for CALR mutations in SVT patients remains a debate. AIM: To synthesize the prevalence of CALR mutations according to the different types (i.e., Budd-Chiari syndrome [BCS] and portal vein thrombosis [PVT]) and characteristics (i.e., with and without myeloproliferative neoplasms [MPNs] and JAK2V617F mutation) of SVT patients. METHODS: Eligible studies were searched by the PubMed and Embase databases. The study quality was assessed according to the STROBE checklist. The proportion of CALR mutations was pooled by using a random-effects model. The heterogeneity and publication bias were calculated. RESULTS: Eleven papers were included. The study quality was moderate to high. The pooled proportion of CALR mutations was 1.21%, 1.41%, and 1.59% in SVT, BCS, and PVT patients, respectively; 1.52%, 1.03%, and 1.82% in these patients without JAK2V617F mutation, respectively; 3.71%, 2.79%, and 7.87% in these patients with MPN, respectively; and 15.16%, 17.22%, and 31.44% in these patients with MPN but without JAK2V617F mutation, respectively. Only the meta-analysis examining the prevalence of CLAR mutations in BCS patients with MPN but without the JAK2V617F mutation showed statistically significant heterogeneity. Statistically significant publication bias was seen only in the meta-analysis examining the prevalence of CALR mutations in SVT patients without the JAK2V617F mutation. CONCLUSION: Screening for CALR mutations may have a role in SVT patients with a high probability of MPN in whom the JAK2V617F mutation has been excluded.