Three-Dimensional Human Pose Estimation from Sparse IMUs through Temporal Encoder and Regression Decoder.

Three-dimensional (3D) pose estimation has been widely used in many three-dimensional human motion analysis applications, where inertia-based path estimation is gradually being adopted. Systems based on commercial inertial measurement units (IMUs) usually rely on dense and complex wearable sensors and time-consuming calibration, causing intrusions to the subject and hindering free body movement. The sparse IMUs-based method has drawn research attention recently. Existing sparse IMUs-based three-dimensional pose estimation methods use neural networks to obtain human poses from temporal feature information. However, these methods still suffer from issues, such as body shaking, body tilt, and movement ambiguity. This paper presents an approach to improve three-dimensional human pose estimation by fusing temporal and spatial features. Based on a multistage encoder-decoder network, a temporal convolutional encoder and human kinematics regression decoder were designed. The final three-dimensional pose was predicted from the temporal feature information and human kinematic feature information. Extensive experiments were conducted on two benchmark datasets for three-dimensional human pose estimation. Compared to state-of-the-art methods, the mean per joint position error was decreased by 13.6% and 19.4% on the total capture and DIP-IMU datasets, respectively. The quantitative comparison demonstrates that the proposed temporal information and human kinematic topology can improve pose accuracy.

Normalization of γ-glutamyl transferase levels is associated with better metabolic control in individuals with nonalcoholic fatty liver disease.

BACKGROUND: The normalization of liver biochemical parameters usually reflects the histological response to treatment for nonalcoholic fatty liver disease (NAFLD). Researchers have not clearly determined whether different liver enzymes exhibit various metabolic changes during the follow-up period in patients with NAFLD. METHODS: We performed a retrospective analysis of patients with NAFLD who were receiving therapy from January 2011 to December 2019. Metabolism indexes, including glucose levels, lipid profiles, uric acid levels and liver biochemical parameters, were measured. Magnetic resonance imaging-based proton density fat fraction (MRI-PDFF) and liver ultrasound were used to evaluate steatosis. All patients received recommendations for lifestyle modifications and guideline-recommended pharmacological treatments with indications for drug therapy for metabolic abnormalities. RESULTS: Overall, 1048 patients with NAFLD were included and received lifestyle modification recommendations and pharmaceutical interventions, including 637 (60.7%) patients with abnormal GGT levels and 767 (73.2%) patients with abnormal ALT levels. Patients with concurrent ALT and GGT abnormalities presented higher levels of metabolism indexes and higher liver fat content than those in patients with single or no abnormalities. After 12 months of follow-up, the cumulative normalization rate of GGT was considerably lower than that of ALT (38% vs. 62%, P < 0.001). Greater weight loss resulted in higher cumulative normalization rates of GGT and ALT. Weight loss (OR = 1.21, 95% CI 1.11-1.32, P < 0.001), ALT normalization (OR = 2.75, 95% CI 1.41-5.36, P = 0.01) and lower TG and HOMA-IR values (OR = 2.03, 95% CI 1.11-3.71, P = 0.02; OR = 2.04, 95% CI 1.07-3.89, P = 0.03) were independent protective factors for GGT normalization. Elevated baseline GGT (OR = 0.99, 95% CI 0.98-0.99, P = 0.01) was a risk factor. CONCLUSIONS: For NAFLD patients with concurrently increased ALT and GGT levels, a lower normalization rate of GGT was observed, rather than ALT. Good control of weight and insulin resistance was a reliable predictor of GGT normalization.

Reconstructing 3D human pose and shape from a single image and sparse IMUs.

Model-based 3D pose estimation has been widely used in many 3D human motion analysis applications, in which vision-based and inertial-based are two distinct lines. Multi-view images in a vision-based markerless capture system provide essential data for motion analysis, but erroneous estimates still occur due to ambiguities, occlusion, or noise in images. Besides, the multi-view setting is hard for the application in the wild. Although inertial measurement units (IMUs) can obtain accurate direction without occlusion, they are usually susceptible to magnetic field interference and drifts. Hybrid motion capture has drawn the attention of researchers in recent years. Existing 3D pose estimation methods jointly optimize the parameters of the 3D pose by minimizing the discrepancy between the image and IMU data. However, these hybrid methods still suffer from the issues such as complex peripheral devices, sensitivity to initialization, and slow convergence. Methods: This article presents an approach to improve 3D human pose estimation by fusing a single image with sparse inertial measurement units (IMUs). Based on a dual-stream feature extract network, we design a model-attention network with a residual module to closely couple the dual-modal feature from a static image and sparse inertial measurement units. The final 3D pose and shape parameters are directly obtained by a regression strategy. Results: Extensive experiments are conducted on two benchmark datasets for 3D human pose estimation. Compared to state-of-the-art methods, the per vertex error (PVE) of human mesh reduces by 9.4 mm on Total Capture dataset and the mean per joint position error (MPJPE) reduces by 7.8 mm on the Human3.6M dataset. The quantitative comparison demonstrates that the proposed method could effectively fuse sparse IMU data and images and improve pose accuracy.

Lipid-Lowering Responses to Dyslipidemia Determine the Efficacy on Liver Enzymes in Metabolic Dysfunction-Associated Fatty Liver Disease with Hepatic Injuries: A Prospective Cohort Study.

Purpose: Effective treatment of dyslipidemia with lipid-lowering agents is pivotal in the management of metabolic-associated fatty liver disease (MAFLD) for preventing cardiovascular complications. We explored the associations between improvements in liver injuries indicated by changes in transaminases and a reduction in lipid levels in MAFLD patients with dyslipidemia and elevated transaminases during lipid-lowering therapies. Methods: This prospective, cohort study enrolled consecutive MAFLD patients with hyperlipidemia and elevated transaminases. Patients were divided into a group receiving lipid-lowering agents and an age-, sex- and baseline lipid level-matched control group without receiving lipid-lowering agents. Clinical visits were performed at the 1st month and then every 3 months for 1 year. Results: This study included 541 MAFLD patients (lipid-lowering group: 325 patients; control group: 216 patients). Compared with controls, there was a substantially greater reduction in alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma glutamyl transpeptidase (GGT), triglyceride (TG), total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-c) in the lipid-lowering group after 12 months (all P < 0.05). The decrease in ALT was positively correlated with the decrease in TC (r = 0.332), TG (r = 0.180), LDL-c (r = 0.253) and apolipoprotein E (ApoE) (r = 0.119), while the decrease in AST was positively correlated with the decrease in TC (r = 0.228) and LDL-c (r = 0.192) (all P<0.05). The greater range of reduction in blood lipids (TC/TG/LDL-c), the higher the transaminase and GGT normalization rate (all P<0.05). Multivariate analysis confirmed that a TG decrease of over 50% remained an independent predictor of transaminase and GGT normalization (OR 2.07, 95% CI 1.12-3.84, P=0.020). Conclusion: Lipid-lowering to target levels might be beneficial to liver injury improvements in MAFLD patients with dyslipidemia when receiving lipid-lowering agents.

Distinct Dose-Dependent Association of Free Fatty Acids with Diabetes Development in Nonalcoholic Fatty Liver Disease Patients.

BACKGROUND: Excessive delivery of free fatty acids (FFAs) to the liver promotes steatosis and insulin resistance (IR), with IR defined as reduced glucose uptake, glycogen synthesis and anti-lipolysis stimulated by normal insulin levels. Whether the associations between FFAs and diabetes development differ between patients with and without nonalcoholic fatty liver disease (NAFLD) remains unclear. METHODS: Consecutive subjects (2,220 NAFLD subjects and 1,790 non-NAFLD subjects according to ultrasound imaging) were enrolled from the First Affiliated Hospital of Sun Yat-sen University between 2009 and 2019. The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated. RESULTS: There was an approximate J-shaped relationship between FFA levels and HOMA-IR in the NAFLD group. Higher FFA concentration quartiles were associated with higher risks of IR (odds ratio [OR], 9.24; 95% confidence interval [CI], 6.43 to 13.36), prediabetes (OR, 10.48; 95% CI, 5.66 to 19.39), and type 2 diabetes mellitus (T2DM; OR, 19.43; 95% CI, 12.75 to 29.81) in the NAFLD group but not in the non-NAFLD group. The cut-off points for the FFA levels increased in a stepwise manner in discriminating IR, prediabetes and T2DM (573, 697, and 715 µmol/L) in the NAFLD group but not in non-NAFLD individuals. CONCLUSION: A distinct dose-dependent relationship of FFA levels was found with IR, prediabetes and T2DM in NAFLD patients. Screening serum FFA levels in NAFLD patients would be valuable in preventing diabetes development.

Precise fibrosis staging with shear wave elastography in chronic hepatitis B depends on liver inflammation and steatosis.

BACKGROUND: Two-dimensional shear wave elastography (2D-SWE) is the latest generation of ultrasound elastography for the non-invasive assessment of liver fibrosis in chronic hepatitis B (CHB). We aimed to identify confounders of 2D-SWE in fibrosis grading. METHODS: A prospective cohort of 440 CHB patients (286 with liver biopsy and 154 with clinical decompensated cirrhosis) was consecutively enrolled from a clinical trial (registration number: ChiCTR-DCD-15006000) aimed at optimizing 2D-SWE assessments from 2015 to 2018. All patients underwent 2D-SWE examination, anthropometric measurement, and serum biomarker assessment. Steatosis was graded by the magnetic resonance imaging-derived proton density fat fraction (MRI-PDFF). RESULTS: Overall, the prevalence of incorrect fibrosis staging by 2D-SWE was 26.1% (n = 115), with 43.5% of patients under-staged and 56.5% over-staged. In multivariate analysis, the steatosis degree was an independent predictor of 2D-SWE discordance in the overall cohort, with moderate-severe steatosis for underestimation (odds ratio, [OR] = 4.3, 95% confidence interval [CI] 1.2-18.2, p = 0.049) and overestimation (OR = 8.2, 95% CI 2.9-23.5, p < 0.001), and mild steatosis for overestimation (OR = 3.7, 95% CI 1.5-9.0, p = 0.004). In patients with liver biopsy, both histological inflammation activity over 2 (OR = 5.0, 95% CI 2.0-25.3, p = 0.048) and moderate-severe steatosis (OR = 5.2, 95% CI 2.1-13.4, p < 0.001) were independent factors associated with discordance. For the risk of 2D-SWE mis-staging, a nomogram that integrated these confounders was established and the area under the receiver operating characteristic curve of the model was 0.861. CONCLUSIONS: Steatosis and inflammation activities were confounders for 2D-SWE. The combination of these confounders could predict mis-staging risks of CHB-related fibrosis with 2D-SWE and may be valuable to decision-making on liver biopsy for fibrosis staging.

The clinical overlap between functional dyspepsia and irritable bowel syndrome based on Rome III criteria.

BACKGROUND: Epidemiological studies suggest considerable overlap between functional dyspepsia (FD) and irritable bowel syndrome (IBS). To date, no surveys have been performed to investigate the clinical overlap between these two disorders using Rome III criteria. Our aim was to investigate the prevalence and risk factors for the overlap of FD and IBS based on Rome III criteria in a large clinical sample. METHODS: Consecutive patients at the general gastroenterology outpatient clinic were requested to complete a self-report questionnaire. FD and IBS were defined by Rome III criteria. RESULTS: Questionnaires were returned by 3014 patients (52.8% female, 89% response rate). FD-IBS overlap was observed in 5.0% of the patients, while 15.2% and 10.9% of the patients were classified as FD alone and IBS alone, respectively. Compared with non-IBS patients, the odds ratio of having FD among IBS patients was 2.09 (95% CI: 1.68-2.59). Patients with FD-IBS overlap had higher severity scores for the postprandial fullness symptom (2.35 +/- 1.49 vs. 1.72 +/- 1.59, P < 0.001) and overall FD symptom (6.65 +/- 2.88 vs. 5.82 +/- 2.76, P = 0.002) than those with FD alone. The only independent risk factor for FD-IBS overlap vs. FD alone was the presence of postprandial fullness symptom (OR 2.67, 95% CI: 1.34-5.31). CONCLUSION: Clinical overlap of FD and IBS according to Rome III criteria is very common. One risk factor for FD-IBS overlap is the presence of postprandial fullness symptom. This study provides clues for future pathophysiological studies of FD and IBS.

[A comparison between Rome III and Rome II criteria in diagnosing irritable bowel syndrome].

OBJECTIVE: To determine the degree of agreement of Rome III and Rome II criteria in diagnosing irritable bowel syndrome (IBS) and to compare the clinical difference between the patients diagnosed with these two criteria. METHODS: 3014 patients in the gastrointestinal outpatient department were enrolled consecutively and interviewed face to face with a standard questionnaire. RESULTS: (1) 480 patients were diagnosed as IBS with Rome III criteria. The overall detection rate was 15.9% (480/3014). The proportion of IBS subtypes was as follows: IBS with constipation 27.9% (134/480), IBS with diarrhea 32.7% (157/480), Mixed IBS 6.7% (32/480), Unsubtyped IBS 32.7% (157/480). No difference was observed between different sex and age groups; with Rome II criteria, 558 patients were diagnosed with a detection rate of 18.5% (558/3014). The proportion of IBS subtypes was as follows: constipation predominant IBS 33.2% (185/558), diarrhea predominant IBS 38.2% (213/558), others 28.7% (160/558). The detection rate was higher in female patients (P = 0.002), but there was no difference between different age groups. The detection rate of Rome III criteria was lower than that of Rome II criteria (P = 0.008). There was a good accordance between these two criteria in the diagnosis of IBS (P < 0.01). (2) Patients classified according to Rome III criteria complained more severe abdominal symptoms (P = 0.04) and abnormal bowel habit (P < 0.001) as well as a higher healthcare seeking rate in the last 3 months (35.6% vs 26.5%, P = 0.02) as compared with those classified according to Rome II criteria. (3) According to Rome III criteria, the severity of bowel habit was different among the four subtypes (C-IBS, M-IBS > D-IBS > U-IBS, P < 0.005) while no difference was observed on the abdominal symptoms and the healthcare seeking rates in the last 3 months. CONCLUSIONS: There is a good accordance between Rome II and Rome III criteria in diagnosing IBS. Compared to Rome II criteria, Rome III criteria has a lower detection rate. It is more practical in the clinical practice with clear definition of symptom frequency and easy way of subtyping IBS. The patients diagnosed with Rome III criteria had more severe symptoms and higher healthcare seeking rate, they are more suitable for clinical trial.

Association of keratin 8/18 variants with non-alcoholic fatty liver disease and insulin resistance in Chinese patients: A case-control study.

AIM: To test the hypothesis that K8/K18 variants predispose humans to non-alcoholic fatty liver disease (NAFLD) progression and its metabolic phenotypes. METHODS: We selected a total of 373 unrelated adult subjects from our Physical Examination Department, including 200 unrelated NAFLD patients and 173 controls of both genders and different ages. Diagnoses of NAFLD were established according to ultrasonic signs of fatty liver. All subjects were tested for population characteristics, lipid profile, liver tests, as well as glucose tests. Genomic DNA was obtained from peripheral blood with a DNeasy Tissue Kit. K8/K18 coding regions were analyzed, including 15 exons and exon-intron boundaries. RESULTS: Among 200 NAFLD patients, 10 (5%) heterozygous carriers of keratin variants were identified. There were 5 amino-acid-altering heterozygous variants and 6 non-coding heterozygous variants. One novel amino-acid-altering heterozygous variant (K18 N193S) and three novel non-coding variants were observed (K8 IVS5-9A→G, K8 IVS6+19G→A, K18 T195T). A total of 9 patients had a single variant and 1 patient had compound variants (K18 N193S+K8 IVS3-15C→G). Only one R341H variant was found in the control group (1 of 173, 0.58%). The frequency of keratin variants in NAFLD patients was significantly higher than that in the control group (5% vs 0.58%, P = 0.015). Notably, the keratin variants were significantly associated with insulin resistance (IR) in NAFLD patients (8.86% in NAFLD patients with IR vs 2.5% in NAFLD patients without IR, P = 0.043). CONCLUSION: K8/K18 variants are overrepresented in Chinese NAFLD patients and might accelerate liver fat storage through IR.

Apolipoprotein C3 (-455T>C) polymorphism confers susceptibility to nonalcoholic fatty liver disease in the Southern Han Chinese population.

AIM: To investigate the relationship between Apolipoprotein C3 (APOC3) (-455T>C) polymorphism and nonalcoholic fatty liver disease (NAFLD) in the Southern Chinese Han population. METHODS: In this prospective case-control study, we recruited 300 NAFLD patients and 300 healthy controls to a cohort representing Southern Chinese Han population at The First Affiliated Hospital, Sun Yat-sen University, from January to December 2012. Polymerase chain reaction-restriction fragment length polymorphism and DNA sequencing were used to genotype the APOC3 (-455T>C) variants. RESULTS: After adjusting for age, gender, and body-mass index, TC and CC genotypes were found to increase the susceptibility to NAFLD compared to the TT genotype, with adjusted odds ratios (ORs) of 1.77 (95%CI: 1.16-2.72) and 2.80 (95%CI: 1.64-4.79), respectively. Further stratification analysis indicated that carriers of the CC genotype was more susceptible to insulin resistance (IR) than those of the TT genotype, with an OR of 3.24 (95%CI: 1.52-6.92). The CC genotype also was associated with a significantly higher risk of hypertension, hypertriglyceridemia, and low levels of high-density lipoprotein cholesterol (HDL) (P < 0.05). No association was found between the APOC3 (-455T>C) polymorphism and obesity, impaired glucose tolerance, hyperuricemia, hypercholesterolemia, or high levels of low-density lipoprotein cholesterol (LDL) (P > 0.05). CONCLUSION: APOC3 (-455T>C) genetic variation is involved in the susceptibility to developing NAFLD, IR, hypertension, hypertriglyceridemia, and low HDL in the Southern Chinese Han population.

Prevalence and features of fatty liver detected by physical examination in Guangzhou.

AIM: To investigate the prevalence of fatty liver discovered upon physical examination of Chinese patients and determine the associated clinical characteristics. METHODS: A total of 3433 consecutive patients who received physical examinations at the Huangpu Division of the First Affiliated Hospital at Sun Yat-sen University in Guangzhou, China from June 2010 to December 2010 were retrospectively enrolled in the study. Results of biochemical tests, abdominal ultrasound, electrocardiography, and chest X-ray were collected. The diagnosis of fatty liver was made if a patient met any two of the three following ultrasonic criteria: (1) liver and kidney echo discrepancy and presence of an increased liver echogenicity (bright); (2) unclear intrahepatic duct structure; and (3) liver far field echo decay. RESULTS: The study population consisted of 2201 males and 1232 females, with a mean age of 37.4 ± 12.8 years. When all 3433 patients were considered, the overall prevalence of hyperlipidemia was 38.1%, of fatty liver was 26.0%, of increased alanine aminotransferase (ALT) and/or aspartate aminotransferase (AST) levels was 11.9%, of gallstone was 11.4%, of hyperglycemia was 7.3%, of hypertension was 7.1%, and of hyperuricemia was 6.2%. Of the 2605 patients who completed the abdominal ultrasonography exam, 677 (26.0%) were diagnosed with fatty liver and the prevalence was higher in males (32.5% vs females: 15.3%, P < 0.001). The overall prevalence of fatty liver increased with age, with the peak prevalence (39.5%) found in the 60 to 70-year-old age group. Among patients between the ages of 18 to 50-year-old, the prevalence of fatty liver was significantly higher in males (20.2% vs females: 8.7%, P < 0.001); the difference in prevalence between the two sexes in patients > 50-year-old did not reach statistical significance. Only 430 of the patients diagnosed with fatty liver had complete information; among those, increased ALT and/or AST levels were detected in only 30%, with all disturbances being mild or moderate. In these 430 patients, the overall prevalence of hypertriglyceridemia was 31.4%, of mixed type hyperlipidemia was 20.9%, of hypercholesterolemia was 12.3%, of hyperglycemia was 17.6%, of hypertension was 16.0%, of hyperuricemia was 15.3%, and of gallstone was 14.4%. Again, the prevalences of hypertriglyceridemia and hyperuricemia were higher in males (hypertriglyceridemia, 36.0% vs females: 12.0%, P < 0.05; hyperuricemia, 17.3% vs females: 7.2%, P < 0.05); in contrast, however, the prevalences of mixed type hyperlipidemia and hypercholesterolemia was higher in females (mixed type hyperlipidemia, 18.7% vs females: 30.1%, P < 0.05, hypercholesterolemia, 9.5% vs females: 24.1%, P < 0.05). Finally, comparison of the fatty liver group to the non-fatty liver group showed that prevalences of hyperlipidemia, hyperglycemia, hypertension, and hyperuricemia were higher in the former (all P < 0.01). CONCLUSION: A high prevalence of fatty liver is detected upon physical examination in Guangzhou, and the primary associated clinical findings are hyperlipidemia, hyperglycemia, hypertension, and hyperuricemia.