Deep Online Video Stabilization with Multi-Grid Warping Transformation Learning.

Video stabilization techniques are essential for most hand-held captured videos due to high-frequency shakes. Several 2D, 2.5D and 3D-based stabilization techniques have been presented previously, but to our knowledge, no solutions based on deep neural networks had been proposed to date. The main reason for this omission is shortage in training data as well as the challenge of modeling the problem using neural networks. In this paper, we present a video stabilization technique using a convolutional neural network. Previous works usually propose an offline algorithm that smoothes a holistic camera path based on feature matching. Instead, we focus on low-latency, real-time camera path smoothing, that does not explicitly represent the camera path, and does not use future frames. Our neural network model, called StabNet, learns a set of mesh-grid transformations progressively for each input frame from the previous set of stabalized camera frames, and creates stable corresponding latent camera paths implicitly. To train the network, we collect a dataset of synchronized steady and unsteady video pairs via a specially designed hand-held hardware. Experimental results show that our proposed online method performs comparatively to traditional offline video stabilization methods without using future frames, while running about 10× faster. More importantly, our proposed StabNet is able to handle low-quality videos such as night-scene videos, watermarked videos, blurry videos and noisy videos, where existing methods fail in feature extraction or matching.

BiggerSelfie: Selfie Video Expansion With Hand-Held Camera.

Selfie photography from the hand-held camera is becoming a popular media type. Although being convenient and flexible, it suffers from low camera motion stability, small field of view, and limited background content. These limitations can annoy users, especially, when touring a place of interest and taking selfie videos. In this paper, we present a novel method to create what we call a BiggerSelfie that deals with these shortcomings. Using a video of the environment that has partial content overlap with the selfie video, we stitch plausible frames selected from the environment video to the original selfie frames and stabilize the composed video content with a portrait-preserving constraint. Using the proposed method, one can easily obtain a stable selfie video with expanded background content by merely capturing some background shots. We show various results and several evaluations to demonstrate the applicability of our method.

High-dose esomeprazole is required for intraesophageal acid control in gastroesophageal reflux disease patients with hiatus hernia.

BACKGROUND AND AIM: The aim of this study was to assess whether the efficacy of proton pump inhibitors (PPI) therapy at a standard dose in esophageal acid control is affected by the presence of hiatus hernia in Chinese gastroesophageal reflux disease patients, and whether a higher dose of PPI is required for acid control. METHODS: Consecutive gastroesophageal reflux disease patients who had typical reflux symptoms and abnormal baseline 24-h esophageal pH and underwent upper endoscopy were enrolled to receive esomeprazole at 40 mg once daily for 4 weeks. Patients underwent the dual-channel 24-h pH test at the end of 4-week therapy. If the 24-h esophageal pH was still abnormal at the end of 4-week therapy, then esomeprazole at 40 mg twice daily was given for another 4 weeks after a washout interval of 1 week, and a 24-h pH test was repeated at the end of the therapy. RESULTS: Overall, 76 patients were included, 13 with hiatus hernia. Of the 76 patients treated with a 40 mg of esomeprazole daily, esophageal acid exposure was normalized in 64 (84.2%). Normalization of acid exposure was achieved by standard PPI therapy in 53.2% (7/13) of patients with hiatus hernia and 90.5% (57/63) of those without (P = 0.004). A double dose of esomeprazole was successful in normalizing the esophageal pH in all 12 non-responders to the standard dose of esomeprazole, including the six patients with hiatus hernia and six patients without. CONCLUSIONS: The standard-dose of esomeprazole fails to normalize the esophageal pH in almost 50% of patients with hiatus hernia, in whom the "double-dose" esomeprazole therapy is required.

Comparison of patients of chronic laryngitis with and without troublesome reflux symptoms.

BACKGROUND AND AIM: Little is known about the difference between patients of chronic laryngitis with and without troublesome reflux symptoms. The aim of this study was to compare the clinical characteristics and response to acid suppression between patients of chronic laryngitis with and without troublesome reflux symptoms. METHODS: Consecutive patients with chronic laryngitis were enrolled. The frequency and severity of reflux and laryngeal symptoms were scored. All the patients underwent laryngoscopy, esophagogastroduodenoscopy and 24-h multichannel intraluminal impedance and pH monitoring before receiving rabeprazole 10 mg b.i.d. for 3 months. Mild typical reflux symptoms (heartburn or regurgitation) occurring ≥ 2 days/week or moderate/severe symptoms occurring ≥ 1 day/week were defined as troublesome reflux symptoms. RESULTS: Compared to patients without troublesome reflux symptoms, those with troublesome reflux symptoms were older and had more episodes of acid and liquid gastroesophageal reflux (GER) and acid and weakly acidic laryngopharyngeal reflux (LPR). They also had higher percentages of both bolus exposure time and acid exposure time of GER and LPR. Patients with troublesome reflux symptoms responded to acid suppression more often at 12 weeks (67.3% vs 20.9%, P < 0.001) and more rapidly (40.8% vs 14.0%, 3 weeks after the start of acid suppression; P = 0.004) compared to those without. CONCLUSION: Difference in reflux profile of GER and LPR between patients with and without troublesome reflux symptoms could partly explain the discrepancy of response to acid suppression among patients with chronic laryngitis. Acid suppression therapy may provide limited therapeutic benefits to patients of chronic laryngitis without troublesome reflux symptoms.

Gastroesophageal and laryngopharyngeal reflux detected by 24-hour combined impedance and pH monitoring in healthy Chinese volunteers.

OBJECTIVE: To define the normative data of gastroesophageal reflux (GER) and laryngopharyngeal reflux (LPR) in healthy Chinese volunteers. METHODS: Healthy volunteers without gastrointestinal and throat disease and symptoms were enrolled and underwent 24-hour impedance-pH monitoring. RESULTS: A total of 37 healthy volunteers completed all the tests (female 54.1%; age 38.0±15.2 years). Of 1862 GER, 49.6% were mixed, 40.9% were liquid and 9.5% were gas. More than half (54.9%) were acid. The median and 95th percentile numbers of GER were 52 and 71, respectively. The number of LPR was eight, of which two were mixed and the others were liquid. Only one episode of reflux was classified as acid LPR. The median and 95th percentile numbers of LPR were 0 and 2, respectively. More liquid reflux were associated with LPR (P=0.02). Age had no influence on GER and LPR reflux profiles. CONCLUSIONS: Few LPR occurred in healthy Chinese volunteers and they were rarely acid when they reached the larynx. LPR was mostly associated with liquid reflux.

Relationship between symptom response and esophageal pH level on standard dose of esomeprazole treatment for gastroesophageal reflux disease.

BACKGROUND: The relationship between symptom elimination and normalization of esophageal acid level of gastroesophageal reflux disease (GERD) on proton-pump inhibitor (PPI) therapy has been questioned. This study aimed to evaluate the relationship between symptom response and gastro-esophageal acidity control in Chinese patients with GERD on esomeprazole therapy, and to assess the role of 24-hour esophageal pH-metry after therapy in GERD patients. METHODS: GERD patients with typical reflux symptoms were enrolled and received esomeprazole 40 mg once daily for 4 weeks. Patients with positive baseline 24-hour esophageal pH-metry were divided into two groups depending on an additional dual-channel 24-hour pH-metry after treatment. The pH- group achieved normalization of esophageal pH level whereas the pH+ group did not. RESULTS: Of the 80 patients studied, 76 had abnormal baseline esophageal pH levels. Of these, 90% (52/58) of symptom-free patients and 67% (12/18) of symptom-persistent patients achieved esophageal pH normalization after therapy (P = 0.030). The mean post-therapy gastric nocturnal percent time of pH < 4.0 was significantly higher in pH+ group than that in pH- group (P < 0.001) after therapy. The multivariate regression analysis identified hiatus hernia (P < 0.001) and persistent reflux symptom (P = 0.004) were two independent factors predicting the low post-therapy esophageal pH level. CONCLUSIONS: Symptom elimination is not always accompanied by esophageal pH normalization, and vice verse. Esophageal pH-metry is recommended for GERD patients with hiatus hernia or with persistent reflux symptoms after PPI therapy.

Prevalence and symptom pattern of pathologic esophageal acid reflux in patients with functional dyspepsia based on the Rome III criteria.

OBJECTIVES: To determine the prevalence and symptom pattern of pathologic esophageal acid reflux (PEAR) in patients with functional dyspepsia (FD) using the Rome III criteria, and to explore the value of a proton pump inhibitor (PPI) test in distinguishing the patients with and those without PEAR among FD patients. METHODS: Consecutive FD patients who fulfilled the Rome III criteria without predominant typical reflux symptoms (i.e., heartburn or regurgitation) were enrolled. All patients underwent upper endoscopy and an ambulatory 24-h pH monitoring. PEAR was defined as the percentage total time for which a pH value <4 was >4.2% in the distal esophagus. Then, patients were treated with rabeprazole 10 mg twice daily for 28 days. The symptom scores were measured by the frequency score multiplied by the severity scores of the predominant symptom before and at the end of the treatment, and the "PPI test" was defined as positive if the overall scores of the predominant dyspeptic symptom in the fourth week decreased by >50% compared with those of the baseline. RESULTS: One hundred eighty-six FD patients were enrolled, with predominant symptoms of epigastric pain (n=68), epigastric burning (n=47), bothersome postprandial fullness (n=54), and early satiation (n=17). The prevalence of PEAR was 31.7%, with the highest percentage (48.9%) in patients with epigastric burning as their predominant symptom. The prevalence of PEAR in patients with postprandial distress syndrome (PDS) and epigastric pain syndrome (EPS) were 36.6% (26/71) and 28.7% (33/115), respectively. Overall, 63.4% were positive for the "PPI test"; the rates were 51.5, 85.0, 66.7, and 41.1% in patients with epigastric pain, epigastric burning, bothersome postprandial fullness, and early satiation as their predominant symptoms, respectively (χ(2)=17.59, P=0.001). The positive rates were 65.5 and 60.6% in patients with PDS and EPS, respectively (χ(2)=0.41, P=0.522). The sensitivity and specificity of the "PPI test" in distinguishing FD patients with PEAR was 83.1 and 45.7%, respectively. CONCLUSIONS: PEAR is present in almost one third of FD patients; the prevalence is ∼50% in those with epigastric burning. The "PPI test" has a limited value in distinguishing the FD patients with and those without PEAR.

Prompt upper endoscopy is an appropriate initial management in uninvestigated chinese patients with typical reflux symptoms.

OBJECTIVES: We sought to investigate the prevalence of clinically significant endoscopic findings (CSEFs) in Chinese patients presenting with uninvestigated typical reflux symptoms in the absence of alarm symptoms, and to evaluate whether prompt endoscopy is an appropriate initial management in these patients. METHODS: Consecutive patients presenting with uninvestigated typical reflux symptoms (i.e., heartburn or acid regurgitation) as chief complaints were recruited for symptom evaluation and upper endoscopy, followed by a 2-week proton pump inhibitor (PPI) therapy. RESULTS: Of 469 patients recruited, CSEFs were observed in 180 (38.4%): 154 (32.8%) with erosive esophagitis (EE), 18 (3.8%) with Barrett's esophagus (BE), 24 (5.1%) with peptic ulcer disease (PUD), and 4 (0.9%) with carcinomas (1 esophageal carcinoma and 3 gastric adenocarcinomas). Multivariate analysis identified that an age >50 years (odds ratio (OR)=1.94, P=0.008), male gender (OR=4.11, P<0.001), being overweight or obese (OR=2.99, P<0.001), and alcohol use (OR=9.96, P<0.001) were independent risk factors for EE; an age >50 years (OR=4.61, P=0.003) and alcohol use (OR=5.50, P=0.003) were independent risk factors for BE; and Helicobacter pylori infection (OR=8.52, P<0.001) and alcohol use (OR=4.08, P=0.004) were independent risk factors for PUD. Symptom evaluation and response to PPI treatment were not correlated with EE, BE, and PUD in these patients. CONCLUSIONS: CSEFs other than gastroesophageal reflux disease are present in a considerable proportion of Chinese patients with uninvestigated typical reflux symptoms but without alarm features. Symptom evaluation is of limited practical value, and thus prompt endoscopy seems to be an appropriate initial management option in these patients.

[Diagnosis of gastroesophageal reflux disease-related noncardiac chest pain].

OBJECTIVE: To assess the values of endoscopy, 24-hour esophageal pH monitoring, symptom index, 24-hour esophageal bile monitoring, and proton pump inhibitor (PPI) trail in the diagnosis of gastroesophageal reflux disease (GERD)-related noncardiac chest pain (NCCP). METHODS: 27 NCCP patients, selected from a consecutive sample of 255 patients who had received coronary angiography owing to chest pain, underwent upper endoscopy, and 24-hour esophageal pH and bile monitoring, then they took esomeprazole 20 mg orally twice daily for 14 days. Symptom score (SI) was used to measure the severity and frequency of chest pain before and after treatment. The patients also underwent an initial questionnaire survey including the general personal data and symptom characters, reflux diagnostic questionnaire (RDQ), self-rating anxiety scale (SAS) survey, and self-rating depression scale (SDS) survey. RESULTS: If only based on erosive esophagitis and/or abnormal 24-hour esophageal pH monitoring, 9 of the 27 patients (33%) were classified as GERD-related NCCP. Otherwise, 74% (20 of the 27 patients were classified as GERD-related NCCP based on erosive esophagitis and/or abnormal 24-hour esophageal pH monitoring, or positive SI or abnormal bile monitoring. 15 of the 20 GERD patients (75%) had a significant symptom improvement on PPI compared with 14% (1/7) of non-GERD patients (P = 0. 009). The calculated sensitivity and specificity of the PPI trail was 75% and 86% respectively. The typical reflux symptoms presented more frequently among GERD patients than non-GERD patients [65% (13/20) vs 29% (2/7), P = 0.185]. There was no significant RDQ score difference between the 2 groups (10 +/- 4 vs 8 +/- 4, P = 0.411). The sensitivity and specificity of RDQ to GERD-related NCCP were 35% and 86% respectively. The SAS scores of GERD and non-GERD-related NCCP patients were both significantly higher than that of the normal controls (P = 0.008 and P = 0.011), while the SDS scores of the GERD and non-GERD-related NCCP patients were both similar to that of the normal controls (P = 0.090 and P = 0.113). CONCLUSION: Combination of endoscopy, 24-hour esophageal pH monitoring, SI, and 24-hour esophageal bile monitoring helped to improve the diagnosis of GERD-related NCCP. PPI trail is the first clinical diagnostic tool in the evaluation of GERD-related NCCP.

Stratification and symptom characteristics of non-erosive reflux disease based on acid and duodenogastroesophageal reflux.

BACKGROUND AND AIM: The subgroups and symptom characteristics of non-erosive reflux disease (NERD) based on acid and duodenogastroesophageal reflux may be different in Chinese patients to Western patients. This study aimed to explore the stratification and symptom characteristics of patients with NERD. METHODS: Patients with typical heartburn and/or acid regurgitation symptoms were enrolled. Each patient filled out a questionnaire. An upper gastrointestinal endoscopy was performed for each patient followed by simultaneous ambulatory 24-h esophageal pH and Bilitec (bilirubin) monitoring. A symptom index (SI) of > or =50% was considered to be positive. RESULTS: Eighty-two consecutive NERD patients were evaluated. Abnormal (NERD pH+) and normal (NERD pH-) 24-h pH tests were found in 24 (29.3%) and 58 (70.7%) patients, respectively. Among 42 NERD pH- patients who reported heartburn symptoms during monitoring, SI was positive in 19 (45.2%) patients (NE-SI+) and negative in 23 (54.8%) patients (NE-SI-). Pathological duodenogastroesophageal reflux (DGER) was found in 43 (52.4%) patients. No significant differences were noted regarding the prevalence of other upper gastrointestinal symptoms, except for acid regurgitation in NERD pH+ and NERD pH- groups. Additionally, no significant differences were seen in the prevalence of other symptoms, except for chest pain, in groups with pathological and normal DGER. CONCLUSION: The proportion of NERD patients with pathological acid reflux was somewhat lower than that reported in Western countries. The role of DGER in NERD may be important. It is difficult to differentiate whether NERD patients have pathological acid or bile reflux according to symptoms.

[Changes of intercellular spaces of esophageal epithelium in patients with gastroesophageal reflux disease before and after esomeprazole treatment].

OBJECTIVE: To observe the changes of dilated intercellular space (DIS) of esophageal epithelium in different subtypes of gastroesophageal reflux disease (GERD) before and after esomeprazole treatment. METHODS: Nineteen patients with GERD, 6 with erosive esophagitis (EE), 7 with NERD with abnormal acid exposure (NERD pH+), and 6 with NERD with normal acid exposure (NERD pH-), were treated with esomeprazole 40 mg daily for two months. During endoscopy before and after treatment, six biopsy specimens were taken from the apparently normal mucosa 3 - 5 cm above the dentate line in the esophagus to undergo HE staining and histological examination and transmission electron microscopy (TEM). Fifteen normal subjects were used as controls. RESULTS: Two months after the treatment 5 cases (83.33%) in the EE group, 6 cases (85.57%) in the NERD pH+ group, and 5 cases (83.33%) in the NERD pH- group showed a complete recovery of DIS and complete disappearance of the symptom of heartburn. Two of the left three patients who showed incomplete recovery of DIS still had heartburn. CONCLUSION: A complete recovery of DIS 2 months after treatment of esomeprazole is always accompanied by resolution of heartburn symptom. No significant difference in complete recovery of DIS can be found among different GERD subgroups.

[A study of diagnosis of primary biliary reflux gastritis].

OBJECTIVE: To study the diadynamic method of primary bile reflux gastritis. METHODS: Endoscopy, histologic examination, symptom analysis and 24-hour gastritic bilirubin monitoring with Bilitec 2000 were performed in 20 healthy subjects (HS) and 42 patients with symptoms of abdominal pain, abdominal distention, nausea and bile vomiting. RESULTS: The total fraction time of bile reflux was higher in 42 patents than in 20 HS [(17.68 +/- 17.89)% vs. (2.92 +/- 2.39)%, P < 0.01]. 23 (55%) of 42 patients showed pathologic duodenogastric reflux (DGR). The incidence of pathologic DGR detected by Bilitec2000 in bile dyed mucous lake, antral mucosal erosion, bile dyed mucous lake accompanied with erosion and bile dyed mucous lake accompanied with middle-hyperemia of antral mucosae found in endoscopy were 86%, 88%, 8/8 and 85% respectively. The diagnosis of bile reflux gastritis (BRG) could be established if the patients with bile dyed mucous lake accompanied with erosion and/or middle-hyperemia of antral mucosae were found to be pathologic DGR. With this criterion, 11 patients were BRG in 42 patients in this study. The active inflammation of antral mucosae in the patients with BRG that without Helicobacter pylori (Hp) infection was more severe than that in the patients with physiologic DGR that without Hp infection significantly (P < 0.05). The symptoms of abdominal distention, nausea and bile vomiting in the patients with BRG were more severe significantly than those in the patients with physiologic DGR respectively (P < 0.05). CONCLUSIONS: Final diagnosis of BRG can be made for the integrity stomach if the patients with bile dyed mucous lake accompanied with erosion and/or middle-hyperemia of antral mucosae were found to be pathologic DGR by Bilitec 2000.