Endoscopic Detection Research Articles

Approach to Patients after Successful Eradication of Helicobacter pylori

<i>Helicobacter pylori</i> (<i>H. pylori</i>) is a well-known gastrointestinal microorganism that causes chronic gastritis and peptic ulcers, which may evolve into gastric cancer. Previous studies have shown that <i>H. pylori</i> eradication inhibits the development of primary and metachronous gastric cancer. Therefore, the Kyoto global consensus recommends eradication therapy for both symptomatic and asymptomatic <i>H. pylori</i>-infected patients. Gastric cancer is detected even in patients after successful <i>H. pylori</i> eradication. Recent studies have reported the role of endoscopy in detection of primary gastric cancers after <i>H. pylori</i> eradication. Development of gastric cancer may be observed several months or even >10 years after successful <i>H. pylori</i> eradication. Therefore, identification of high-risk patients in whom extensive surveillance may prove beneficial represents a clinical dilemma. In this review, the characteristics of gastric cancer patients who have undergone successful <i>H. pylori</i>-eradication therapy are summarized.

Conspectus on nanodiagnostics as an incipient platform for detection of oral potentially malignant disorders and oral squamous cell carcinoma.

Conspectus on nanodiagnostics as an incipient platform for detection of oral potentially malignant disorders and oral squamous cell carcinoma.

Appendiceal lesions in serrated polyposis patients are easily overlooked but only seldomly lead to colorectal cancer.

Serrated polyposis syndrome (SPS) is the most prevalent colonic polyposis syndrome and is associated with an increased colorectal cancer risk. A recent study in resected appendices of SPS patients reported that 6/23 (26.1 %) of identified serrated polyps had histological dysplasia. We evaluated the prevalence and clinical relevance of appendiceal lesions in a large SPS cohort. Prospective data from 2007 to 2020 for a cohort of 199 SPS patients were analyzed. Data were retrieved from endoscopy and pathology reports. Patients who underwent (pre)clearance colonoscopies, surveillance colonoscopies, or colorectal surgery including the appendix were separately evaluated for the presence of appendiceal lesions. The primary outcome was the prevalence of adenocarcinomas and serrated polyps/adenomas with advanced histology in the surgery group. 171 patients were included, of whom 110 received endoscopic surveillance and 34 underwent surgery. Appendiceal lesion prevalence in the surgery group was 14 /34 (41.2 %, 95 %CI 24.7 %-59.3 %); none were advanced on histology. Detection rates in the (pre)clearance group were 1 /171 (0.6 %, 95 %CI 0.01 %-3.2 %) for advanced and 3 /171 (1.8 %, 95 %CI 0.04 %-5.0 %) for nonadvanced appendiceal lesions, all of which were sessile serrated lesions. During 522 patient-years of surveillance, no advanced appendiceal lesions were detected at endoscopy, and in 1 /110 patients (0.9 %, 95 %CI 0.02 %-5.0 %) was a nonadvanced lesion detected. Appendiceal lesions are common in SPS patients. The discrepancy between the endoscopic detection rate of appendiceal lesions and the reported prevalence in surgically resected appendices suggests a substantial miss-rate of appendiceal lesions during colonoscopy. Advanced appendiceal lesions are however rare and no appendiceal adenocarcinomas occurred, implying limited clinical relevance of these lesions.

MvHOTA: A multi-view higher order tracking accuracy metric to measure temporal and spatial associations in multi-point tracking

ABSTRACT Multi-point tracking is a challenging task that involves detecting points in the scene and tracking them across time. Here, metrics from Multi-object tracking (MOT) methods are shown to perform better than frame-based F-measures. The recently proposed HOTA metric, used for benchmarks such as the KITTI dataset, better evaluates the performance over metrics like MOTA, DetA, and IDF1. While HOTA takes into account temporal associations, it does not provide a tailored means to analyse the spatial associations of a dataset in a multi-camera setup. Moreover, there are differences in evaluating the detection task for points vs. objects (point distances vs. bounding box overlap). Therefore, we propose a multi-view higher-order tracking metric mvHOTA ,to determine the accuracy of multi-point (multi-instance and multi-class) tracking methods while taking into account temporal and spatial associations. We demonstrate its use in evaluating the tracking performance on an endoscopic point detection dataset from a previously organised surgical data science challenge. Furthermore, we compare with other adjusted MOT metrics for this use-case, discuss the properties of mvHOTA, and show how the proposed multi-view Association and the Occlusion index (OI) facilitate analysis of methods with respect to handling of occlusions. The code is available at https://github.com/Cardio-AI/mvhota.

Intramural dissection of the esophagus is a rare complication of eosinophilic esophagitis in children

Eosinophilic esophagitis (EoE) is characterized by endoscopic signs and detection of more than 15 eosinophils per high-power field of view in biopsy specimens of the esophageal mucosa. Due to deep chronic eosinophilic inflammation, it can cause complications: esophageal stenosis, rarely perforation. Intramural esophageal dissection (IED) is a previously undescribed complication of EoE in children. We present a case of IED in a 15-year-old boy, which clinically manifested itself as sudden epigastric pain radiating to the back and nausea against the background of acute respiratory viral infection. According to endoscopy, CT and X-ray contrast study, a double-barreled esophagus with two cavities was detected - the esophagus itself and a blind "pocket", histologically in which granulation tissue and active inflammation with a large number of neutrophils were found. The initial histological examination did not reveal significant eosinophilic infiltration in the esophagus. The boy received a course of esomeprazole and was re-examined 6 months later. The radiographic picture was the same, and endoscopic examination revealed characteristic signs of EoE: longitudinal grooves and erosions, transverse rings, and narrowing of the esophageal lumen. In the middle third, the picture of a double-barreled esophagus was still preserved, but without signs of acute detachment, but with a dense septum between the two cavities of the esophagus. Histologically, the infiltrate contained more than 50 eosinophils in the field of view. The diagnosis was eosinophilic esophagitis complicated by RTI. A strict hypoallergenic diet was prescribed, excluding obligate allergens, budesonide 1 mg 2 times a day in the form of a gel. Endoscopic dissection of the esophageal septum was performed, its patency was restored. After 2 months, the boy was examined again while continuing drug therapy and diet. He had no complaints, radiologically the patency of the esophagus was not impaired. Endoscopically, the esophagus was freely passable, there was only a small fold at the site of the operation, longitudinal grooves and small erosions in the lower third remained. Histologically, the number of eosinophils decreased significantly. Therapy was continued. This clinical case demonstrates the possibility of developing RDI as a complication of EoE. The course of EoE can be asymptomatic, and RDI is the first manifestation of the disease. The endoscopic picture of a double-barreled esophagus is a consequence of dissection. The combination of a hypoallergenic diet, topical steroids and endoscopic septectomy is effective in the treatment of EoE with RDI.

Effects of COVID-19 Pandemic and Post-Vaccination Period on Gastroenterology Practice in Turkey.

The sudden and intense burden due to the novel coronavirus (coronavirus disease 2019) pandemic has changed the priority of many health services. The highly contagious new variants and vaccination services have greatly impacted the daily practice of gastroenterologists. In the present study, we tried to evaluate the change in the daily practice of Turkish gastroenterologists in the coronavirus disease 2019 pandemic and the post-vaccination periods. A questionnaire was prepared using Google forms and sent online to 214 gastroenterologists registered with the Turkish Gastroenterology Association. Approximately half of the gastroenterologists moved their endoscopy unit or gastroenterology clinic to another location in the hospital during the pandemic and about one-third of the endoscopy units were completely closed. About 56% of the respondents reported the interruption of endoscopic procedures and inpatient services. In total, 85% of gastroenterologists thought that follow-up and treatment of chronic patients were disrupted due to patients obtaining their medicine directly from pharmacies. After the first case in Turkey, 74% of gastroenterologists worked in coronavirus disease 2019 patient care services, 28% of gastroenterologists were infected with coronavirus disease 2019, and 3% believed they had a cross-infected patient(s). A total of 78% of gastroenterologists reported that patient management became difficult due to the change in the priority of other departments, and 60% of gastroenterologists confirmed that they experienced a decrease in income. In the post-vaccination period, 70% of gastroenterologists returned to their pre-pandemic working schedule and 31% noticed an increase in endoscopic cancer detection. Prolongation of the pandemic has seriously damaged the practice of gastroenterology and multidisciplinary patient management.

68Ga]Ga-FAPI-04 PET/CT on assessing Crohn's disease intestinal lesions.

Fibrosis and inflammation are major pathological changes of Crohn's disease (CD). Early detection and accurate severity evaluation of CD are critical for patient's prognosis. Endoscopy is widely used to evaluate CD progression. Herein, we evaluated the efficacy of [68Ga]Ga-FAPI-04 PET/CT to identify lesions and assess the progression of CD. All CD patients received computed tomography enterography (CTE), [68Ga]Ga-FAPI-04 PET/CT examination, and ileocolonoscopy within 1week. Two independent gastroenterologists computed the Crohn's disease activity index (CDAI) of all patients. Two radiology physicians assessed the CTE images separately, and the CTE scores were calculated. Lastly, two nuclear medicine physicians independently examined the [68Ga]Ga-FAPI-04 PET/CT images. Once the FAPI uptake of the intestinal segment was equal or higher relative to the liver (considered FAPI-positive), the target-to-background ratio (TBR) and global FAPI PET/CT score were computed, representing the independent intestinal activity and activity of all intestinal segments, respectively. Levels of fecal calprotectin (FCP) and C-reactive protein (CRP) were determined before the endoscopy. The Crohn's disease endoscopy index of severity (CDEIS) and the simple endoscopic score for Crohn's disease (SES-CD) were calculated during the endoscopy. Finally, all data were obtained and analyzed. There were 74 intestinal segments in 16 patients were assessed. [68Ga]Ga-FAPI-04 PET/CT identified 42 of 45 endoscopically lesioned segments (endoscopic lesions detection sensitivity: 93.3%), while CTE identified 39 of them (endoscopic lesions detection sensitivity: 86.7%). According to the receiver operating characteristic (ROC) analysis, [68Ga]Ga-FAPI-04 PET/CT showed better performance in the detection of endoscopic lesions compared with CTE (P < 0.05). The TBR was significantly associated with the CTE score (r = 0.81; (95% CI): 0.736-0.869; P < 0.0001) and SES-CD values (r = 0.86; (95% CI): 0.776-0.908; P < 0.0001). In addition, the global FAPI PET/CT score was significantly correlated with FCP (r = 0.52; 95% CI, 0.02-0.81; P = 0.039), CRP (r = 0.60; 95% CI, 0.13-0.85; P = 0.014), CDEIS (r = 0.55; 95% CI, 0.06-0.83; P = 0.028), and CDAI (r = 0.81; 95% CI, 0.50-0.93;P < 0.0001). In summary, [68Ga]Ga-FAPI-04 PET/CT correlated well with endoscopic, CTE, clinical, and biomarkers of CD. It was also highly sensitive in the detection of different classes of lesions in all intestinal segments, and unlike other examinations, this technique required no patient fasting or bowel preparation. Therefore, [68Ga]Ga-FAPI-04 PET/CT may be a promising method for assessing the activity of CD.

Simple medical rigid endoscopic relative percent distortion detection system

Simple medical rigid endoscopic relative percent distortion detection system

Inside Back Cover, Volume 2, Number 1, January 2023

Inside Back Cover: The cover image is based on the Research Article Slippery quartz surfaces for anti-fouling optical windows by Li et al. An antifouling transparent window is fabricated on a silica glass surface by infusing lubricant into its porous microstructure prepared by femtosecond laser ablation. The as-prepared surface exhibits great repellence to various liquids, good antifogging ability and stability after immersion into corrosive solutions. Such remarkable characteristics can protect the lens from contamination so as to achieve uninterrupted inspection during endoscopic detection procedure. (DOI: 10.1002/dro2.41)

Morphology, Histopathology, and Anatomical Distribution of Sporadic Colorectal Polyps in Chinese Patients.

There are limited data regarding the morphology, histopathology, and anatomical distribution of sporadic colorectal polyps in Chinese patients. We evaluated these characteristics of sporadic polyps to guide the endoscopic detection and excision of colorectal polyps. This was a retrospective single-center observational study involving 7381 patients with sporadic colorectal polyps. All polyps were removed endoscopically. The morphology and histopathology of polyps were evaluated according to the Paris classification and the World Health Organization classification, respectively. A total of 22,174 polyps removed endoscopically from 7322 patients were included. In the sigmoid colon, 24.70% of colorectal polyps occurred, followed by the transverse colon (18.58%) in frequency. 0-Is type polyps accounted for 60.60% of all sporadic colorectal polyps. Polyps with 0-Ip, 0-Isp, and 0-IIa types were frequently found in the sigmoid colon, but laterally spreading lesions usually occurred in the ascending colon (24.61%) and rectum (20.51%). Irrespective of the Paris classification and anatomical location, as the polyps enlarge, the proportion of adenomatous polyps gradually increases while the proportion of serrated lesions decreases. Polyps with size ≥1 cm located in the left-sided colon were more likely to have villous/tubulovillous or high-grade dysplasia histology than those located in the right-sided colon, and about 1% of them were demonstrated with adenocarcinoma. Sigmoid colon should be detected adequately during colonoscopy, and polyps with size ≥1 cm should be treated carefully, especially in the left-sided colon.

Ntestinal Metaplasia in Barrett’s Oesophagus: A Clinicopathological Study

Introduction: The incidence of Oesophageal Adenocarcinoma (EAC) has increased at a faster rate than any other cancer in the developed nations. Despite advances in treatment, five year survival rate for EAC is &lt;15%. Till date, Barrett’s Oesophagus (BE) is the only known precursor of EAC increasing its risk by greater than 30 to 60 fold. Most important risk factor for development of dysplasia and EAC is specialised Intestinal Metaplasia (IM) in BE. Aim: To find the association between clinical, endoscopic and histopathological features and presence of IM in patients with endoscopically suspected BE. Materials and Methods: This was an institution based descriptive study with a cross-sectional design conducted in the Departments of Pathology and Surgery, in a tertiary care centre of North Bengal for four years (2017-2021), among patients attending surgery and medicine Outpatient Department (OPDs) or Inpatient Department (IPDs), suspected on clinical basis and subsequent endoscopic detection of BE utilising Prague criteria. Periodic Acid-Schiff (PAS) and Alcian Blue (pH 2.5) stains were used to detect complete or incomplete IM and results were analysed using appropriate statistical software. Results: Among 120 cases included in the study, 72 (60%) had Short Segment Barrett’s Oesophagus (SSBE) and 48 (40%) Long Segment Barrett’s Oesophagus (LSBE). Hiatal hernia was significantly more frequent in LSBE patients (32 out of 48) compared to patients with SSBE (24 out of 72). The associations of tobacco and alcohol abuse with microscopically proven BE were statistically significant with p-values of 0.005 and 0.004, respectively. The association of IM with the increasing length of Columnar Lined Oesophagus (CLE) was statistically significant (p-value=0.004). Conclusion: Tobacco and alcohol abuse, presence of hiatal hernia (particularly in LSBE patients) were significantly associated with BE. Increasing length of CLE is more commonly associated with IM. Incomplete IM was observed more commonly in LSBE cases whereas complete IM was detected more frequently in cases diagnosed as SSBE by endoscopy

Repurposed Drugs in Gastric Cancer.

Gastric cancer (GC) is one of the major causes of death worldwide, ranking as the fifth most incident cancer in 2020 and the fourth leading cause of cancer mortality. The majority of GC patients are in an advanced stage at the time of diagnosis, presenting a poor prognosis and outcome. Current GC treatment approaches involve endoscopic detection, gastrectomy and chemotherapy or chemoradiotherapy in an adjuvant or neoadjuvant setting. Drug development approaches demand extreme effort to identify molecular mechanisms of action of new drug candidates. Drug repurposing is based on the research of new therapeutic indications of drugs approved for other pathologies. In this review, we explore GC and the different drugs repurposed for this disease.

Performance and comparison of artificial intelligence and human experts in the detection and classification of colonic polyps

ObjectiveThe main aim of this study was to analyze the performance of different artificial intelligence (AI) models in endoscopic colonic polyp detection and classification and compare them with doctors with different experience.MethodsWe searched the studies on Colonoscopy, Colonic Polyps, Artificial Intelligence, Machine Learning, and Deep Learning published before May 2020 in PubMed, EMBASE, Cochrane, and the citation index of the conference proceedings. The quality of studies was assessed using the QUADAS-2 table of diagnostic test quality evaluation criteria. The random-effects model was calculated using Meta-DISC 1.4 and RevMan 5.3.ResultsA total of 16 studies were included for meta-analysis. Only one study (1/16) presented externally validated results. The area under the curve (AUC) of AI group, expert group and non-expert group for detection and classification of colonic polyps were 0.940, 0.918, and 0.871, respectively. AI group had slightly lower pooled specificity than the expert group (79% vs. 86%, P < 0.05), but the pooled sensitivity was higher than the expert group (88% vs. 80%, P < 0.05). While the non-experts had less pooled specificity in polyp recognition than the experts (81% vs. 86%, P < 0.05), and higher pooled sensitivity than the experts (85% vs. 80%, P < 0.05).ConclusionThe performance of AI in polyp detection and classification is similar to that of human experts, with high sensitivity and moderate specificity. Different tasks may have an impact on the performance of deep learning models and human experts, especially in terms of sensitivity and specificity.

Application of artificial intelligence in the endoscopic diagnosis of early gastric cancer, atrophic gastritis, and Helicobacter pylori infection.

Gastric cancer (GC) is one of the most serious health problems worldwide. Chronic atrophic gastritis (CAG) is most commonly caused by Helicobacter pylori (H. pylori) infection. Currently, endoscopic detection of early gastric cancer (EGC) and CAG remains challenging for endoscopists, and the diagnostic accuracy of H. pylori infection by endoscopy is approximately 70%. Artificial intelligence (AI) can assist endoscopic diagnosis including detection, prediction of depth of invasion, boundary delineation, and anatomical location of EGC, and has achievable diagnostic ability even comparable to experienced endoscopists. In this review we summarized various AI-assisted systems in the diagnosis of EGC, CAG, and H. pylori infection.

Robotic large‐area optical biopsy imaging for automated detection of gastrointestinal cancers tested in tissue phantoms and ex vivo porcine bowel

AbstractGastrointestinal endoscopy is a subjective procedure that frequently requires tissue samples for diagnosis. Contact optical biopsy (OB) techniques have the aim of providing direct diagnosis of endoscopic areas without excising tissue samples but lack the wide‐area coverage required for locating and resecting lesions. This article presents a large‐area robotically deployed OB imaging platform for endoscopic detection of colorectal cancer as an add‐on for conventional endoscopes. In vitro, in silicon colon phantoms, the platform achieves an optical resolution of 0.5 line pairs per millimeter, while resolving simulated cancer lesions down to 0.75 mm diameter across large‐area images (55‐103 cm2). Large‐area OB images were generated in an ex vivo porcine colon. The platform allows centimeter‐sized large‐area OB imaging in vitro and ex vivo with submillimeter resolution, including automatic data segmentation of simulated cancer areas. The ability for robotic actuation and spectrum collection is also shown for ex vivo animal colon. If successful, this technology could widen access to user‐independent high‐quality endoscopy and early detection of gastrointestinal cancers.

Investigation of the New Advantages of Colonoscope Insertion with an Endoscopic Position Detection Unit.

The use of an endoscopic position detection unit (UPD) enables better and more objective understanding of the shape and position of the colonoscope. Here, we investigated the reproducibility of the insertion of a colonoscope with UPD. Study participants were 122 patients who received a colonoscopy with UPD twice for the purpose of large bowel screening and surveillance. The mean age of participants was 69.7 ± 10.4 years, and the male-to-female ratio was 9.2:1. The colonoscope insertion technique was primarily based on the shaft-holding, shortening insertion method. The cecal intubation time was recorded; the method used for passing through the sigmoid/descending colon junction (SDJ) and the hepatic flexure. The mean cecal intubation time was 990 ± 511 s. The cecal intubation time and the patterns for passing through the SDJ and hepatic flexure were significantly correlated between the first and second colonoscopies. Use of a UPD revealed good reproducibility of colonoscope insertion. This is the first time we have proved that both time and pattern are inserted in much the same way for the first and second times. In patients' conducted UPD combination TCS after the second time, it was suggested that individual tailor-made insertions are possible based on the information from the first time.

An Improved Method of Polyp Detection Using Custom YOLOv4-Tiny

Automatic detection of Wireless Endoscopic Images can avoid dangerous possible diseases such as cancers. Therefore, a number of articles have been published on different methods to enhance the speed of detection and accuracy. We also present a custom version of the YOLOv4-tiny for Wireless Endoscopic Image detection and localization that uses a You Only Look Once (YOLO) version to enhance the model accuracy. We modified the YOLOv4-tiny model by replacing the CSPDarknet-53-tiny backbone structure with the Inception-ResNet-A block to enhance the accuracy of the original YOLOv4-tiny. In addition, we implemented a new custom data augmentation method to enhance the data quality, even for small datasets. We focused on maintaining the color of medical images because the sensitivity of medical images can affect the efficiency of the model. Experimental results showed that our proposed method obtains 99.4% training accuracy; compared with the previous models, this is more than a 1.2% increase. An original model used for both detection and the segmentation of medical images may cause a high error rate. In contrast, our proposed model could eliminate the error rate of the detection and localization of disease areas from wireless endoscopic images.

Automatic detection of early gastric cancer in endoscopy based on Mask region-based convolutional neural networks (Mask R-CNN)(with video)

The artificial intelligence (AI)-assisted endoscopic detection of early gastric cancer (EGC) has been preliminarily developed. The currently used algorithms still exhibit limitations of large calculation and low-precision expression. The present study aimed to develop an endoscopic automatic detection system in EGC based on a mask region-based convolutional neural network (Mask R-CNN) and to evaluate the performance in controlled trials. For this purpose, a total of 4,471 white light images (WLIs) and 2,662 narrow band images (NBIs) of EGC were obtained for training and testing. In total, 10 of the WLIs (videos) were obtained prospectively to examine the performance of the RCNN system. Furthermore, 400 WLIs were randomly selected for comparison between the Mask R-CNN system and doctors. The evaluation criteria included accuracy, sensitivity, specificity, positive predictive value and negative predictive value. The results revealed that there were no significant differences between the pathological diagnosis with the Mask R-CNN system in the WLI test (χ2 = 0.189, P=0.664; accuracy, 90.25%; sensitivity, 91.06%; specificity, 89.01%) and in the NBI test (χ2 = 0.063, P=0.802; accuracy, 95.12%; sensitivity, 97.59%). Among 10 WLI real-time videos, the speed of the test videos was up to 35 frames/sec, with an accuracy of 90.27%. In a controlled experiment of 400 WLIs, the sensitivity of the Mask R-CNN system was significantly higher than that of experts (χ2 = 7.059, P=0.000; 93.00% VS 80.20%), and the specificity was higher than that of the juniors (χ2 = 9.955, P=0.000, 82.67% VS 71.87%), and the overall accuracy rate was higher than that of the seniors (χ2 = 7.009, P=0.000, 85.25% VS 78.00%). On the whole, the present study demonstrates that the Mask R-CNN system exhibited an excellent performance status for the detection of EGC, particularly for the real-time analysis of WLIs. It may thus be effectively applied to clinical settings.

Application of an artificial intelligence system for endoscopic diagnosis of superficial esophageal squamous cell carcinoma.

BACKGROUNDUpper gastrointestinal endoscopy is critical for esophageal squamous cell carcinoma (ESCC) detection; however, endoscopists require long-term training to avoid missing superficial lesions. AIMTo develop a deep learning computer-assisted diagnosis (CAD) system for endoscopic detection of superficial ESCC and investigate its application value. METHODSWe configured the CAD system for white-light and narrow-band imaging modes based on the YOLO v5 algorithm. A total of 4447 images from 837 patients and 1695 images from 323 patients were included in the training and testing datasets, respectively. Two experts and two non-expert endoscopists reviewed the testing dataset independently and with computer assistance. The diagnostic performance was evaluated in terms of the area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity. RESULTSThe area under the receiver operating characteristics curve, accuracy, sensitivity, and specificity of the CAD system were 0.982 [95% confidence interval (CI): 0.969-0.994], 92.9% (95%CI: 89.5%-95.2%), 91.9% (95%CI: 87.4%-94.9%), and 94.7% (95%CI: 89.0%-97.6%), respectively. The accuracy of CAD was significantly higher than that of non-expert endoscopists (78.3%, P < 0.001 compared with CAD) and comparable to that of expert endoscopists (91.0%, P = 0.129 compared with CAD). After referring to the CAD results, the accuracy of the non-expert endoscopists significantly improved (88.2% vs 78.3%, P < 0.001). Lesions with Paris classification type 0-IIb were more likely to be inaccurately identified by the CAD system. CONCLUSIONThe diagnostic performance of the CAD system is promising and may assist in improving detectability, particularly for inexperienced endoscopists.

Deep learning for classification and localization of early gastric cancer in endoscopic images

Gastric cancer, as a malignant tumor, is one of the most common cancer-related deaths worldwide with high mortality and incidence rates. Therefore, the endoscopic detection of gastric cancer at an early stage is essential. In this paper, we propose an automatic diagnosis method of early gastric cancer (EGC) based on deep learning (DL) techniques. Specifically, with the new annotated endoscopic image dataset collected from a single-center, this paper designs several DL architectures to realize the automatic analysis of EGC images. Particularly, a guided-attention deep network was introduced, derived from ResNet-50, for the accurate score prediction of EGC and the extraction of feature information. Furthermore, we combined a lightweight attention module and multi-scale feature extractor with U-Net for estimating the pixel-level segmentation of EGC pathological regions. Experiment results on the presented dataset showed outstanding performance in the involved classification and segmentation assignments with an accuracy of 98.84% and an intersection over union (IOU) of 0.64, revealing the potential applications of DL in aiding and improving EGC diagnosis using endoscopic images.