Comparison of clinical utility of deep learning-based systems for small-bowel capsule endoscopy reading.

BACKGROUND AND AIM: Convolutional neural network (CNN) systems that automatically detect abnormalities from small-bowel capsule endoscopy (SBCE) images are still experimental, and no studies have directly compared the clinical usefulness of different systems. We compared endoscopist readings using an existing and a novel CNN system in a real-world SBCE setting. METHODS: Thirty-six complete SBCE videos, including 43 abnormal lesions (18 mucosal breaks, 8 angioectasia, and 17 protruding lesions), were retrospectively prepared. Three reading processes were compared: (A) endoscopist readings without CNN screening, (B) endoscopist readings after an existing CNN screening, and (C) endoscopist readings after a novel CNN screening. RESULTS: The mean number of small-bowel images was 14 747 per patient. Among these images, existing and novel CNN systems automatically captured 24.3% and 9.4% of the images, respectively. In this process, both systems extracted all 43 abnormal lesions. Next, we focused on the clinical usefulness. The detection rates of abnormalities by trainee endoscopists were not significantly different across the three processes: A, 77%; B, 67%; and C, 79%. The mean reading time of the trainees was the shortest during process C (10.1 min per patient), followed by processes B (23.1 min per patient) and A (33.6 min per patient). The mean psychological stress score while reading videos (scale, 1-5) was the lowest in process C (1.8) but was not significantly different between processes B (2.8) and A (3.2). CONCLUSIONS: Our novel CNN system significantly reduced endoscopist reading time and psychological stress while maintaining the detectability of abnormalities. CNN performance directly affects clinical utility and should be carefully assessed.

Efficacy of Early Video Capsule Endoscopy for Acute Overt Lower Gastrointestinal Bleeding with Colonic Diverticulosis: A Prospective Observational Study.

BACKGROUND/AIMS: Although most patients with presumptive colonic diverticular bleeding (CDB) do not undergo a small bowel investigation in clinical practice, no prospective study supports this management. We evaluated the utility of early small bowel capsule endoscopy (CE) after negative colonoscopy results. METHODS: This prospective study evaluated the diagnostic yield of early small bowel CE (≤3 days from visit) for consecutive patients with acute-onset hematochezia, when colonoscopy found colonic diverticulosis but did not identify the definite bleeding source (n = 51; presumptive CDB). As a matched control for comparing clinical outcomes, presumptive CDB patients without CE (n = 51) were retrospectively extracted. RESULTS: On CE for the prospective cohort, the rates of total positive findings, P2 findings (high bleeding potential according to the P classification), and blood pooling in the colon were 57%, 12% (ulceration, 8%; angioectasia, 4%), and 24%, respectively. The rates of rebleeding within 30 and 365 days were 16% and 29% in the prospective cohort with CE, respectively, and were not significantly different from those in the retrospective cohort without CE (10% and 25%, respectively). In addition, thromboembolism and mortality within 30 and 365 days were not significantly different between those with and without CE. CONCLUSION: Early CE detected a suspected small bowel bleeding source in 12% of acute-onset presumptive CDB patients but did not significantly improve major clinical outcomes. Therefore, routine CE is unnecessary for presumptive CDB patients after colonoscopy (UMIN000026676).

Automatic detection of various abnormalities in capsule endoscopy videos by a deep learning-based system: a multicenter study.

BACKGROUND AND AIMS: A deep convolutional neural network (CNN) system could be a high-level screening tool for capsule endoscopy (CE) reading but has not been established for targeting various abnormalities. We aimed to develop a CNN-based system and compare it with the existing QuickView mode in terms of their ability to detect various abnormalities. METHODS: We trained a CNN system using 66,028 CE images (44,684 images of abnormalities and 21,344 normal images). The detection rate of the CNN for various abnormalities was assessed per patient, using an independent test set of 379 consecutive small-bowel CE videos from 3 institutions. Mucosal breaks, angioectasia, protruding lesions, and blood content were present in 94, 29, 81, and 23 patients, respectively. The detection capability of the CNN was compared with that of QuickView mode. RESULTS: The CNN picked up 1,135,104 images (22.5%) from the 5,050,226 test images, and thus, the sampling rate of QuickView mode was set to 23% in this study. In total, the detection rate of the CNN for abnormalities per patient was significantly higher than that of QuickView mode (99% vs 89%, P < .001). The detection rates of the CNN for mucosal breaks, angioectasia, protruding lesions, and blood content were 100% (94 of 94), 97% (28 of 29), 99% (80 of 81), and 100% (23 of 23), respectively, and those of QuickView mode were 91%, 97%, 80%, and 96%, respectively. CONCLUSIONS: We developed and tested a CNN-based detection system for various abnormalities using multicenter CE videos. This system could serve as an alternative high-level screening tool to QuickView mode.

Automatic detection of different types of small-bowel lesions on capsule endoscopy images using a newly developed deep convolutional neural network.

BACKGROUND : Previous computer-aided detection systems for diagnosing lesions in images from wireless capsule endoscopy (WCE) have been limited to a single type of small-bowel lesion. We developed a new artificial intelligence (AI) system able to diagnose multiple types of lesions, including erosions and ulcers, vascular lesions, and tumors. METHODS : We trained the deep neural network system RetinaNet on a data set of 167 patients, which consisted of images of 398 erosions and ulcers, 538 vascular lesions, 4590 tumors, and 34 437 normal tissues. We calculated the mean area under the receiver operating characteristic curve (AUC) for each lesion type using five-fold stratified cross-validation. RESULTS : The mean age of the patients was 63.6 years; 92 were men. The mean AUCs of the AI system were 0.996 (95 %CI 0.992 - 0.999) for erosions and ulcers, 0.950 (95 %CI 0.923 - 0.978) for vascular lesions, and 0.950 (95 %CI 0.913 - 0.988) for tumors. CONCLUSION : We developed and validated a new computer-aided diagnosis system for multiclass diagnosis of small-bowel lesions in WCE images.

Automatic detection of blood content in capsule endoscopy images based on a deep convolutional neural network.

BACKGROUND AND AIM: Detecting blood content in the gastrointestinal tract is one of the crucial applications of capsule endoscopy (CE). The suspected blood indicator (SBI) is a conventional tool used to automatically tag images depicting possible bleeding in the reading system. We aim to develop a deep learning-based system to detect blood content in images and compare its performance with that of the SBI. METHODS: We trained a deep convolutional neural network (CNN) system, using 27 847 CE images (6503 images depicting blood content from 29 patients and 21 344 images of normal mucosa from 12 patients). We assessed its performance by calculating the area under the receiver operating characteristic curve (ROC-AUC) and its sensitivity, specificity, and accuracy, using an independent test set of 10 208 small-bowel images (208 images depicting blood content and 10 000 images of normal mucosa). The performance of the CNN was compared with that of the SBI, in individual image analysis, using the same test set. RESULTS: The AUC for the detection of blood content was 0.9998. The sensitivity, specificity, and accuracy of the CNN were 96.63%, 99.96%, and 99.89%, respectively, at a cut-off value of 0.5 for the probability score, which were significantly higher than those of the SBI (76.92%, 99.82%, and 99.35%, respectively). The trained CNN required 250 s to evaluate 10 208 test images. CONCLUSIONS: We developed and tested the CNN-based detection system for blood content in CE images. This system has the potential to outperform the SBI system, and the patient-level analyses on larger studies are required.

Clinical usefulness of a deep learning-based system as the first screening on small-bowel capsule endoscopy reading.

BACKGROUND AND AIM: To examine whether our convolutional neural network (CNN) system based on deep learning can reduce the reading time of endoscopists without oversight of abnormalities in the capsule-endoscopy reading process. METHODS: Twenty videos of the entire small-bowel capsule endoscopy procedure were prepared, each of which included 0-5 lesions of small-bowel mucosal breaks (erosions or ulcerations). At another institute, two reading processes were compared: (A) endoscopist-alone readings and (B) endoscopist readings after the first screening by the proposed CNN. In process B, endoscopists read only images detected by CNN. Two experts and four trainees independently read 20 videos each (10 for process A and 10 for process B). Outcomes were reading time and detection rate of mucosal breaks by endoscopists. Gold standard was findings at the original institute by two experts. RESULTS: Mean reading time of small-bowel sections by endoscopists was significantly shorter during process B (expert, 3.1 min; trainee, 5.2 min) compared to process A (expert, 12.2 min; trainee, 20.7 min) (P < 0.001). For 37 mucosal breaks, detection rate by endoscopists did not significantly decrease in process B (expert, 87%; trainee, 55%) compared to process A (expert, 84%; trainee, 47%). Experts detected all eight large lesions (>5 mm), but trainees could not, even when supported by the CNN. CONCLUSIONS: Our CNN-based system for capsule endoscopy videos reduced the reading time of endoscopists without decreasing the detection rate of mucosal breaks. However, the reading level of endoscopists should be considered when using the system.

Automatic detection of erosions and ulcerations in wireless capsule endoscopy images based on a deep convolutional neural network.

BACKGROUND AND AIMS: Although erosions and ulcerations are the most common small-bowel abnormalities found on wireless capsule endoscopy (WCE), a computer-aided detection method has not been established. We aimed to develop an artificial intelligence system with deep learning to automatically detect erosions and ulcerations in WCE images. METHODS: We trained a deep convolutional neural network (CNN) system based on a Single Shot Multibox Detector, using 5360 WCE images of erosions and ulcerations. We assessed its performance by calculating the area under the receiver operating characteristic curve and its sensitivity, specificity, and accuracy using an independent test set of 10,440 small-bowel images including 440 images of erosions and ulcerations. RESULTS: The trained CNN required 233 seconds to evaluate 10,440 test images. The area under the curve for the detection of erosions and ulcerations was 0.958 (95% confidence interval [CI], 0.947-0.968). The sensitivity, specificity, and accuracy of the CNN were 88.2% (95% CI, 84.8%-91.0%), 90.9% (95% CI, 90.3%-91.4%), and 90.8% (95% CI, 90.2%-91.3%), respectively, at a cut-off value of 0.481 for the probability score. CONCLUSIONS: We developed and validated a new system based on CNN to automatically detect erosions and ulcerations in WCE images. This may be a crucial step in the development of daily-use diagnostic software for WCE images to help reduce oversights and the burden on physicians.

Etiology and long-term rebleeding of endoscopic ulcerative lesions in the small bowel in patients with obscure gastrointestinal bleeding: A multicenter cohort study.

BACKGROUND: Among patients with obscure gastrointestinal bleeding (OGIB), endoscopic ulcerative lesions in the small bowel have diverse etiologies and often cause rebleeding. Certain characteristics of patients or ulcerations may be reasonable indications for diagnostic balloon-assisted endoscopy (BAE) to assess etiology and may be risks of rebleeding; however, these characteristics are unclear. We aimed to elucidate appropriate indications for diagnostic BAE and predictors of long-term rebleeding in patients with small bowel ulcerative lesions. METHODS: We conducted a multicenter retrospective cohort study of 68 patients with OGIB, in whom small bowel ulcerative lesions were detected by capsule endoscopy (n = 60) and/or BAE (n = 43). Patients' characteristics, including medications and endoscopic findings, were evaluated. Predictors of the need for diagnostic BAE to determine ulceration etiology were identified by logistic regression analysis. Rebleeding risks were evaluated using Cox proportional hazards analysis. RESULTS: Single ulcerations were diagnosed in 26 patients, and multiple ulcerations were diagnosed in 42 patients. Among 43 patients who underwent BAE, ulceration etiology was identified in 12 (28%) patients. In the etiology identification, BAE was more useful for a single ulceration than for multiple ulcerations (P < 0.001). Among the 68 patients, rebleeding occurred in 14 (21%) patients during a mean follow-up period of 17 months. Aspirin use and multiple ulcerations were significant predictors of rebleeding (P < 0.05). CONCLUSIONS: When we manage small bowel ulcerative lesions in OGIB patients, a single ulceration is a reasonable indication for the diagnostic BAE. The rebleeding rate was lower for single ulcerations than for multiple ulcerations.

The incidence of post-colonoscopy colorectal cancer: a retrospective long-term cohort study using a colonoscopy database.

PURPOSE: The cumulative incidence of post-colonoscopy colorectal cancer remains unclear. Our aims were to estimate the incidence of and identify risk factors associated with post-colonoscopy colorectal cancer. METHODS: We conducted a retrospective cohort study using the colonoscopy database of the Department of Gastroenterology, the University of Tokyo Hospital Records from1995-2012. A cohort of 2544 patients, who received multiple colonoscopies without colorectal cancer findings at first colonoscopy, was selected. The primary outcome was post-colonoscopy colorectal cancer; data were censored at the date of final colonoscopy. We assessed patients' background characteristics, colonoscopy findings, and cancer characteristics, including location and size. The cumulative incidence of colorectal cancer was evaluated, and a Cox proportional hazards model was used to estimate hazard ratios (HRs). RESULTS: Colorectal cancer was identified in seven (0.77/1000 person-years) patients during the mean follow-up period of 3.6 years (maximum, 17 years). The cumulative incidence of colorectal cancer was 0, 0.47, 0.62, and 0.62% at 1, 5, 10, and 15 years, respectively. Cancer was identified in the rectum in five of seven patients. Polyp size >10 mm (HR 5.7, p = 0.023) and intubation time >30 min (HR 11.6, p = 0.003) at first colonoscopy were associated significantly with an increased incidence of post-colonoscopy colorectal cancer. CONCLUSIONS: Although several factors were associated with an increased risk of post-colonoscopy colorectal cancer, the incidence of this disease might be low in patients who received at least twice colonoscopy. High proportion of rectal cancer in post-colonoscopy colorectal cancer should be noted.

Improved Object Detection Artificial Intelligence Using the Revised RetinaNet Model for the Automatic Detection of Ulcerations, Vascular Lesions, and Tumors in Wireless Capsule Endoscopy.

The use of computer-aided detection models to diagnose lesions in images from wireless capsule endoscopy (WCE) is a topical endoscopic diagnostic solution. We revised our artificial intelligence (AI) model, RetinaNet, to better diagnose multiple types of lesions, including erosions and ulcers, vascular lesions, and tumors. RetinaNet was trained using the data of 1234 patients, consisting of images of 6476 erosions and ulcers, 1916 vascular lesions, 7127 tumors, and 14,014,149 normal tissues. The mean area under the receiver operating characteristic curve (AUC), sensitivity, and specificity for each lesion were evaluated using five-fold stratified cross-validation. Each cross-validation set consisted of between 6,647,148 and 7,267,813 images from 217 patients. The mean AUC values were 0.997 for erosions and ulcers, 0.998 for vascular lesions, and 0.998 for tumors. The mean sensitivities were 0.919, 0.878, and 0.876, respectively. The mean specificities were 0.936, 0.969, and 0.937, and the mean accuracies were 0.930, 0.962, and 0.924, respectively. We developed a new version of an AI-based diagnostic model for the multiclass identification of small bowel lesions in WCE images to help endoscopists appropriately diagnose small intestine diseases in daily clinical practice.

Detection of circulating colorectal cancer cells by a custom microfluid system before and after endoscopic metallic stent placement.

Although the detection of circulating tumor cells (CTCs) should be crucial for future personalized medicine, no efficient and flexible methods have been established. The current study established a polymeric custom-made chip for capturing CTCs with a high efficiency and flexibility. As an example of clinical application, the effects of self-expandable metallic stent (SEMS) placement on the release of cancer cells into the blood of patients with colorectal cancer and bowel obstruction were analyzed. This was assessed as the placement of SEMS may cause mechanical damage and physical force to malignant tissue, increasing the risk of cancer cell release into the bloodstream. The present study examined the number of CTCs using a custom-made chip, before, at 24 h after and at 4 days after SEMS placement in patients with colorectal cancer. The results revealed that, among the 13 patients examined, the number of CTCs was increased in three cases at 24 h after SEMS placement. However, this increase was temporary. The number of CTCs also decreased at 4 days after stent placement in most cases. The CTC chip of the current study detected the number of CD133-positive cancer stem-like cells, which did not change, even in the patient whose total number of CTCs temporarily increased. The results indicated that this custom-made microfluid system can efficiently and flexibly detect CTCs, demonstrating its potential for obtaining information during the management of patients with cancer.

Analysis of predictive factors for R0 resection and immediate bleeding of cold snare polypectomy in colonoscopy.

BACKGROUND: Factors associated with efficacy and safety of cold snare polypectomy (CSP) are not well established. The aim is to elucidate the predictors of R0 resection and immediate bleeding of CSP. METHODS: We retrospectively reviewed a database of patients who underwent CSP for subcentimetric polyps at the University of Tokyo Hospital in Japan. Using the data regarding the characteristics of patients and polyps, such as location, size, and macroscopic appearance; use of narrow band imaging with magnification (NBI-M); and endoscopists' experience, we revealed the predictive factors associated with R0 resection and immediate post-CSP bleeding by univariate and multivariate analyses. RESULTS: In total, 399 polyps, in 200 patients without antithrombotics, were removed. Failure of tissue retrieval was noted in 4% of resected lesions. There was no intramucosal carcinoma observed. The overall rate of R0 resection was 46%. Multivariate analysis elucidated that the observation of the polyp with NBI-M was an independent predictor associated with R0 resection (odds ratio [OR] 1.90; p = 0.024). Although immediate post-CSP bleeding occurred in 19 polyps (4.8%), no delayed bleeding or perforation was observed. Multivariate analysis revealed protruded lesion as an independent risk factor for immediate bleeding (OR 3.54; p = 0.018). CONCLUSIONS: A higher rate of R0 resection with CSP can be achieved by performing colonoscopy with NBI-M, than with white-light imaging. Macroscopic protruding appearance of a polyp is a risk factor for immediate bleeding.