Search

We have developed the computer-aided detection (CADe) system using an original deep learning algorithm based on a convolutional neural network for assisting endoscopists in detecting colorectal lesions during colonoscopy. The aim of this study was to clarify whether adenoma miss rate (AMR) could be reduced with CADe assistance during screening and surveillance colonoscopy. This study was a multicenter randomized controlled trial. Patients aged 40 to 80 years who were referred for colorectal screening or surveillance at four sites in Japan were randomly assigned at a 1:1 ratio to either the “standard colonoscopy (SC)-first group” or the “CADe-first group” to undergo a back-to-back tandem procedure. Tandem colonoscopies were performed on the same day for each participant by the same endoscopist in a preassigned order. All polyps detected in each pass were histopathologically diagnosed after biopsy or resection. A total of 358 patients were enrolled and 179 patients were assigned to the SC-first group or CADe-first group. The AMR of the CADe-first group was significantly lower than that of the SC-first group (13.8% vs. 36.7%, P

Artificial intelligence (AI) tools aimed at improving polyp detection have been shown to increase the adenoma detection rate during colonoscopy. However, it is unknown how increased polyp detection rates by AI affect the burden of patient surveillance after polyp removal.We conducted a pooled analysis of 9 randomized controlled trials (5 in China, 2 in Italy, 1 in Japan, and 1 in the United States) comparing colonoscopy with or without AI detection aids. The primary outcome was the proportion of patients recommended to undergo intensive surveillance (ie, 3-year interval). We analyzed intervals for AI and non-AI colonoscopies for the U.S. and European recommendations separately. We estimated proportions by calculating relative risks using the Mantel-Haenszel method.A total of 5796 patients (51% male, mean 53 years of age) were included; 2894 underwent AI-assisted colonoscopy and 2902 non-AI colonoscopy. When following U.S. guidelines, the proportion of patients recommended intensive surveillance increased from 8.4% (95% CI, 7.4%-9.5%) in the non-AI group to 11.3% (95% CI, 10.2%-12.6%) in the AI group (absolute difference, 2.9% [95% CI, 1.4%-4.4%]; risk ratio, 1.35 [95% CI, 1.16-1.57]). When following European guidelines, it increased from 6.1% (95% CI, 5.3%-7.0%) to 7.4% (95% CI, 6.5%-8.4%) (absolute difference, 1.3% [95% CI, 0.01%-2.6%]; risk ratio, 1.22 [95% CI, 1.01-1.47]).The use of AI during colonoscopy increased the proportion of patients requiring intensive colonoscopy surveillance by approximately 35% in the United States and 20% in Europe (absolute increases of 2.9% and 1.3%, respectively). While this may contribute to improved cancer prevention, it significantly adds patient burden and healthcare costs.

We developed a computer-aided detection (CADe) system to detect and localize colorectal lesions by modifying You-Only-Look-Once version 3 (YOLO v3) and evaluated its performance in two different settings. The test dataset was obtained from 20 randomly selected patients who underwent endoscopic resection for 69 colorectal lesions at the Jikei University Hospital between June 2017 and February 2018. First, we evaluated the diagnostic performances using still images randomly and automatically extracted from video recordings of the entire endoscopic procedure at intervals of 5 s, without eliminating poor quality images. Second, the latency of lesion detection by the CADe system from the initial appearance of lesions was investigated by reviewing the videos. A total of 6531 images, including 662 images with a lesion, were studied in the image-based analysis. The AUC, sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.983, 94.6%, 95.2%, 68.8%, 99.4%, and 95.1%, respectively. The median time for detecting colorectal lesions measured in the lesion-based analysis was 0.67 s. In conclusion, we proved that the originally developed CADe system based on YOLO v3 could accurately and instantaneously detect colorectal lesions using the test dataset obtained from videos, mitigating operator selection biases.

Background Our group previously proved that the human enteric nervous system can be visualized with confocal laser endomicroscopy after topical application of cresyl violet using surgically resected intestine specimens. The present report documents the first in vivo visualization of the human enteric nervous system with confocal laser endomicroscopy using local cresyl violet staining. The aim of this study was to evaluate the technical feasibility and clinical efficiency of confocal laser endomicroscopy in patients with Hirschsprung’s disease and allied disorders in vivo. Methods Confocal laser endomicroscopy was performed in vivo in two patients to confirm the presence of the enteric nervous system during surgery in patients with Hirschsprung’s disease and allied disorders. Cresyl violet was gently injected from the serosal side into the muscular layer of the intestine, and scanning was performed within 30 min. Then, the scanned intestines were resected, and the visualized area of the specimens was pathologically evaluated. Results The ganglion cell nuclei and the enteric nervous system network were clearly visualized intraoperatively in both cases. The morphological findings were similar to the pathological findings of the enteric nervous system in both cases although the period of visibility was brief. Conclusion This study demonstrated the first, real-time observation of the enteric nervous system in humans using confocal laser endomicroscopy and suggest the potential to identify the enteric nervous system intra-operatively during surgery for Hirschsprung’s disease and allied disorders.

Artificial intelligence (AI) has been attracting considerable attention as an important scientific topic in the field of medicine. Deep-leaning (DL) technologies have been applied more dominantly than other traditional machine-learning methods. They have demonstrated excellent capability to retract visual features of objectives, even unnoticeable ones for humans, and analyze huge amounts of information within short periods. The amount of research applying DL-based models to real-time computer-aided diagnosis (CAD) systems has been increasing steadily in the GI endoscopy field. An array of published data has already demonstrated the advantages of DL-based CAD models in the detection and characterization of various neoplastic lesions, regardless of the level of the GI tract. Although the diagnostic performances and study designs vary widely, owing to a lack of academic standards to assess the capability of AI for GI endoscopic diagnosis fairly, the superiority of CAD models has been demonstrated for almost all applications studied so far. Most of the challenges associated with AI in the endoscopy field are general problems for AI models used in the real world outside of medical fields. Solutions have been explored seriously and some solutions have been tested in the endoscopy field. Given that AI has become the basic technology to make machines react to the environment, AI would be a major technological paradigm shift, for not only diagnosis but also treatment. In the near future, autonomous endoscopic diagnosis might no longer be just a dream, as we are witnessing with the advent of autonomously driven electric vehicles.

Our previous studies have shown the diagnostic utility of a newly developed dual-focus endoscope with narrow-band imaging (DF-NBI) and simplified dyad criteria for detection of superficial esophageal squamous cell carcinoma (SESCC). This clinical trial aimed to study the diagnostic efficacy of DF-NBI with dyad criteria for detecting SESCC compared to white light imaging (WLI). This was a single-arm prospective comparative trial. We enrolled 170 consecutive high-risk patients for esophageal squamous cell carcinoma. Patients were initially examined with WLI by one independent endoscopist and then the other performed DF-NBI blinded to the WLI diagnosis to avoid a carry-over effect. Lesions showing proliferation and/or various shapes of intrapapillary capillary loops (IPCL) under DF-NBI (i.e., dyad criteria) were endoscopically diagnosed as SESCC including high-grade intraepithelial neoplasia. The primary endpoint was sensitivity of WLI and DF-NBI for detecting SESCC. The secondary endpoints were the diagnostic performance (i.e., specificity and accuracy) and inter/intra-observer concordance of DF-NBI with dyad criteria. A total 77 SESCCs were detected. The sensitivity of DF-NBI for SESCC was significantly higher than that of WLI (91% vs. 51%, P

Background and Aim In preceding studies, we identified that the myenteric plexus (MP) could be visualized with confocal laser endomicroscopy (CLE) by applying neural fluorescent probes lacking clinical safety profiling data from the submucosal side. In this study, we evaluated the technical feasibility of MP visualization using probe-based CLE (pCLE) from the serosal side with cresyl violet (CV), which has been used clinically for chromoendoscopy. Methods The dye affinity of CV for MP was first explored in an in vivo transgenic mouse model using neural crest derivatives labeled with green fluorescent protein. We also tested the feasibility of CV-assisted visualization of MP in human surgical specimens, wherein the tissue dying and pCLE observation were performed from the serosal side. In the human study, rate of MP visualization by pCLE was evaluated as the primary outcome. We also evaluated the sensitivity and specificity of MP visualization by pCLE, using pathological presence/absence of MP as the gold standard. Results We confirmed the dye affinity of CV to MP in all tested models. The MP appeared as brightly stained ladder-like structures with pCLE, and in the human study, MP was visualized in 12/14 (85.7%) samples, with 92.3% sensitivity and 100% specificity. In positive cases showing the ladder-like structure of MP by pCLE, the mean maximum and minimum widths of nerve strands were 54.3 (± 23.6) and 19.7 (± 6.0) μm, respectively. A ganglion was detected by pCLE in 10 cases (10/12, 83.3%). Conclusions This study demonstrated the technical feasibility of visualizing the MP in real time by CV-assisted pCLE (UMIN-CTR number, UMIN000015056).

Background Hirschsprung's disease is a congenital abnormality of the enteric nervous system (ENS) presenting severe constipation soon after birth due to the lack of ganglion cells in the distal gut. Surgery for Hirschsprung's disease requires an intraoperative histopathological diagnosis to assess the extent of aganglionosis. Confocal laser endomicroscopy (CLE) is a novel endoscopic technique allowing real-time, in vivo analysis of cellular details during ongoing endoscopy. In this study, we evaluated the possibility of a new application of CLE to provide real-time observations of the ENS in patients with Hirschsprung's disease. In this preclinical feasibility study, we assessed the visualization of the ENS by CLE using surgically resected intestines. Methods The subjects were nine patients who underwent pull-through surgery for Hirschsprung's disease between September 2014 and March 2016. The colon specimens were stained with 0.1% cresyl violet and evaluated using CLE. We compared the CLE findings with those of the histopathological examination. Key results The ENS was clearly visualized as a ladder-like structure in the ganglionic segment but was not observed in the aganglionic segment. Of the 69 samples, corresponding positive and negative results for both CLE and the histopathology were obtained in 61 (88%). In addition, CLE was able to visualize unique, wavy structures comprising thick nerve bundles characteristic of the aganglionic/transition zone in Hirschsprung's disease. Conclusions and inferences As a novel tool for visualizing the human ENS, CLE has the potential to revolutionize how pediatric surgeons identify the level of ganglionosis during surgery for Hirschsprung's disease and may be a superior alternative to intraoperative histopathological diagnosis.

Blood group O of ABO blood group system is considered as a risk factor for various bleeding events, but the relationship with endoscopic treatment-associated bleeding has yet to be investigated. This study aimed to evaluate whether blood group O is associated with delayed bleeding after colorectal endoscopic resection. This was a retrospective observational study based on medical records at four university hospitals in Japan. We reviewed the records for consecutive patients who underwent colorectal endoscopic resection from January 2014 through December 2017. The primary outcome was the incidence of delayed bleeding, defined as hematochezia or melena, requiring endoscopy, transfusion, or any hemostatic intervention up to 28 days after endoscopic resection. Multivariate logistic regression analysis was performed to adjust the impact of blood group O on the delayed bleeding. Among 10,253 consecutive patients who underwent colorectal endoscopic resection during the study period, 8625 patients met the criteria. In total, delayed bleeding occurred in 255 patients (2.96%). The O group had significantly more bleeding events compared with the non-O group (A, B, and AB) (relative risk, 1.62 [95% confidence interval, 1.24–2.10]; P

Endoscopy is a standard method for the diagnosis and detection of colorectal lesions. As a method to enhance the detectability of lesions, the effectiveness of pancolonic chromoendoscopy with indigocarmine has been reported. On the other hand, computer-aided diagnosis (CAD) has attracted attention. However, existing CAD systems are mainly for white light imaging (WLI) endoscopy, and the effect of the combination of CAD and indigocarmine dye spraying is not clear. Besides, it is difficult to gather a lot of indigocarmine dye-sprayed (IC) images for training. Here, we propose image-to-image translation from WLI to virtual indigocarmine dye-sprayed (VIC) images based on unpaired cycle-consistent Generative Adversarial Networks. Using this generator as preprocess part, we constructed detection models to evaluate the effectiveness of VIC translation for localization and classification of lesions. We also compared the localization and classification performance with and without image augmentation by using generated VIC images. Our results show that the model trained on IC and VIC images had the highest performance in both localization and classification. Therefore, VIC images are useful for the augmentation of IC images.

In preceding studies, we identified that the myenteric plexus (MP) could be visualized with confocal laser endomicroscopy (CLE) by applying neural fluorescent probes lacking clinical safety profiling data from the submucosal side. In this study, we evaluated the technical feasibility of MP visualization using probe-based CLE (pCLE) from the serosal side with cresyl violet (CV), which has been used clinically for chromoendoscopy.The dye affinity of CV for MP was first explored in an in vivo transgenic mouse model using neural crest derivatives labeled with green fluorescent protein. We also tested the feasibility of CV-assisted visualization of MP in human surgical specimens, wherein the tissue dying and pCLE observation were performed from the serosal side. In the human study, rate of MP visualization by pCLE was evaluated as the primary outcome. We also evaluated the sensitivity and specificity of MP visualization by pCLE, using pathological presence/absence of MP as the gold standard.We confirmed the dye affinity of CV to MP in all tested models. The MP appeared as brightly stained ladder-like structures with pCLE, and in the human study, MP was visualized in 12/14 (85.7%) samples, with 92.3% sensitivity and 100% specificity. In positive cases showing the ladder-like structure of MP by pCLE, the mean maximum and minimum widths of nerve strands were 54.3 (± 23.6) and 19.7 (± 6.0) μm, respectively. A ganglion was detected by pCLE in 10 cases (10/12, 83.3%).This study demonstrated the technical feasibility of visualizing the MP in real time by CV-assisted pCLE (UMIN-CTR number, UMIN000015056).