In a set of 8,641 colonoscopy images containing 4,088 unique polyps, the CNN identified polyps with a cross-validation accuracy of 96.4% and an area under the receiver operating characteristic curve of 0.991. The CNN system detected and localized polyps well within real-time constraints using an ordinary desktop machine with a contemporary graphics processing unit. This system could increase the ADR and decrease interval colorectal cancers but requires validation in large multicenter trials.

No abstract

Helicobacter pylor is associated with an inflammatory reaction in the stomach and duodenum, yet the mechanism of this inflammatory infiltrate is unknown. The ability of Helicobacter pylon to secrete a factor that attracts leucocytes is investigated. Helicobacterpylori conditioned supernatant attracted neutrophils and monocytes with 50-100% of the activity of control chemotactic factor, 10-8M formyl-methionine-leucine-phenylalanine. Strains derived from individuals with ulcer or non-ulcer associated H pylori infections displayed similar chemotactic activity. Preliminary characterisation shows that the factor has a molecular weight of less than 3000, is heat stable, is acid resistant, and can be diluted at least 10-fold. Checkerboard analysis confirmed that the activity was chemotactic rather than chemokinetic. This chemotactic activity could play a role in the pathogenesis ofHelicobacter pylori gastritis. (Gut 1992; 33: Chemotaxis was determined in multiwell chambers (Neuro Probe, Inc, Cabin John, MD, USA)7 using a 5 [i pore size filter (Nucleopore Corp, Pleasanton, CA) that separated the upper well, containing the leucocytes, from the lower well which contained the putative chemotactic factors. Formyl-methionine-leucine-phenylalanine (FMLP), a well characterised chemotactic peptide for monocytes and neutrophils, was used at 10-8 M as the positive control, and negative control using phosphate buffered saline were run simultaneously. All experiments were performed in triplicate. After one hour of incubation at 300C, the filters were removed, fixed in methanol, and stained with giemsa. The number of cells that had migrated to the lower aspect of the filter, toward the putative chemotactic factor were counted microscopically to determine the average number of cells per high power field (cells/hpf). The ratio of the activity of the H pylori supernatant compared to the activity of FMLP is used to determine the % of FMLP activity.Checkerboard analysis' was carried out to distinguish chemotactic from chemokinetic activity. Serial dilutions of the test factor were placed in the lower well of the chamber, and similar concentrations of the factor were used to dilute the leucocytes used in the upper well of the chamber. Directional movement of the cells in response to a concentration gradient indicates chemotaxis, whereas increased locomotion of the cells induced by the factor unrelated to the concentration gradient indicates chemokinesis. Results are expressed as the per cent of the undiluted activity. ResultsThe chemotactic activity of Helicobacter pylon supernatants is summarised in Figure 1. The supernatants showed chemotactic activity for 1020

Introduction-Reliable in situ diagnosis of diminutive (≤ 5 mm) colorectal polyps could allow for "resect and discard" and "diagnose and leave" strategies resulting in one billion dollars in cost savings per year in the U.S. alone. Current methodologies have failed to consistently meet the Preservation and Incorporation of Valuable endoscopic Innovations (PIVI) initiative thresholds. Convolutional neural networks (CNN) have the potential to predict polyp pathology and achieve PIVI thresholds in real-time.Methods-We developed a CNN-based optical pathology (OP) model using Tensorflow and pretrained on ImageNet, capable of operating at 77 frames per second. 6223 images of unique colorectal polyps of known pathology, location, size, and light source (white light [WL] or narrow band imaging [NBI]) underwent 5-fold cross training (80%) and validation (20%). Separate fresh validation was performed on 634 polyp images. Surveillance intervals were calculated, comparing OP vs. true pathology (TP).Results-In the original validation set, the negative predictive value (NPV) for adenomas was 97% among diminutive rectum/rectosigmoid polyps. Results were independent of use of NBI or WL. Surveillance interval concordance comparing OP and TP was 93%. In the fresh validation set, NPV was 97% among diminutive polyps in the rectum and rectosigmoid and surveillance concordance was 94%. Conclusion:This study demonstrates the feasibility of in situ diagnosis of colorectal polyps using CNN. Our model exceeds PIVI thresholds for both "resect and discard" and "diagnose and