Florian Grimpen

Gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS): a new autosomal dominant syndrome

Objective The purpose of this study was the clinical and pathological characterisation of a new autosomal dominant gastric polyposis syndrome, gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS). Methods Case series were examined, documenting GAPPS in three families from Australia, the USA and Canada. The affected families were identified through referral to centralised clinical genetics centres. Results The report identifies the clinical and pathological features of this syndrome, including the predominant dysplastic fundic gland polyp histology, the exclusive involvement of the gastric body and fundus, the apparent inverse association with current Helicobacter pylori infection and the autosomal dominant mode of inheritance. Conclusions GAPPS is a unique gastric polyposis syndrome with a significant risk of gastric adenocarcinoma. It is characterised by the autosomal dominant transmission of fundic gland polyposis, including areas of dysplasia or intestinal-type gastric adenocarcinoma, restricted to the proximal stomach, and with no evidence of colorectal or duodenal polyposis or other heritable gastrointestinal cancer syndromes.

Impact of colonoscopy bowel preparation on intestinal microbiota.

The gut microbiota is important in maintaining human health, but numerous factors have the potential to alter its composition. Our aim was to examine the impact of a standard bowel preparation on the intestinal microbiota using two different techniques. Fifteen subjects undergoing colonoscopy consumed a bowel preparation comprised of 10 mg bisacodyl and 2 L polyethylene glycol. The microbiota of stool samples, collected one month before, one week before (pre-colonoscopy), and one week, one month, and three to six months after colonoscopy (post-colonoscopy) was evaluated. Two samples were taken three to six months apart from five healthy subjects who did not undergo colonoscopy. Universal primers targeting the V2–V3 region of the 16S rRNA gene were used to PCR amplify all samples for denaturing gradient gel electrophoresis (PCR-DGGE). Pre- and post-colonoscopy samples were compared using Dice’s similarity coefficients. Three samples from ten subjects who underwent colonoscopy, and both samples from the five subjects who didn’t, were used for high-throughput sequencing of the V1–V3 region of the 16S rRNA gene. Samples were curated and analysed in Mothur. Results of the DGGE analyses show that the fecal microbiota of a small number of subjects had short-term changes. High-throughput sequencing results indicated that the variation between the samples of subjects who underwent colonoscopy was no greater than the variation observed between samples from subjects who did not. We conclude that bowel preparation does not have a lasting effect on the composition of the intestinal microbiota for the majority of subjects.

Australian clinical practice guidelines for the diagnosis and management of Barrett's esophagus and early esophageal adenocarcinoma

Barretts esophagus (BE), a common condition, is the only known precursor to esophageal adenocarcinoma (EAC). There is uncertainty about the best way to manage BE as most people with BE never develop EAC and most patients diagnosed with EAC have no preceding diagnosis of BE. Moreover, there have been recent advances in knowledge and practice about the management of BE and early EAC. To aid clinical decision making in this rapidly moving field, Cancer Council Australia convened an expert working party to identify pertinent clinical questions. The questions covered a wide range of topics including endoscopic and histological definitions of BE and early EAC; prevalence, incidence, natural history, and risk factors for BE; and methods for managing BE and early EAC. The latter considered modification of lifestyle factors; screening and surveillance strategies; and medical, endoscopic, and surgical interventions. To answer each question, the working party systematically reviewed the literature and developed a set of recommendations through consensus. Evidence underpinning each recommendation was rated according to quality and applicability.

Uninformative Frame Detection in Colonoscopy Through Motion, Edge and Color Features

Colonoscopy is performed by using a long endoscope inserted in the colon of patients to inspect the internal mucosa. During the intervention, clinicians observe the colon under bright light to diagnose pathology and guide intervention. We are developing a computer aided system to facilitate navigation and diagnosis. One essential step is to estimate the camera pose relative to the colon from video frames. However, within every colonoscopy video is a large number of frames that provide no structural information e.g. blurry or out of focus frames or those close to the colon wall. This hampers our camera pose estimation algorithm. To distinguish uninformative frames from informative ones, we investigated several features computed from each frame: corner and edge features matched with the previous frame, the percentage of edge pixels, and the mean and standard deviation of intensity in hue-saturation-value color space. A Random Forest classifier was used for classification. The method was validated on four colonoscopy videos that were manually classified. The resulting classification had a sensitivity of 75i¾ź% and specificity of 97i¾ź% for detecting uninformative frames. The proposed features not only compared favorably to existing techniques for detecting uninformative frames, but they also can be utilized for the camera navigation purpose.

Unsupervised Learning of Endoscopy Video Frames’ Correspondences from Global and Local Transformation

Inferring the correspondences between consecutive video frames with high accuracy is essential for many medical image processing and computer vision tasks (e.g. image mosaicking, 3D scene reconstruction). Image correspondences can be computed by feature extraction and matching algorithms, which are computationally expensive and are challenged by low texture frames. Convolutional neural networks (CNN) can estimate dense image correspondences with high accuracy, but lack of labeled data especially in medical imaging does not allow end-to-end supervised training. In this paper, we present an unsupervised learning method to estimate dense image correspondences (DIC) between endoscopy frames by developing a new CNN model, called the EndoRegNet. Our proposed network has three distinguishing aspects: a local DIC estimator, a polynomial image transformer which regularizes local correspondences and a visibility mask which refines image correspondences. The EndoRegNet was trained on a mix of simulated and real endoscopy video frames, while its performance was evaluated on real endoscopy frames. We compared the results of EndoRegNet with traditional feature-based image registration. Our results show that EndoRegNet can provide faster and more accurate image correspondences estimation. It can also effectively deal with deformations and occlusions which are common in endoscopy video frames without requiring any labeled data.

Learning Camera Pose from Optical Colonoscopy Frames Through Deep Convolutional Neural Network (CNN)

Optical colonoscopy is performed by insertion of a long flexible colonoscope into the colon. Estimating the position of the colonoscope tip with respect to the colon surface is important as it would help localization of cancerous polyps for subsequent surgery and facilitate navigation. Knowing camera pose is also essential for 3D automatic scene reconstruction, which could support clinicians inspecting the whole colon surface thereby reducing missed polyps. This paper presents a method to estimate the pose of the colonoscope camera with six degrees of freedom (DoF) using deep convolutional neural network (CNN). Because obtaining a ground truth to train the CNN for camera pose from actual colonoscopy videos is extremely challenging, we trained the CNN using realistic synthetic videos generated with a colonoscopy simulator, which could generate the exact camera pose parameters. We validated the trained CNN on unseen simulated video datasets and on actual colonoscopy videos from 10 patients. Our results showed that the colonoscopy camera pose could be estimated with higher accuracy and speed than feature based computer vision methods such as the classical structure from motion (SfM) pipeline. This paper demonstrates that transfer learning from surgical simulation to actual endoscopic based surgery is a possible approach for deep learning technologies.