A multi-centre polyp detection and segmentation dataset for generalisability assessment.

Polyps in the colon are widely known cancer precursors identified by colonoscopy. Whilst most polyps are benign, the polyp's number, size and surface structure are linked to the risk of colon cancer. Several methods have been developed to automate polyp detection and segmentation. However, the main issue is that they are not tested rigorously on a large multicentre purpose-built dataset, one reason being the lack of a comprehensive public dataset. As a result, the developed methods may not generalise to different population datasets. To this extent, we have curated a dataset from six unique centres incorporating more than 300 patients. The dataset includes both single frame and sequence data with 3762 annotated polyp labels with precise delineation of polyp boundaries verified by six senior gastroenterologists. To our knowledge, this is the most comprehensive detection and pixel-level segmentation dataset (referred to as PolypGen) curated by a team of computational scientists and expert gastroenterologists. The paper provides insight into data construction and annotation strategies, quality assurance, and technical validation.

An Expert Consensus to Standardize Assessment of Bowel Cleansing for Clinical Trials of Bowel Preparations for Crohn's Disease.

BACKGROUND: Despite regular need for colonoscopy in patients with Crohn's disease (CD), the efficacy and tolerability of bowel preparation (BP) agents is rarely assessed in this population. Assessing BP quality with existing scales may be challenging in CD due to presence of inflammation, bowel resection, and strictures. AIMS: To provide recommendations for assessing BP quality in clinical trials for CD using a modified Research and Development/University of California, Los Angeles appropriateness process. METHODS: Based on systematic reviews and a literature search, 110 statements relating to BP quality assessment in CD were developed. A panel of 15 gastroenterologists rated the statements as appropriate, uncertain, or inappropriate using a 9-point Likert scale. RESULTS: Panelists considered it appropriate that central readers, either alone or with local assessment, score BP quality in clinical trials. Central readers should be trained on scoring BP quality and local endoscopists on performing high-quality video recording. Both endoscope insertion and withdrawal phases should be reviewed to score BP quality in each colonic segment and segments should align with endoscopic disease activity indices. The Harefield Cleansing Scale and the Boston Bowel Preparation Scale were considered appropriate. The final score should be calculated as the average of all visualized segments. Both total and worst segment scores should also be assessed. CONCLUSIONS: We developed a framework for assessing BP quality in patients with CD based on expert feedback. This framework could support the development or refinement of BP quality scales and the integration of BP quality assessment in future CD studies.

A deep learning framework for quality assessment and restoration in video endoscopy.

Endoscopy is a routine imaging technique used for both diagnosis and minimally invasive surgical treatment. Artifacts such as motion blur, bubbles, specular reflections, floating objects and pixel saturation impede the visual interpretation and the automated analysis of endoscopy videos. Given the widespread use of endoscopy in different clinical applications, robust and reliable identification of such artifacts and the automated restoration of corrupted video frames is a fundamental medical imaging problem. Existing state-of-the-art methods only deal with the detection and restoration of selected artifacts. However, typically endoscopy videos contain numerous artifacts which motivates to establish a comprehensive solution. In this paper, a fully automatic framework is proposed that can: 1) detect and classify six different artifacts, 2) segment artifact instances that have indefinable shapes, 3) provide a quality score for each frame, and 4) restore partially corrupted frames. To detect and classify different artifacts, the proposed framework exploits fast, multi-scale and single stage convolution neural network detector. In addition, we use an encoder-decoder model for pixel-wise segmentation of irregular shaped artifacts. A quality score is introduced to assess video frame quality and to predict image restoration success. Generative adversarial networks with carefully chosen regularization and training strategies for discriminator-generator networks are finally used to restore corrupted frames. The detector yields the highest mean average precision (mAP) of 45.7 and 34.7, respectively for 25% and 50% IoU thresholds, and the lowest computational time of 88 ms allowing for near real-time processing. The restoration models for blind deblurring, saturation correction and inpainting demonstrate significant improvements over previous methods. On a set of 10 test videos, an average of 68.7% of video frames successfully passed the quality score (≥0.9) after applying the proposed restoration framework thereby retaining 25% more frames compared to the raw videos. The importance of artifacts detection and their restoration on improved robustness of image analysis methods is also demonstrated in this work.

Cost savings in colonoscopy with artificial intelligence-aided polyp diagnosis: an add-on analysis of a clinical trial (with video).

BACKGROUND AND AIMS: Artificial intelligence (AI) is being implemented in colonoscopy practice, but no study has investigated whether AI is cost saving. We aimed to quantify the cost reduction using AI as an aid in the optical diagnosis of colorectal polyps. METHODS: This study is an add-on analysis of a clinical trial that investigated the performance of AI for differentiating colorectal polyps (ie, neoplastic versus non-neoplastic). We included all patients with diminutive (≤5 mm) rectosigmoid polyps in the analyses. The average colonoscopy cost was compared for 2 scenarios: (1) a diagnose-and-leave strategy supported by the AI prediction (ie, diminutive rectosigmoid polyps were not removed when predicted as non-neoplastic), and (2) a resect-all-polyps strategy. Gross annual costs for colonoscopies were also calculated based on the number and reimbursement of colonoscopies conducted under public health insurances in 4 countries. RESULTS: Overall, 207 patients with 250 diminutive rectosigmoid polyps (104 neoplastic, 144 non-neoplastic, and 2 indeterminate) were included. AI correctly differentiated neoplastic polyps with 93.3% sensitivity, 95.2% specificity, and 95.2% negative predictive value. Thus, 105 polyps were removed and 145 were left under the diagnose-and-leave strategy, which was estimated to reduce the average colonoscopy cost and the gross annual reimbursement for colonoscopies by 18.9% and US$149.2 million in Japan, 6.9% and US$12.3 million in England, 7.6% and US$1.1 million in Norway, and 10.9% and US$85.2 million in the United States, respectively, compared with the resect-all-polyps strategy. CONCLUSIONS: The use of AI to enable the diagnose-and-leave strategy results in substantial cost reductions for colonoscopy.

Guidelines for the management of hereditary colorectal cancer from the British Society of Gastroenterology (BSG)/Association of Coloproctology of Great Britain and Ireland (ACPGBI)/United Kingdom Cancer Genetics Group (UKCGG).

Heritable factors account for approximately 35% of colorectal cancer (CRC) risk, and almost 30% of the population in the UK have a family history of CRC. The quantification of an individual's lifetime risk of gastrointestinal cancer may incorporate clinical and molecular data, and depends on accurate phenotypic assessment and genetic diagnosis. In turn this may facilitate targeted risk-reducing interventions, including endoscopic surveillance, preventative surgery and chemoprophylaxis, which provide opportunities for cancer prevention. This guideline is an update from the 2010 British Society of Gastroenterology/Association of Coloproctology of Great Britain and Ireland (BSG/ACPGBI) guidelines for colorectal screening and surveillance in moderate and high-risk groups; however, this guideline is concerned specifically with people who have increased lifetime risk of CRC due to hereditary factors, including those with Lynch syndrome, polyposis or a family history of CRC. On this occasion we invited the UK Cancer Genetics Group (UKCGG), a subgroup within the British Society of Genetic Medicine (BSGM), as a partner to BSG and ACPGBI in the multidisciplinary guideline development process. We also invited external review through the Delphi process by members of the public as well as the steering committees of the European Hereditary Tumour Group (EHTG) and the European Society of Gastrointestinal Endoscopy (ESGE). A systematic review of 10 189 publications was undertaken to develop 67 evidence and expert opinion-based recommendations for the management of hereditary CRC risk. Ten research recommendations are also prioritised to inform clinical management of people at hereditary CRC risk.

Surgery versus surveillance in ulcerative colitis patients with endoscopically invisible low-grade dysplasia: a cost-effectiveness analysis.

BACKGROUND AND AIMS: There is uncertainty regarding the optimal management of endoscopically invisible (flat) low-grade dysplasia in ulcerative colitis. Such a finding does not currently provide an automatic indication for colectomy; however, a recommendation of surveillance instead of surgery is controversial. The aim of this study was to determine the clinical and cost-effectiveness of colonoscopic surveillance versus colectomy for endoscopically invisible low-grade dysplasia of the colon in ulcerative colitis. METHODS: A Markov model was used to evaluate the costs and health outcomes of surveillance and surgery over a 20-year timeframe. Outcomes evaluated were life years gained and quality-adjusted life years (QALYs). Cohorts of patients aged 25 to 75 were modeled, including estimates from a validated surgical risk calculator and considering none, 1, or both of 2 key comorbidities: heart failure and obstructive airway disease. RESULTS: Surveillance is associated with more life years and QALYs compared with surgery from age 61 for those with no comorbidities, age 51 for those with 1 comorbidity and age 25 for those with 2 comorbidities. At the current United Kingdom National Institute for Health and Care Excellence threshold of $25,800 per QALY, ongoing surveillance was cost-effective at age 65 in those without comorbidities and at age 60 in those with either 1 or more comorbidities. CONCLUSIONS: Surveillance can be recommended from age 65 for those with no comorbidities; however, in younger patients with typical postsurgical quality of life, colectomy may be more effective clinically and more cost-effective. The results were sensitive to the colorectal cancer incidence rate in patients under surveillance and to quality of life after surgery.

Cost-effective and Large-scale synthesis of 16:0 Lysophosphatidic Acid.

Lysophosphatidic acid (LPA) is a bioactive compound that has gained attention due to its role in neoplastic diseases. Popularity of the compound has necessitated the use of large quantities of the phospholipid for in vivo and in vitro testing but methods for generating LPA require the use costly procedures, namely phosphoramidite coupling reagents. Additionally there has been no reported large-scale synthesis of LPA. In the present study we report the cost-effective and large-scale synthesis of 16:0 LPA.

Optical diagnosis of small colorectal polyps at routine colonoscopy (Detect InSpect ChAracterise Resect and Discard; DISCARD trial): a prospective cohort study.

BACKGROUND: Accurate optical diagnosis of small (<10 mm) colorectal polyps in vivo, without formal histopathology, could make colonoscopy more efficient and cost effective. The aim of this study was to assess whether optical diagnosis of small polyps is feasible and safe in routine clinical practice. METHODS: Consecutive patients with a positive faecal occult blood test or previous adenomas undergoing surveillance at St Mark's Hospital (London, UK), from June 19, 2008, to June 16, 2009, were included in this prospective study. Four colonoscopists with different levels of experience predicted polyp histology using optical diagnosis with high-definition white light, followed by narrow-band imaging without magnification and chromoendoscopy, as required. The primary outcome was accuracy of polyp characterisation using optical diagnosis compared with histopathology, the current gold standard. Accuracy of optical diagnosis to predict the next surveillance interval was also assessed and compared with surveillance intervals predicted by current guidelines using histopathology. This study is registered with ClinicalTrials.gov, NCT00888771. FINDINGS: 363 polyps smaller than 10 mm were detected in 130 patients, of which 278 polyps had both optical and histopathological diagnosis. By histology, 198 of these polyps were adenomas and 80 were non-neoplastic lesions (of which 62 were hyperplastic). Optical diagnosis accurately diagnosed 186 of 198 adenomas (sensitivity 0.94; 95% CI 0.90-0.97) and 55 of 62 hyperplastic polyps (specificity 0.89; 0.78-0.95), with an overall accuracy of 241 of 260 (0.93, 0.89-0.96) for polyp characterisation. Using optical diagnosis alone, 82 of 130 patients could be given a surveillance interval immediately after colonoscopy, and the same interval was found after formal histopathology in 80 patients (98%) using British guidelines and in 78 patients (95%) using US multisociety guidelines. INTERPRETATION: For polyps less than 10 mm in size, in-vivo optical diagnosis seems to be an acceptable strategy to assess polyp histopathology and future surveillance intervals. Dispensing with formal histopathology for most small polyps found at colonoscopy could improve the efficiency of the procedure and lead to substantial savings in time and cost. FUNDING: Leigh Family Trust, London, UK.

Novel approaches in colorectal endoscopy: what do we need biopsies for?

Magnification chromoendoscopy, narrow band imaging (NBI) and confocal endomicroscopy can all provide accurate assessment of small and diminutive colonic lesions for neoplastic change that approaches the accuracy of standard histopathology. It is likely that there will be a move to use these techniques in clinical practice for small and particularly diminutive, non-depressed lesions in the near future. Non-neoplastic lesions would be left in situ, and neoplastic lesions resected and disposed of without histopathological assessment. Histopathology would be reserved for larger lesions, indeterminate lesions or lesions where invasion was suspected. There are potentially significant cost savings and patient benefits, with a focussing of histopathological expertise on higher risk lesions, particularly in the era of bowel cancer screening. These techniques may also help target biopsies in colitis surveillance, removing the need for large numbers of random samples. However, in order to convince patients, histopathologists and those funding healthcare of the validity of this approach, further trial data will be needed, with an accreditation process for endoscopists wishing to take on this responsibility.

CT colonography and cost-effectiveness.

CT colonography (CTC) is increasingly advocated as an effective initial screening tool for colorectal cancer. Nowadays, policy-makers are increasingly interested in cost-effectiveness issues. A number of studies assessing the cost-effectiveness of CTC have been published to date. The majority of findings indicate that CTC is probably not cost-effective when colonoscopy is available, but this conclusion is sensitive to a number of key parameters. This review discusses the findings of these studies, and considers those factors which most influence final conclusions, notably intervention costs, compliance rates, effectiveness of colonoscopy, and the assumed prevalence and natural history of diminutive advanced polyps.