ID: 3526290 ARTIFICIAL INTELLIGENCE (AI) IN ENDOSCOPY - DEEP LEARNING FOR SCORING OF ULCERATIVE COLITIS DISEASE ACTIVITY UNDER MULTIPLE SCORING SYSTEMS

2021 ◽  
Vol 93 (6) ◽  
pp. AB196-AB197
Author(s):  
Michael F. Byrne ◽  
James E. East ◽  
Marietta Iacucci ◽  
Remo Panaccione ◽  
Rakesh Kalapala ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Ulcerative Colitis ◽  
Deep Learning ◽  
Disease Activity ◽  
Scoring Systems

scholarly journals DOP13 Artificial Intelligence (AI) in endoscopy - Deep learning for detection and scoring of Ulcerative Colitis (UC) disease activity under multiple scoring systems

2021 ◽  
Vol 15 (Supplement_1) ◽  
pp. S051-S052
Author(s):  
M Byrne ◽  
J East ◽  
M Iacucci ◽  
R Panaccione ◽  
R Kalapala ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Ulcerative Colitis ◽  
Deep Learning ◽  
Disease Activity ◽  
Tissue Characterization ◽  
Scoring Systems ◽  
Ground Truth ◽  
The Body ◽  
Graphic User Interface ◽  
Ground Truth Data

Abstract Background Computer vision & deep learning(DL)to assess & help with tissue characterization of disease activity in Ulcerative Colitis(UC)through Mayo Endoscopic Subscore(MES)show good results in central reading for clinical trials.UCEIS(Ulcerative Colitis Endoscopic Index of Severity)being a granular index,may be more reflective of disease activity & more primed for artificial intelligence(AI). We set out to create UC detection & scoring,in a single tool & graphic user interface(GUI),improving accuracy & precision of MES & UCEIS scores & reducing the time elapsed between video collection,quality assurance & final scoring.We apply DL models to detect & filter scorable frames,assess quality of endoscopic recordings & predict MES & UCEIS scores in videos of patients with UC Methods We leveraged>375,000frames from endoscopy cases using Olympus scopes(190&180Series).Experienced endoscopists & 9 labellers tagged~22,000(6%)images showing normal, disease state(MES orUCEIS subscores)& non-scorable frames.We separate total frames in 3 categories:training(60%),testing(20%)&validation(20%).Using a Convolutional Neural Network(CNN)Inception V3,including a biopsy & post-biopsy detector,an out-of-the-body framework & blue light algorithm.Similar architecture for detection with multiple separate units & corresponding dense layers taking CNN to provide continuous scores for 5 separate outputs:MES,aggregate UCEIS & individual components Vascular Pattern,Bleeding & Ulcers. Results Multiple metrics evaluate detection models.Overall performance has an accuracy of~88% & a similar precision & recall for all classes. MAE(distance from ground truth)& mean bias(over/under-prediction tendency)are used to assess the performance of the scoring model.Our model performs well as predicted distributions are relatively close to the labelled,ground truth data & MAE & Bias for all frames are relatively low considering the magnitude of the scoring scale. To leverage all our models,we developed a practical tool that should be used to improve efficiency & accuracy of reading & scoring process for UC at different stages of the clinical journey. Conclusion We propose a DL approach based on labelled images to automate a workflow for improving & accelerating UC disease detection & scoring using MES & UCEIS scores. Our deep learning model shows relevant feature identification for scoring disease activity in UC patients, well aligned with both scoring guidelines,performance of experts & demonstrates strong promise for generalization.Going forward, we aim to continue developing our detection & scoring tool. With our detailed workflow supported by deep learning models, we have a driving function to create a precise & potentially superhuman level AI to score disease activity

scholarly journals Training and deploying a deep learning model for endoscopic severity grading in ulcerative colitis using multicenter clinical trial data

2021 ◽  
Vol 14 ◽  
pp. 263177452199062
Author(s):  
Benjamin Gutierrez Becker ◽  
Filippo Arcadu ◽  
Andreas Thalhammer ◽  
Citlalli Gamez Serna ◽  
Owen Feehan ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Ulcerative Colitis ◽  
Clinical Trials ◽  
Deep Learning ◽  
Mayo Clinic ◽  
Automated System ◽  
Phase Iii ◽  
Plain Language ◽  
Data Set ◽  
Endoscopic Videos

Introduction: The Mayo Clinic Endoscopic Subscore is a commonly used grading system to assess the severity of ulcerative colitis. Correctly grading colonoscopies using the Mayo Clinic Endoscopic Subscore is a challenging task, with suboptimal rates of interrater and intrarater variability observed even among experienced and sufficiently trained experts. In recent years, several machine learning algorithms have been proposed in an effort to improve the standardization and reproducibility of Mayo Clinic Endoscopic Subscore grading. Methods: Here we propose an end-to-end fully automated system based on deep learning to predict a binary version of the Mayo Clinic Endoscopic Subscore directly from raw colonoscopy videos. Differently from previous studies, the proposed method mimics the assessment done in practice by a gastroenterologist, that is, traversing the whole colonoscopy video, identifying visually informative regions and computing an overall Mayo Clinic Endoscopic Subscore. The proposed deep learning–based system has been trained and deployed on raw colonoscopies using Mayo Clinic Endoscopic Subscore ground truth provided only at the colon section level, without manually selecting frames driving the severity scoring of ulcerative colitis. Results and Conclusion: Our evaluation on 1672 endoscopic videos obtained from a multisite data set obtained from the etrolizumab Phase II Eucalyptus and Phase III Hickory and Laurel clinical trials, show that our proposed methodology can grade endoscopic videos with a high degree of accuracy and robustness (Area Under the Receiver Operating Characteristic Curve = 0.84 for Mayo Clinic Endoscopic Subscore ⩾ 1, 0.85 for Mayo Clinic Endoscopic Subscore ⩾ 2 and 0.85 for Mayo Clinic Endoscopic Subscore ⩾ 3) and reduced amounts of manual annotation. Plain language summary Patient, caregiver and provider thoughts on educational materials about prescribing and medication safety Artificial intelligence can be used to automatically assess full endoscopic videos and estimate the severity of ulcerative colitis. In this work, we present an artificial intelligence algorithm for the automatic grading of ulcerative colitis in full endoscopic videos. Our artificial intelligence models were trained and evaluated on a large and diverse set of colonoscopy videos obtained from concluded clinical trials. We demonstrate not only that artificial intelligence is able to accurately grade full endoscopic videos, but also that using diverse data sets obtained from multiple sites is critical to train robust AI models that could potentially be deployed on real-world data.

scholarly journals P087 Biomarkers of elastin degradation differentiate between clinically inactive and active disease in ulcerative colitis patients

2020 ◽  
Vol 14 (Supplement_1) ◽  
pp. S178-S179
Author(s):  
M Pehrsson ◽  
V Domislović ◽  
M A Karsdal ◽  
M Brinar ◽  
A Barisic ◽  
...  
Keyword(s):  
Ulcerative Colitis ◽  
Disease Activity ◽  
Disease Severity ◽  
Matrix Protein ◽  
Scoring Systems ◽  
Cathepsin G ◽  
Tissue Remodelling ◽  
Montreal Classification ◽  
Elastin Degradation ◽  
Active Disease

Abstract Background In ulcerative colitis (UC), the state of chronic inflammation results in increased matrix metalloprotease (MMP) and serine protease activity, which effectively leads to a higher degree of intestinal tissue remodelling, including components of the extracellular matrix (ECM). One of these components is elastin a matrix protein of the interstitial matrix in the lamina propria and submucosa, providing tissue resilience and elasticity. As such, we investigated whether elastin degradation in UC patients was associated with disease activity and severity, potentially enabling patient differentiation based on elastin degradation. Methods Twenty-nine UC patients and 29 healthy donors were included in the study. Disease activity was determined according to the partial Mayo score (pMayo >1) and the Mayo Endoscopic Score (MES). Disease severity and extension was assessed using the Montreal classification. Disease severity was additionally assessed using the Trulove and Witt’s (TW) clinical score. The biomarkers of elastin degradation included: MMP-7 (ELM-7) cathepsin-G (EL-CG) and proteinase-3 (ELP-3), measured in serum by ELISA. One-way ANOVA (Kruskal–Wallis) correcting for the false discovery rate were applied for the statistical analysis. Results TW: ELP-3 levels in moderate-to-severe UC patients were significantly elevated in comparison with HD (p < 0.001). Partial Mayo: EL-CG levels in patients with active UC were significantly elevated in comparison with HD (p < 0.01), and UC patients in remission (p < 0.01). ELP-3 levels were likewise significantly elevated in active UC patients compared with HD (p < 0.001), and UC patients in remission (p < 0.01). Montreal classification: ELM-7 was significantly elevated in active UC compared with HD (p < 0.05), and UC patients in remission (p < 0.05). EL-CG were also significantly elevated in active UC compared with HD (p < 0.05), and UC patients in remission (p < 0.05). ELP-3 was significantly elevated in active UC compared with HD (p < 0.01). According to the MES score, ELP-3 levels in moderate-to-severe UC patients were significantly elevated in comparison to HD (p < 0.01). Conclusion The data presented in this study demonstrate an association between biomarkers of proteolytic elastin degradation and disease activity in UC patients especially the protease-3-derived biomarker, ELP-3, showed significant association with active UC in all the clinical scoring systems as well as the MES score. Utilising these minimally invasive elastin degradation biomarkers could serve as surrogate markers for monitoring of disease activity and potentially aid the differentiation of patients with an active disease from patients in remission or with a lower disease activity for UC.

scholarly journals P141 Scoring endoscopy in pediatric inflammatory bowel disease: a way to improve quality

2020 ◽  
Vol 14 (Supplement_1) ◽  
pp. S208-S208
Author(s):  
L Norsa ◽  
A Ferrari ◽  
S Arrigo ◽  
M Bramuzzo ◽  
M Deganello Saccomani ◽  
...  
Keyword(s):  
Ulcerative Colitis ◽  
Crohn’S Disease ◽  
Crohn's Disease ◽  
Disease Activity ◽  
Intraclass Correlation ◽  
Scoring Systems ◽  
Median Number ◽  
Kappa Statistics ◽  
Low Grade ◽  
Inflammatory Bowel

Abstract Background The aim of mucosal healing (MH) as a therapeutic target in paediatric inflammatory bowel diseases (IBD) has emphasised the role of the endoscopy. There is a great variability in evaluating mucosal lesions among different operators, especially in paediatric patients. This multicentre prospective study aims to evaluate the interobserver agreement among paediatric endoscopists in using validated endoscopic scores of IBD in children. Methods Fifteen videos of follow-up ileocolonoscopies in children with IBD (8 ulcerative colitis –UC-, 7 Crohn’s disease –CD-) were selected from 3 different referral sites in Italy. Eleven paediatric endoscopists from different centres were asked to evaluate all videos as independent and blinded readers. The scoring systems used were ulcerative colitis Endoscopic Index of Severity (UCEIS) for UC and simple endoscopic score for Crohn’s disease (SES-CD) for CD. Kappa statistics and intraclass correlation coefficients were used to measure agreement. Furthermore, an experienced adult gastroenterologist evaluated the same videos and scores them. His results were compared with paediatric endoscopists’ findings. Results The median age of the participants was 40 (interquartile range: 6) with a median experience of 12 (14) years in centres with a median number of 140 (230) of paediatric IBDs. Intercluster correlation agreement was 0.298 (95% CI: 0.13–0.55) for UC and 0.266 (0.11–0.52) for CD. When a disease activity categorisation was adopted (remission, moderate, mild and severe) Fleiss’ kappa coefficient was 0.408 (0.29–0.53) for UC and 0.552 (0.43–0.73) for CD (Figure 1). When stratified for item vascular pattern of UC was the most reliable item IC: 0.624 (0.321–0.854). The comparison between paediatric and expert gastroenterologist’s scores is shown in Figure 2. In the multivariate analysis none of the reviewer characteristic affected the readers’ errors. Conclusion This pilot multicentre study shows that there is a low level of agreement among paediatric endoscopists in evaluating children with IBDs. Agreement improved after using a disease activity categorisation, with better results for CD. Regardless to experience, all readers showed a low-grade accordance with adult gastroenterologist. According to these findings, the use of scoring systems should be implemented for all paediatric endoscopists. Future specific training programs should be considered to pursue this goal.

ID: 3527087 PRACTICAL DEEP LEARNING TOOL FOR SCORING OF ULCERATIVE COLITIS DISEASE ACTIVITY IN CENTRAL READING

2021 ◽  
Vol 93 (6) ◽  
pp. AB199
Author(s):  
Michael F. Byrne ◽  
James E. East ◽  
Marietta Iacucci ◽  
Simon P. Travis ◽  
Rakesh Kalapala ◽  
...  
Keyword(s):  
Ulcerative Colitis ◽  
Deep Learning ◽  
Disease Activity ◽  
Learning Tool

scholarly journals A simple clinical colitis activity index

Gut ◽  
1998 ◽  
Vol 43 (1) ◽  
pp. 29-32 ◽  
Author(s):  
R S Walmsley ◽  
R C S Ayres ◽  
R E Pounder ◽  
R N Allan
Keyword(s):  
Ulcerative Colitis ◽  
Disease Activity ◽  
Activity Index ◽  
Laboratory Data ◽  
Scoring Systems ◽  
Initial Assessment ◽  
Mucosal Inflammation ◽  
Clinical Criteria ◽  
Complex Index ◽  
Simple Index

Background—The appropriate medical treatment of patients with ulcerative colitis is determined largely by the severity of symptoms. Hospital assessment of the severity of disease activity includes investigation of laboratory indices and sigmoidoscopic assessment of mucosal inflammation.Aims—To develop a simplified clinical colitis activity index to aid in the initial evaluation of exacerbations of colitis.Methods—The information for development of the simple index was initially evaluated in 63 assessments of disease activity in patients with ulcerative colitis where disease activity was evaluated using the Powell-Tuck Index (which includes symptoms, physical signs, and sigmoidoscopic appearance). The new index was then further evaluated in 113 assessments in a different group of patients, by comparison with a complex index utilising clinical and laboratory data, as well as five haematological and biochemical markers of disease severity.Results—The newly devised Simple Clinical Colitis Activity Index, consisting of scores for five clinical criteria, showed a highly significant correlation with the Powell-Tuck Index (r=0.959, p<0.0001) as well as the complex index (r=0.924, p<0.0001) and all laboratory markers (p=0.0003 to p<0.0001).Conclusions—This new Simple Colitis Activity Index shows good correlation with existing more complex scoring systems and therefore could be useful in the initial assessment of patients with ulcerative colitis.

Practical Deep Learning Tool For Scoring Of Ulcerative Colitis Disease Activity In Central Reading

2021 ◽  
Author(s):  
MF Byrne ◽  
JE East ◽  
M Iacucci ◽  
SP Travis ◽  
R Kalapala ◽  
...  
Keyword(s):  
Ulcerative Colitis ◽  
Deep Learning ◽  
Disease Activity ◽  
Learning Tool
Sign in / Sign up

Export Citation Format

Share Document