Abstract
ID: 3522405 DETECTION OF EARLY ESOPHAGEAL NEOPLASIA IN BARRETT’S ESOPHAGUS USING REAL TIME ARTIFICIAL INTELLIGENCE: A MULTICENTER EXTERNAL VIDEO VALIDATION STUDY
Gastrointestinal endoscopy, Vol.93(6 Supplement), pp.AB195-AB195
06/2021
DOI: 10.1016/j.gie.2021.03.438
Abstract
Background
The gold standard and most widely used approach for screening and surveillance of Barrett’s esophagus (BE) is esophagogastroduodenoscopy. However, the visual detection of early esophageal neoplasia (high grade dysplasia and T1 stage adenocarcinoma) in BE with white light and virtual chromoendoscopy is still often difficult. We previously designed a convolutional neural artificial intelligence algorithm to detect and localize dysplasia seen on still images. The aim of this study is to assess if this system can detect early esophageal neoplasia in BE in endoscopic video.
Methods
The AI system, a deep convolutional neural network with base architecture based on Xception with pre-trained weights on ImageNet, was trained on over 4000 unique images from 150 Barrrett’s Esophagus patients. 40 video clips (ranging from 1min to 6 mins) were used for validation from 40 unique patients. 20 patients had atleast 1 dysplastic lesion and 20 patients had non-dysplastic Barretts Esophagus. The videos, containing both white light and narrow band imaging endoscopy, were obtained from two outside institutions and therefore were completely unique to the algorithm’s training database. Dysplastic lesions in the video were identified and time stamped by two expert endoscopists and were scored on a scale of subtlety from S1 to S5 (least subtle). The algorithm was designed to conduct continuous real time assessment with generation of atleast 30 predictions per second during video flow. For a prediction to be considered a “clinical prediction” for dysplasia the algorithm had to display a bounding box over the region for a continuous 3 seconds. To assess false positive (FP) rate and true negative (TN) rate in the non-dysplastic videos, the ratio of FP frames to TN frames were compared as well as the presence of false positive clinical predictions.
Results
In the dysplastic videos, the algorithm detected 19/20 lesions (95% per lesion sensitivity). The lesion that was not detected was an S1 lesion. In 2 dysplastic videos there were two extra “lesions” identified that were considered false positive clinical predictions. These were both protuberant areas of inflammatory tissue. In the non-dysplastic videos, there were 52419 true negative frames and 1308 false positive frames and thus a 2.4% false positive rate, however there were no false positive clinical predictions. The per patient negative predictive value was 100%.
Conclusion
This external validation study shows promising results for a real-time AI algorithm demonstrating high sensitivity for dysplastic lesion detection while maintaining a low rate of false positive predictions. Strengths of this system include a true real-time analysis that does not require freezing endoscopy to generate predictions. The algorithm appears ready for prospective live real-time testing.
Details
- Title: Subtitle
- ID: 3522405 DETECTION OF EARLY ESOPHAGEAL NEOPLASIA IN BARRETT’S ESOPHAGUS USING REAL TIME ARTIFICIAL INTELLIGENCE: A MULTICENTER EXTERNAL VIDEO VALIDATION STUDY
- Creators
- Jason B. SamarasenaVani J. KondaArvind J. TrindadeKimberly R. CavaliereKenneth ChangRintaro HashimotoEfren RaelAnastasia ChahineJennifer M. KolbAlyssa Y. ChoiAndrew NinhTyler DaoJames RequaWilliam E. Karnes
- Resource Type
- Abstract
- Publication Details
- Gastrointestinal endoscopy, Vol.93(6 Supplement), pp.AB195-AB195
- DOI
- 10.1016/j.gie.2021.03.438
- ISSN
- 0016-5107
- eISSN
- 1097-6779
- Language
- English
- Date published
- 06/2021
- Academic Unit
- Internal Medicine
- Record Identifier
- 9984696766202771
Metrics
1 Record Views