RESUMEN
INTRODUCTION: Comprehensive and standardized colonoscopy reports are crucial in colorectal cancer prevention, monitoring, and research. This study investigates adherence to national and international guidelines by analyzing reporting practices among 21 endoscopists in 7 German centers, with a focus on polyp reporting. METHODS: We identified and assessed German, European, American, and World Health Organization-provided statements to identify key elements in colonoscopy reporting. Board-certified gastroenterologists rated the relevance of each element and estimated their reporting frequency. Adherence to the identified report elements was evaluated for 874 polyps from 351 colonoscopy reports ranging from March 2021 to March 2022. RESULTS: We identified numerous recommendations for colonoscopy reporting. We categorized the reasoning behind those recommendations into clinical relevance, justification, and quality control and research. Although all elements were considered relevant by the surveyed gastroenterologists, discrepancies were observed in the evaluated reports. Particularly diminutive polyps or attributes which are rarely abnormal (e.g., surface integrity) respectively rarely performed (e.g., injection) were sparsely documented. Furthermore, the white light morphology of polyps was inconsistently documented using either the Paris classification or free text. In summary, the analysis of 874 reported polyps revealed heterogeneous adherence to the recommendations, with reporting frequencies ranging from 3% to 89%. CONCLUSION: The inhomogeneous report practices may result from implicit reporting practices and recommendations with varying clinical relevance. Future recommendations should clearly differentiate between clinical relevance and research and quality control or explanatory purposes. Additionally, the role of computer-assisted documentation should be further evaluated to increase report frequencies of non-pathological findings and diminutive polyps.
Asunto(s)
Pólipos del Colon , Colonoscopía , Neoplasias Colorrectales , Adhesión a Directriz , Humanos , Colonoscopía/normas , Colonoscopía/estadística & datos numéricos , Colonoscopía/métodos , Adhesión a Directriz/estadística & datos numéricos , Pólipos del Colon/patología , Pólipos del Colon/diagnóstico , Alemania , Neoplasias Colorrectales/diagnóstico , Neoplasias Colorrectales/patología , Guías de Práctica Clínica como Asunto , Pautas de la Práctica en Medicina/estadística & datos numéricos , Pautas de la Práctica en Medicina/normas , Mejoramiento de la Calidad , Gastroenterólogos/estadística & datos numéricos , Gastroenterólogos/normas , Documentación/normas , Documentación/estadística & datos numéricos , Documentación/métodosRESUMEN
BACKGROUND : Reliable documentation is essential for maintaining quality standards in endoscopy; however, in clinical practice, report quality varies. We developed an artificial intelligence (AI)-based prototype for the measurement of withdrawal and intervention times, and automatic photodocumentation. METHOD: A multiclass deep learning algorithm distinguishing different endoscopic image content was trained with 10â557 images (1300 examinations, nine centers, four processors). Consecutively, the algorithm was used to calculate withdrawal time (AI prediction) and extract relevant images. Validation was performed on 100 colonoscopy videos (five centers). The reported and AI-predicted withdrawal times were compared with video-based measurement; photodocumentation was compared for documented polypectomies. RESULTS: Video-based measurement in 100 colonoscopies revealed a median absolute difference of 2.0 minutes between the measured and reported withdrawal times, compared with 0.4 minutes for AI predictions. The original photodocumentation represented the cecum in 88 examinations compared with 98/100 examinations for the AI-generated documentation. For 39/104 polypectomies, the examiners' photographs included the instrument, compared with 68 for the AI images. Lastly, we demonstrated real-time capability (10 colonoscopies). CONCLUSION : Our AI system calculates withdrawal time, provides an image report, and is real-time ready. After further validation, the system may improve standardized reporting, while decreasing the workload created by routine documentation.
Asunto(s)
Inteligencia Artificial , Endoscopía Gastrointestinal , Humanos , Colonoscopía , Algoritmos , DocumentaciónRESUMEN
BACKGROUND: Measurement of colorectal polyp size during endoscopy is mainly performed visually. In this work, we propose a novel polyp size measurement system (Poseidon) based on artificial intelligence (AI) using the auxiliary waterjet as a measurement reference. METHODS: Visual estimation, biopsy forceps-based estimation, and Poseidon were compared using a computed tomography colonography-based silicone model with 28 polyps of defined sizes. Four experienced gastroenterologists estimated polyp sizes visually and with biopsy forceps. Furthermore, the gastroenterologists recorded images of each polyp with the waterjet in proximity for the application of Poseidon. Additionally, Poseidon's measurements of 29 colorectal polyps during routine clinical practice were compared with visual estimates. RESULTS: In the silicone model, visual estimation had the largest median percentage error of 25.1â% (95â%CI 19.1â%-30.4â%), followed by biopsy forceps-based estimation: median 20.0â% (95â%CI 14.4â%-25.6â%). Poseidon gave a significantly lower median percentage error of 7.4â% (95â%CI 5.0â%-9.4â%) compared with other methods. During routine colonoscopies, Poseidon presented a significantly lower median percentage error (7.7â%, 95â%CI 6.1â%-9.3â%) than visual estimation (22.1â%, 95â%CI 15.1â%-26.9â%). CONCLUSION: In this work, we present a novel AI-based method for measuring colorectal polyp size with significantly higher accuracy than other common sizing methods.
Asunto(s)
Pólipos del Colon , Colonografía Tomográfica Computarizada , Neoplasias Colorrectales , Humanos , Pólipos del Colon/diagnóstico por imagen , Pólipos del Colon/patología , Inteligencia Artificial , Colonoscopía/métodos , Colonografía Tomográfica Computarizada/métodos , Instrumentos Quirúrgicos , Neoplasias Colorrectales/diagnóstico por imagen , Neoplasias Colorrectales/patologíaRESUMEN
Standardized assessment of the gastroesophageal valve during endoscopy, attainable via the Hill classification, is important for clinical assessment and therapeutic decision making. The Hill classification is associated with the presence of hiatal hernia (HH), a common endoscopic finding connected to gastro-esophageal reflux disease. A novel efficient medical artificial intelligence (AI) training pipeline using active learning (AL) is designed. We identified 21,970 gastroscopic images as training data and used our AL to train a model for predicting the Hill classification and detecting HH. Performance of the AL and traditionally trained models were evaluated on an external expert-annotated image collection. The AL model achieved accuracy of 76%. A traditionally trained model with 125% more training data achieved 77% accuracy. Furthermore, the AL model achieved higher precision than the traditional one for rare classes, with 0.54 versus 0.39 (p < 0.05) for grade 3 and 0.72 versus 0.61 (p < 0.05) for grade 4. In detecting HH, the AL model achieved 94% accuracy, 0.72 precision and 0.74 recall. Our AL pipeline is more efficient than traditional methods in training AI for endoscopy.