Reliability of Paris Classification for superficial neoplastic gastric lesions improves with training and narrow band imaging

Ribeiro, Helena; Libânio, Diogo; Castro, Raquel; Ferreira, Aníbal; Barreiro, Pedro; Carvalho, Pedro Boal; Capela, Tiago Lima; Pimentel‐Nunes, Pedro; Santos, Cristina; Dinis-Ribeiro, Mário

doi:10.1055/a-0828-7541

Cited by 13 publications

(11 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This revealed that polypoid type I (57 %) and types IIc (37 %) and III (40 %) had a higher risk of submucosal invasion. In comparison, forms IIa and IIb (29% and 20%) showed a lower probability of [5,29].…”

Section: Paris Classificationmentioning

confidence: 97%

Automated classification of polyps using deep learning architectures and few-shot learning

Krenzer

Heil

Fitting

et al. 2022

Preprint

View full text Add to dashboard Cite

Background: Colorectal cancer is a leading cause of cancer-related deaths worldwide. The best method to prevent CRC is a colonoscopy. However, not all colon polyps have the risk of becoming cancerous. Therefore, polyps are classified using different classification systems. After the classification, further treatment and procedures are based on the classification of the polyp. Nevertheless, classification is not easy. Therefore, we suggest two novel automated classifications system assisting gastroenterologists in classifying polyps based on the NICE and Paris classification. Methods: We build two classification systems. One is classifying polyps based on their shape (Paris). The other classifies polyps based on their texture and surface patterns (NICE). A two-step process for the Paris classification is introduced: First, detecting and cropping the polyp on the image, and secondly, classifying the polyp based on the cropped area with a transformer network. For the NICE classification, we design a few-shot learning algorithm based on the Deep Metric Learning approach. The algorithm creates an embedding space for polyps, which allows classification from a few examples to account for the data scarcity of NICE annotated images in our database. Results: For the Paris classification, we achieve an accuracy of 89.35 %, surpassing all papers in the literature and establishing a new state-of-the-art and baseline accuracy for other publications on a public data set. For the NICE classification, we achieve a competitive accuracy of 81.13 % and demonstrate thereby the viability of the few-shot learning paradigm in polyp classification in data-scarce environments. Additionally, we show different ablations of the algorithms. Finally, we further elaborate on the explainability of the system by showing heat maps of the neural network explaining neural activations. Conclusion: Overall we introduce two polyp classification systems to assist gastroenterologists. We achieve state-of-the-art performance in the Paris classification and demonstrate the viability of the few-shot learning paradigm in the NICE classification, addressing the prevalent data scarcity issues faced in medical machine learning.

show abstract

Section: Paris Classificationmentioning

confidence: 97%

Automated classification of polyps using deep learning architectures and few-shot learning

Krenzer

Heil

Fitting

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Since it is difficult to increase diagnostic accuracy using a single modality Cheng et al 72 Risk scores should be used to estimate the probability of noncurative resection, and this data can be shared with patients to help them make more informed decisions. 65,73 Some authors have proposed prediction models for noncurative resection of EGC. 74,75 Kim et al 74 analyzed 1639 patients who had undergone ESD for EGCs and 272 were identified as being noncuratively treated with ESD.…”

Section: How To Decrease Adverse Events?mentioning

confidence: 99%

“…Risk scores should be used to estimate the probability of noncurative resection, and this data can be shared with patients to help them make more informed decisions 65,73 . Some authors have proposed prediction models for noncurative resection of EGC 74,75 .…”

Section: How To Improve Patient Selection?mentioning

confidence: 99%

The future of endoscopic resection for early gastric cancer

Ortigão

Libânio

Dinis-Ribeiro

2022

Journal of Surgical Oncology

View full text Add to dashboard Cite

Endoscopic resection for early gastric cancer is recommended when the risk of lymph node metastasis is negligible and should be performed through submucosal dissection due to well‐established short‐ and long‐term results. To overcome technical difficulties and decrease adverse events some techniques have been studied. This review outlines current strategies for improving patient selection and highlights innovative techniques that help minimize adverse events. Moreover, we discuss how to improve management after curative and noncurative resections.

show abstract

“…Kaise et al investigated depressed lesions using NBI and concluded that abnormality in vascular and mucosal pattern were specific for cancer but the reproducibility was only moderate (k=0.48) (89). Multiple studies have concluded NBI improves detection when compared to WLE (90,91). Meticulous examination of the mucosa pattern for irregular features is vitally important for detection of pre-malignant changes in clinical practice.…”

Section: Endoscopic Features Of Gastric Dysplasiamentioning

confidence: 99%

“…The Paris classification observer agreement is moderate to good. Training using this classification in combination with NBI improves accuracy of lesion detection and observer agreement (91). Both UK and international guidelines for the pre-malignant stomach recommend the use of high definition with image enhanced technology and where possible magnification.…”

Section: Endoscopy Assessmentmentioning

confidence: 99%

Identifying the pre-malignant stomach: from guidelines to practice

White¹,

Banks²

2022

Transl Gastroenterol Hepatol

View full text Add to dashboard Cite

Gastric adenocarcinoma develops after stepwise progression from normal mucosa through to adenocarcinoma most commonly after being triggered by Helicobacter pylori (H. pylori) infection. As disease is often diagnosed late, the prognosis for gastric adenocarcinoma is poor. Identifying pre-malignant mucosal lesions such as atrophic gastritis, intestinal metaplasia and dysplasia is one strategy adopted by clinicians to reduce cancer related mortality. Surveillance of high-risk individuals and endoscopic resection of dysplastic lesions is recommended by international and UK guidelines. The early detection and endoscopic management reduce the need for invasive surgery. The advancement of image enhanced endoscopy technology, endoscopic training, risk stratification and histological assessment has proven pivotal to the management of pre-malignant lesions. In this review we outline the development of a high-risk stomach, endoscopic assessment and review practical guidelines on identifying pre-malignant gastric mucosa.

show abstract

Reliability of Paris Classification for superficial neoplastic gastric lesions improves with training and narrow band imaging

Cited by 13 publications

References 21 publications

Automated classification of polyps using deep learning architectures and few-shot learning

Automated classification of polyps using deep learning architectures and few-shot learning

The future of endoscopic resection for early gastric cancer

Identifying the pre-malignant stomach: from guidelines to practice

Contact Info

Product

Resources

About