Refine
Has Fulltext
- yes (21)
Is part of the Bibliography
- yes (21)
Document Type
- Journal article (20)
- Conference Proceeding (1)
Keywords
- deep learning (4)
- automation (3)
- endoscopy (3)
- fully convolutional neural networks (3)
- gastroenterology (3)
- historical document analysis (3)
- machine learning (3)
- artificial intelligence (2)
- background knowledge (2)
- data warehouse (2)
- electronic health records (2)
- information extraction (2)
- medieval manuscripts (2)
- neume notation (2)
- object detection (2)
- ontology (2)
- Acceptance Evaluation (1)
- Akzeptanz-Evaluation (1)
- Aufwandsanalyse (1)
- Authoring Tools (1)
- Automatisierte Prüfungskorrektur (1)
- Autorensystem (1)
- Blended Learning (1)
- Brüder Grimm Privatbibliothek (1)
- CADe (1)
- CaseTrain (1)
- Convolutional Neural Network (1)
- Cost Analysis (1)
- Digital Humanities (1)
- Domänenadaption (1)
- Educational Measurement (I2.399) (1)
- Entscheidungsfindung (1)
- Erkennung handschriftlicher Artefakte (1)
- Ethik (1)
- Figurenerkennung (1)
- Grimm brothers personal library (1)
- Künstliche Intelligenz (1)
- Literatur (1)
- Multiple-Choice Examination (1)
- Multiple-Choice Prüfungen (1)
- Named-Entity-Recognition (1)
- Optical Music Recognition (1)
- Patient Simulation (1)
- Problem Based Learning (1)
- Problembasiertes Lernen (1)
- Self-Evaluation Programs (I2.399.780) (1)
- Teaching (1)
- Trainingsfall (1)
- annotation (1)
- baseline detection (1)
- clinical data warehouse (1)
- clinical study (1)
- colonoscopy (1)
- convolutional neural network (1)
- deep metric learning (1)
- document analysis (1)
- eHealth (1)
- electronic data capture (1)
- fault detection (1)
- few-shot learning (1)
- glaucoma progression (1)
- handwritten artefact recognition (1)
- historical printings (1)
- hospital data (1)
- image classification (1)
- key-insight extraction (1)
- laterality (1)
- layout recognition (1)
- medical records (1)
- medication extraction (1)
- model-based diagnosis (1)
- multiple myeloma (1)
- nano-satellite (1)
- natural language processing (1)
- nycthemeral intraocular pressure (1)
- optical character recognition (1)
- optical music recognition (1)
- polyp (1)
- radiology (1)
- real world evidence (1)
- real-time (1)
- regelbasierte Nachbearbeitung (1)
- right-left comparison (1)
- rule based post processing (1)
- secondary data usage (1)
- statistics and numerical data (1)
- table extraction (1)
- table understanding (1)
- text line detection (1)
- transformer (1)
- video object detection (1)
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.