K90.0
Conditions
Interventions
Group 1: All consecutive patients with positive celiac serology on a normal gluten-containing diet undergoing upper endoscopy within a two-year period from January 2015 onwards will be enrolled. The i
Sponsors
St. Anna Kinderspital Medizinische Universität Wien Austria
Eligibility
Sex/Gender
All
Age
9 Months to 18 Years
Inclusion criteria
Inclusion criteria: Subjects All consecutive patients with positive celiac serology on a normal gluten-containing diet undergoing upper endoscopy within a two-year period from January 2015 onwards will be enrolled. Controls All patients endoscopied because of non-celiac upper abdominal symptoms will constitute the control group expected to have a normal duodenal mucosa.
Exclusion criteria
Exclusion criteria: In the new ESPGHAN criteria 2012 the diagnosis of CD is even possible without villous atrophy provided that celiac autoantibody tests are positive in case of a Marsh 2 lesion. Since the endoscopic aspect of the intestinal mucosa being affected by a Marsh 2 lesion can be expected not to be different from the normal mucosa we are planning to exclude these children exhibiting Marsh 2 lesions. We only very rarely come across Marsh 2 as a result of the histological examination; approximately 1 of 100 children newly diagnosed with CD in a 2-year-period expectedly show “only” Marsh 2 as the result of the inflammation induced by an untreated CD
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| After anonymization, the endoscopic still images and video clips of all study participants will undergo feature extraction methods with regard to an automated classification in non-celiac vs. celiac disease. For the evaluation of the diagnostic characteristics of the emerging automated assessment tools, histological report of the small intestinal mucosa biopsies will serve as the gold standard. | — |
Secondary
| Measure | Time frame |
|---|---|
| Furthermore, it will be examined if the use of the narrow band imaging (NBI) technique, a optical/digital chromoendoscopy, can further improve the diagnostic reliability of automated analysis. Various established feature extraction and classification strategies will be applied for automated differentiation between presence and absence of villous atrophy in the endoscopic video sequences. A central question in the proposed study will be whether a further improvement of diagnostic reliability of automated texture analysis methods can be achieved using NBI imagery. For example, in the context of colonoscopy it has been shown that NBI is capable of improving inter-observer agreement and can lead to improved automated diagnosis results as compared to white-light coloscopy. We will clarify if this can be achieved in the CD case as well. Additionally, it is also planned to merge features taken in white light with features extracted from NBI images of the same mucosal region to further increase the discriminative information. Eventually, we will apply virtual chromoendoscopy (as applied by the systems developed by Pentax and Fujitsu, i.e. I-Scan and FICE) and compare the results to the NBI-based scheme. Increasing video resolution to reveal more mucosal details is of course desirable due to the limited field of view of endoscopes. For this purpose, we will use several good-quality frames in a superresolution approach to generate a high resolution still image onto which “classical” feature extraction and classification can be performed. The idea of superresolution is that each of these video frames represents a slightly different perspective of a specific mucosal region. After fusing information of all these low resolution images into one single image, the mucosal region of interest will appear in much higher “virtual” resolution. Preliminary work in the area of colon polyp mucosa classificat | — |
Countries
Austria
Contacts
Public ContactAndreas Vécsei
St. Anna Kinderspital Medizinische Universität Wien Austria
Outcome results
None listed