Article Text


PTU-021 3D Reconstruction in Capsule Endoscopy; a Feasibility Study
  1. A Koulaouzidis1,
  2. A Karargyris2,
  3. E Rondonotti3,
  4. J N Plevris1
  1. 1Centre for Liver & Digestive Disorders, Royal Infirmary of Edinburgh, Edinburgh, UK
  2. 2National Institutes of Health, National Library of Medicine, Bethesda, United States
  3. 3Gastroenterology Unit, Ospedale Valduce, Como, Italy


Introduction Three-dimensional (3D) imaging in small-bowel capsule endoscopy (SBCE) is not currently feasible due to hardware limitations. However, there has been increasing use of software algorithms to enable 3D reconstruction in this setting. The aim of this study was to explore a) whether innovative software offers accurate 3D reconstruction of monocular images and b) if its application leads to enhanced lesion visualisation in SBCE.

Methods Feasibility study; first, a phantom was designed to test the accuracy of 3D reconstruction by comparing images of red, yellow and white phantom models to their 2D counterparts. Thereafter, a total of 192 SBCE images (84 PillCam®/108 MiroCam®; vascular: n = 50; inflammatory: n = 73; protruding structures/lesions: n = 69) were reviewed with the aid of a purpose-built 3D reconstruction software. Seven endoscopists rated visualisation improved or non-improved (compared to original 2D images) (Figure 1). Finally, the following sub-group analyses were performed: type of finding (vascular vs inflammatory vs protruding), colour of finding and SBCE equipment used (PillCam® vs Mirocam®).

Results Overall, phantom experiments showed that the 3D reconstruction software was accurate in predicting the protruding or non-protruding nature for 90% of red, 70% of yellow and 45% of white phantom models. Furthermore, it offered enhanced visualisation for 56% of vascular, 23% of inflammatory and < 10% of protruding structures/lesions (P = 0.007, 0.172 and 0.008, respectively). Lastly,when the images were categorised according to the predominant colour of the lesion, 3D software application enhanced 29/54 (53.7%) of red, 12/55 (21.8%) of white, 5/29 (17.3%) of red+white and 5/54 cases (9.2%) of lesions with colour similar to that of the surrounding mucosa, P < 0.0001.

Conclusion Application of a 3D reconstruction software in SBCE leads to image enhancement for a significant proportion (56%) of vascular, but less so for inflammatory and protruding lesions. Its integration, as adjunct tool, in CE reviewing software is desirable.

Disclosure of Interest A. Koulaouzidis Grant/Research Support from: Research support from Given Imanging Ltd, A. Karargyris: None Declared, E. Rondonotti: None Declared, J. Plevris: None Declared

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.