3D reconstruction of human anatomy from cross-sectional imaging has recently gained increasing importance in several medical fields thus designating the 3D bones reconstruction accuracy, critical for the success of the whole surgical intervention. The 3D anatomic model quality depends on the quality of the reconstructed image, on the quality of the images segmentation step and on the error introduced by the iso-surface triangulation algorithm. The influence of image processing procedures and relative parametrization has been largely studied in the scientific literature; however, the analysis of the direct impact of the quality of the reconstructed medical images is still lacking. In this paper, a comparative study on the influence of both image reconstruction al-gorithm (standard and iterative) and applied kernel is reported. Research was per-formed on the 3D reconstruction of a pig tibia, by using Philips Brilliance 64 CT scanner. At the stage of scanning and at the stage of 3D reconstruction, the same procedures were followed, while only image reconstruction algorithm and kernel were changed. The influence of such selection on the accuracy of bone geometry was assessed by comparing it against the 3D model obtained with a professional 3D scanner. Results show an average error in reconstructing the geometry of around 0.1 mm with a vari-ance of 0.08 mm. The presented study highlights new opportunities to control the de-viations on the geometry accuracy of the bones structures at the stage of cross sec-tional imaging generation.

3D reconstruction of human anatomy from cross-sectional imaging has recently gained increasing importance in several medical fields thus designating the 3D bones reconstruction accuracy, critical for the success of the whole surgical intervention. The 3D anatomic model quality depends on the quality of the reconstructed image, on the quality of the images segmentation step and on the error introduced by the iso-surface triangulation algorithm. The influence of image processing procedures and relative parametrization has been largely studied in the scientific literature; however, the analysis of the direct impact of the quality of the reconstructed medical images is still lacking. In this paper, a comparative study on the influence of both image reconstruction algorithm (standard and iterative) and applied kernel is reported. Research was performed on the 3D reconstruction of a pig tibia, by using Philips Brilliance 64 CT scanner. At the stage of scanning and at the stage of 3D reconstruction, the same procedures were followed, while only image reconstruction algorithm and kernel were changed. The influence of such selection on the accuracy of bone geometry was assessed by comparing it against the 3D model obtained with a professional 3D scanner. Results show an average error in reconstructing the geometry of around 0.1 mm with a variance of 0.08 mm. The presented study highlights new opportunities to control the deviations on the geometry accuracy of the bones structures at the stage of cross sectional imaging generation.

Accuracy assessment of CT-based 3D bone surface reconstruction / Luca Puggelli, Francesca Uccheddu, Yary Volpe, Rocco Furferi, Daniele Di Feo. - ELETTRONICO. - (2019), pp. 487-496. [10.1007/978-3-030-12346-8_47]

Accuracy assessment of CT-based 3D bone surface reconstruction

Luca Puggelli;Francesca Uccheddu;Yary Volpe;Rocco Furferi;Daniele Di Feo
2019

Abstract

3D reconstruction of human anatomy from cross-sectional imaging has recently gained increasing importance in several medical fields thus designating the 3D bones reconstruction accuracy, critical for the success of the whole surgical intervention. The 3D anatomic model quality depends on the quality of the reconstructed image, on the quality of the images segmentation step and on the error introduced by the iso-surface triangulation algorithm. The influence of image processing procedures and relative parametrization has been largely studied in the scientific literature; however, the analysis of the direct impact of the quality of the reconstructed medical images is still lacking. In this paper, a comparative study on the influence of both image reconstruction algorithm (standard and iterative) and applied kernel is reported. Research was performed on the 3D reconstruction of a pig tibia, by using Philips Brilliance 64 CT scanner. At the stage of scanning and at the stage of 3D reconstruction, the same procedures were followed, while only image reconstruction algorithm and kernel were changed. The influence of such selection on the accuracy of bone geometry was assessed by comparing it against the 3D model obtained with a professional 3D scanner. Results show an average error in reconstructing the geometry of around 0.1 mm with a variance of 0.08 mm. The presented study highlights new opportunities to control the deviations on the geometry accuracy of the bones structures at the stage of cross sectional imaging generation.
2019
Advances on Mechanics, Design Engineering and Manufacturing II
487
496
3D reconstruction of human anatomy from cross-sectional imaging has recently gained increasing importance in several medical fields thus designating the 3D bones reconstruction accuracy, critical for the success of the whole surgical intervention. The 3D anatomic model quality depends on the quality of the reconstructed image, on the quality of the images segmentation step and on the error introduced by the iso-surface triangulation algorithm. The influence of image processing procedures and relative parametrization has been largely studied in the scientific literature; however, the analysis of the direct impact of the quality of the reconstructed medical images is still lacking. In this paper, a comparative study on the influence of both image reconstruction al-gorithm (standard and iterative) and applied kernel is reported. Research was per-formed on the 3D reconstruction of a pig tibia, by using Philips Brilliance 64 CT scanner. At the stage of scanning and at the stage of 3D reconstruction, the same procedures were followed, while only image reconstruction algorithm and kernel were changed. The influence of such selection on the accuracy of bone geometry was assessed by comparing it against the 3D model obtained with a professional 3D scanner. Results show an average error in reconstructing the geometry of around 0.1 mm with a vari-ance of 0.08 mm. The presented study highlights new opportunities to control the de-viations on the geometry accuracy of the bones structures at the stage of cross sec-tional imaging generation.
Luca Puggelli, Francesca Uccheddu, Yary Volpe, Rocco Furferi, Daniele Di Feo
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificatore per citare o creare un link a questa risorsa: https://hdl.handle.net/2158/1128875
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 7
  • ???jsp.display-item.citation.isi??? ND
social impact