3D surface reconstruction of coronary arteries from cardiovascular angiography to detect location of heart vessels

Complications and difficulties are common in the medical field. One of the important difficulties that a surgeon may face nowadays is that when a vessel or more vessels goes inside the surface of the heart. Coronary artery vessels naturally lie on the surface of the heart but sometimes a coronary artery vessel goes inside the heart muscle and stays there for awhile and then comes out again. In this case a surgeon will not be able to locate the artery during a bypass operation. Working on 2D projections can easily mislead the comprehension and the interpretation of the structures like in the case of stenosis quantification. Different acquisition techniques make it possible to obtain 3D models of the vessels network reconstructed from 2D angiographies. The problem is that the coronary angiograms can show the arteries through contrast dye projection but still cannot show the exact location of a coronary artery vessel. The proposed procedure to solve this problem in this thesis consists of three steps. The first step is the coronary artery trees extraction. We proposed an algorithm to extract tree vessels from cardiovascular angiography by removing the background and highlighting the coronary artery tree vessels. The second step is 3D reconstruction for coronary artery tree vessels. Since extracted vessels from step one will be in 2D which will offer little information to surgeons and is difficult to study as well, we need to reconstruct them in 3D to simplify their diagnosing and analyzing. The third step in the procedure is the surface fitting. The 3D model reconstructed from step two will offer more information about the coronary artery tree vessels but it will again be difficult to decide whether a vessel is in or out the heart’s surface. Therefore, we need to build a 3D surface out of the 3D cloud of points obtained from step two to simplify the detection of said vessels. To best confirm the location of a vessel and determine the depth of that vessel inside the heart, we added a step at the end of the research to measure the curvatures of the reconstructed surface and the maximum depth of a vessel inside that surface as well. This step will measure the depths of all curvatures and highlight the maximum. To surgeons, 1 cm (10 mm) depth of a vessel inside the heart could cause a problem during surgery; therefore, our approach would set out an alarm to warn the surgeon if that depth (10 mm) or more is present. We have tested the approach to raw of clinical data sets and the results show that our proposed approach is capable of detecting the location of vessels in about 98%. From those results we can conclude that our approach is robust and can act as a tool in surgery planning and scientific researches purposes.

Preview PDF FSKTM 2012 10R.pdf Download (901kB) | Preview

Actions (login required)

View Item