Entezari P, Kino A, Honarmand AR, Galizia MS, Yang Y, Collins J, Yaghmai V, Carr JC. Analysis of the thoracic aorta using a semi-automated post processing tool.
Eur J Radiol 2013;
82:1558-64. [PMID:
23680155 DOI:
10.1016/j.ejrad.2013.03.024]
[Citation(s) in RCA: 14] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/03/2013] [Revised: 03/21/2013] [Accepted: 03/24/2013] [Indexed: 11/26/2022]
Abstract
OBJECTIVE
To evaluates a semi-automated method for Thoracic Aortic Aneurysm (TAA) measurement using ECG-gated Dual Source CT Angiogram (DSCTA).
METHODS
This retrospective HIPAA compliant study was approved by our IRB. Transaxial maximum diameters of outer wall to outer wall were studied in fifty patients at seven anatomic locations of the thoracic aorta: annulus, sinus, sinotubular junction (STJ), mid ascending aorta (MAA) at the level of right pulmonary artery, proximal aortic arch (PROX) immediately proximal to innominate artery, distal aortic arch (DIST) immediately distal to left subclavian artery, and descending aorta (DESC) at the level of diaphragm. Measurements were performed using a manual method and semi-automated software. All readers repeated their measurements. Inter-method, intra-observer and inter-observer agreements were evaluated according to intraclass correlation coefficient (ICC) and Bland-Altman plot. The number of cases with manual contouring or center line adjustment for the semi-automated method and also the post-processing time for each method were recorded.
RESULTS
The mean difference between semi-automated and manual methods was less than 1.3mm at all seven points. Strong inter-method, inter-observer and intra-observer agreement was recorded at all levels (ICC ≥ 0.9). The maximum rate of manual adjustment of center line and contour was at the level of annulus. The average time for manual post-processing of the aorta was 19 ± 0.3 min, while it took 8.26 ± 2.1 min to do the measurements with the semi-automated tool (Vitrea version 6.0.0.1 software). The center line was edited manually at all levels, with most corrections at the level of annulus (60%), while the contour was adjusted at all levels with highest and lowest number of corrections at the levels of annulus and DESC (75% and 0.07% of the cases), respectively.
CONCLUSION
Compared to the commonly used manual method, semi-automated measurement of vessel dimensions is feasible in the thoracic aorta with the advantage of reduced post-processing time.
Collapse