Method details

Download PDF file of article
Automatic Coronary Arteries Stenoses Detection in 3D CT Angiography



Abstract:
Cardiovascular lesions are the world's leading cause of mortality. Early detection of these diseases using less invasive techniques provides better therapeutic outcome, as well as reduces costs and risks, compared to an interventionist approach. In this work we propose a two-step severe cardiac stenoses detection approach. First we used the geometric model of the stenosis to detect suspicious areas and then we applied a false positives (FPs) removal step, based on lesions appearance properties. The algorithm was tested, in the context of the MICCAI challenge, on a 42 3D cardiac CT angiography database provided by the organizers. For the detection, a sensitivity of 47% and a PPV of 14% is obtained as compared to QCA, while a sensitivity of 35% and a PPV of 8% is achieved as compared to CTA.

Detection confusion tables

Calc. cat.QCA (per segment)CTA (per lesion)
 TPFPFNTP+FPTP+FNTPFPFNTP+FPTP+FN
All  13  94  15  107  28  20  196  27  216  47 
The results of this method are based on the following centerlines: Rcadia team auto.

Detection (QCA per segment / CTA per lesion)

Calc. cat.QCA
Sens.
QCA
P.P.V.
CTA
Sens.
CTA
P.P.V.
Avg. rank
 %rank%rank%rank%rank 
0 (0 - 0.1)  20.0  11.0  6.7  14.0  14.3  14.0  3.7  14.0  13.2 
1 (0.1 - 10)  100.0  1.0  7.7  11.0  80.0  4.0  13.3  14.0  7.5 
2 (11 - 100)  42.9  11.0  9.4  17.0  36.4  12.0  5.6  16.0  14.0 
3 (101 - 400)  60.0  8.0  15.8  14.0  47.4  9.0  12.0  15.0  11.5 
4 (400+)  40.0  12.0  22.2  13.0  40.0  11.0  16.7  11.0  11.8 
All  46.4  13.0  12.1  16.0  42.6  12.0  9.3  15.0  14.0 
For ranking  46.4  13.0  12.1  16.0  42.6  12.0  9.3  15.0  14.0 
These results are based on 30 datasets and 19 submissions.