Aortic aneurysms are a primary vascular disease causing thousands of deaths every year in Europe and the US. Whereas the open surgical intervention is accompanied with a higher encumbrance during and after the operation and with a longer time of convalescence, EVAR comes with a number of post-operative complications, leading to a higher rate of re-interventions as compared to open surgery. The goal of the project EndoPredictor is to conceive and develop a highly automated software system for the prediction of post-operative complications (inclusively “5-year survival”, re-interventions, etc.) after endovascular abdominal aortic aneurysm repair. From pre-operative CT image data we derive geometric and bio-mechanical parameters derived from simulations of blood flow with high predictive value, which we integrate in a statistical prediction system for the disease progression after EVAR. The developed algorithms shall be evaluated on CT data from 75-100 patients.

Fig. A: reconstructed stent graft surface of different follow-ups (courtesy Mattes Medical Imaging), Fig. B: Visualized arterial blood flow through an aneurysm (using MEDVIS 3D)

endopredictor01

Aortic abdominal aneurysms (AAA) are a primary vascular disease causing thousands of deaths every year [Raut et al. 2013]. In over 60% of cases this type of malformation occurs among men [Raut et al. 2013]. In 2006, endovascular aortic repair (EVAR) procedures for the first time exceeded open surgical AAA repairs in the United States [Schanzer et al. 2011].Whereas the open surgical intervention is accompanied with a higher encumbrance during and after the operation and with a longer time of convalescence, EVAR comes with a number of post-operative complications, leading to a higher rate of re-interventions (approx. 20%) as compared to open surgery (approx. 10%) [Malina 2015]. The assignment of patients to EVAR or to open surgery is usually based on elementary measurements of distances and angles (e.g., the aortic diameter at the lowest renal artery) [Schanzer et al. 2011].

The goal of the project EndoPredictor is to conceive and develop a highly automated software system for the prediction of post-operative complications (inclusively “5-year survival”, re-interventions, etc.) after endovascular abdominal aortic aneurysm repair. From pre-operative CT image data we derive geometric and bio-mechanical parameters derived from simulations of blood flow [Altnji et al. 2015, Prasad et al. 2011] with high predictive value. In addition, we calculate displacement fields of stent and aneurysm from post-operative CT data in order to estimate migration and deformation of the stent graft (covered wire frame to reconstruct the healthy vessel path) as well as the deformation of the aneurysm. All these parameters we integrate in a statistical prediction system for the disease progression after EVAR. The developed algorithms shall be evaluated on CT data from 75-100 patients. The evaluation will be based on CT scans taken before and after the intervention as well as during follow up after 1, 3 and 5 years. We collect the parameter sets computed from the CT scans in a central database. On the one hand, we perform statistical tests on the collected data for identifying the most important predictors for post-operative complications. On the other hand, a classifier shall be trained and validated which allows automated prediction for future patient data in a clinical environment.