Key points are not available for this paper at this time.
The use of Artificial Intelligence (AI) technologies in healthcare has evolved rapidly in recent years, but surgical applications of AI remain limited relative to more diagnostic disciplines. This case demonstrates novel AI technology that may be applied to the surgical management of abdominal aortic aneurysms (AAA). We present a patient with an AAA below the size-threshold for surgical repair that was clinically determined to be at high-risk of rupture alongside the results of an AI-based assessment of the biomechanical properties of the patient's aortic wall. The patient received open aortic repair. The use of Artificial Intelligence (AI) technologies in healthcare has evolved rapidly in recent years, but surgical applications of AI remain limited relative to more diagnostic disciplines. This case demonstrates novel AI technology that may be applied to the surgical management of abdominal aortic aneurysms (AAA). We present a patient with an AAA below the size-threshold for surgical repair that was clinically determined to be at high-risk of rupture alongside the results of an AI-based assessment of the biomechanical properties of the patient's aortic wall. The patient received open aortic repair. A 56-year-old man was referred to the Vascular Surgery clinic after a screening abdominal ultrasound identified an abdominal aortic aneurysm (AAA). Computed Tomography (CT) Angiogram revealed an abdominal aorta with a maximal transverse diameter of 4.6cm and an anteroposterior diameter of 3.6cm. Mild aortic margin irregularity was also noted. His past medical history was remarkable for significant smoking. Given his small body habitus (height: 1.58m, weight: 55kg, BSA: 1.55m2Forneris A. Kennard J. Ismaguilova A. Shepherd R.D. Studer D. Bromley A. et al.Linking Aortic Mechanical Properties, Gene Expression and Microstructure: A New Perspective on Regional Weakening in Abdominal Aortic Aneurysms.Front Cardiovasc Med. 2021; 8: 1-14https://doi.org/10.3389/fcvm.2021.631790Crossref Scopus (6) Google Scholar) and unusual aortic appearance on CT an urgent surgical AAA repair was recommended by the treating vascular surgeon. The patient provided informed consent for surgery with complete aneurysm excision and participation in a research study of an AI-based technology that assesses the biomechanical properties of the aorta to estimate focal aortic wall weakness. Pre-operative AI analysis was performed on the AAA using electrocardiography-gated dynamic CT as the source imaging. Simpleware ScanIP (Synopsys) was used to delineate the aortic geometry through a proprietary AI-based segmentation protocol.1Abdolmanafi A. Forneris A. Moore R.D. Di Martino E.S. Deep-learning method for fully automatic segmentation of the abdominal aortic aneurysm from computed tomography imaging.Front Cardiovasc Med. 2022; 91040053https://doi.org/10.3389/fcvm.2022.1040053Crossref Scopus (4) Google Scholar A system of 24 unique patches was created by sectioning both the lumen and outer wall perpendicular to the lumen centerline and the 3D geometry was extracted as a triangulated surface mesh (Figure 1). Unsteady computational fluid dynamic simulations were performed using a semi-implicit method for pressure linked equations algorithm for pressure-velocity coupling and a second order implicit transient formulation. Blood was assumed to be an isotropic and incompressible Newtonian fluid while the arterial wall was assumed to be rigid. Three biomarkers were then examined: intraluminal thrombus thickness, time averaged wall shear stress, and strain. These biomarkers are described in greater detail and validated in previous publications, demonstrating their individual contributions to aortic wall deterioration.2Forneris A. Kennard J. Ismaguilova A. Shepherd R.D. Studer D. Bromley A. et al.Linking Aortic Mechanical Properties, Gene Expression and Microstructure: A New Perspective on Regional Weakening in Abdominal Aortic Aneurysms.Front Cardiovasc Med. 2021; 8: 1-14https://doi.org/10.3389/fcvm.2021.631790Crossref Scopus (6) Google Scholar,3Forneris A. Beddoes R. Benovoy M. Faris P. Moore R.D. Di Martino E.S. AI powered assessment of biomarkers for growth prediction of abdominal aortic aneurysms.JVS-Vascular Science. 2023; 4: 1-8https://doi.org/10.1016/j.jvssci.2023.100119Abstract Full Text Full Text PDF Scopus (1) Google Scholar Regional aortic weakness was calculated as a composite of these biomarkers over each of the patches and represents a novel approach to aortic wall strength assessment. Results from intraluminal thrombus thickness, time averaged wall shear stress, and strain analysis are demonstrated along with the composite regional aortic weakness index in Figure 2. Increased intraluminal thrombus and low time averaged wall shear stress were found within the AAA, which is consistent with hemodynamic instability, local tissue weakening, and increased aneurysm growth – a surrogate marker for increased risk of rupture.2Forneris A. Kennard J. Ismaguilova A. Shepherd R.D. Studer D. Bromley A. et al.Linking Aortic Mechanical Properties, Gene Expression and Microstructure: A New Perspective on Regional Weakening in Abdominal Aortic Aneurysms.Front Cardiovasc Med. 2021; 8: 1-14https://doi.org/10.3389/fcvm.2021.631790Crossref Scopus (6) Google Scholar,3Forneris A. Beddoes R. Benovoy M. Faris P. Moore R.D. Di Martino E.S. AI powered assessment of biomarkers for growth prediction of abdominal aortic aneurysms.JVS-Vascular Science. 2023; 4: 1-8https://doi.org/10.1016/j.jvssci.2023.100119Abstract Full Text Full Text PDF Scopus (1) Google Scholar The impact of strain on this patient's AAA rupture risk prediction was minimal – no areas of very elevated strain were identified. High regional aortic weakness (greater than 6.5) was found within the AAA, related to the increased intraluminal thrombus and low time averaged wall shear stress. Previously published data found regional aortic weakness of 6.5 or greater was predictive of rapid relative aortic wall growth at 1 year follow-up.3Forneris A. Beddoes R. Benovoy M. Faris P. Moore R.D. Di Martino E.S. AI powered assessment of biomarkers for growth prediction of abdominal aortic aneurysms.JVS-Vascular Science. 2023; 4: 1-8https://doi.org/10.1016/j.jvssci.2023.100119Abstract Full Text Full Text PDF Scopus (1) Google Scholar The treating surgeon remained blind to these results until open aortic repair was completed to reduce risk of bias. Knowledge of the biomechanical properties of the aorta has the potential to influence surgical decision making. For example, awareness of focal weakness proximal or distal to an AAA could encourage more extensive repair. Following AI-based analysis and standard preoperative investigations, open aortic repair was completed given the patient's young age and to allow complete aneurysm excision. A transperitoneal approach via a midline incision extending from the xiphoid to the pubis was used. A Conjoint Health Research Ethics Board approved protocol allowed for modification of the standard endoaneurysmorrhaphy technique for open aneurysm repair: complete aortic excision was performed for tissue analysis following appropriate exposure and dissection. Aortic reconstruction was performed with end-to-end placement of a bifurcated Dacron graft, complete coverage of the graft with omental flap, and confirmation of distal perfusion. The patient had an uneventful post-operative recovery. The use of AI in healthcare research has expanded rapidly in the last five years, predominantly in diagnostic specialties. Less emphasis has been placed on how AI may improve surgical decision making, particularly in vascular surgery.4Javidan A.P. Li A. Lee M.H. Forbes T.L. Naji F. A systematic review and bibliometric analysis of appliucations of artificial intelligence and machine learning in vascular surgery.Ann Vasc Surg. 2022; 85: 395-405https://doi.org/10.1016/j.avsg.2022.03.019Abstract Full Text Full Text PDF PubMed Scopus (17) Google Scholar In this case, we present a tool that may improve the ability of clinicians to recognize AAA's with biomechanical properties that increase the risk of rupture.
Vergouwen et al. (2024) studied this question.