Motivation The degeneration of the vascular wall tissue induces a - PDF document

Lesson 12b University of Bergamo Engineering and Management for Health FOR CHRONIC DISEASES MEDICAL SUPPORT SYSTEMS LESSON 12B Aortic stiffness: reconstruction of the compliance profile from TAC images acquired over the cardiac cycle. Ettore Lanzarone April 15, 2020 Motivation The degeneration of the vascular wall tissue induces a change of the arterial stiffness, i.e., the capability of the vessel to distend under the pulsatile hemodynamic load. In the literature, the aortic stiffness is usually computed following a simple approach, in which only the maximum and the minimum values of both arterial diameter and blood pressure over the cardiac cycle are considered. However, this is not directly related to the stiffness. In these two lessons, we will develop an approach to estimate the stiffness, and its spatial variation, of a given arterial region exploiting: • patient-specific geometrical data derived from computed tomography angiography (CTA); • pressure waveforms generated using a lumped parameter model of the arterial circulation. The arterial stiffness is computed linking the kinematic information derived from CTA with pressure waveforms. The final objective is to create a decision support tool non-invasive diagnosis of arterial (aortic) diseases. 1

Lesson 12b Structure The proposed methodology includes: 1. medical imaging analysis; 2. generation of aortic blood pressure waveforms; 3. estimation of the aortic stiffness. Medical imaging analysis The aim is to acquire the aortic diameter evolution along with the cardiac cycle. Thus, the information is anatomo-functional and not only anatomic. It requires several images and the comparison between them. In a formal way, the cardiac cycle T is discretized considering I +1 equally spaced time instants t i with: • I = 0, …, I t i  [0, T ] • t i = i  T / I and • t 0 = 0 and t I = T Due to the cyclic behavior, the image at t I is actually not acquired (the image at t 0 is used twice: at the beginning and the ending of the cardiac cycle) Thus I images are acquired; they are CTA images. 2

Lesson 12b Medical imaging analysis Computed Tomography Angiography (CTA) is a computed tomography technique used to visualize arterial and venous vessels throughout the body. Using a contrast injected into the blood vessels, images highlight the volume occupied by blood. Thus, they can be used to visualize the vessels of the heart, the aorta and other large blood vessels, the lungs, and the kidneys. CTA is typically used to search for blockages, aneurysms (dilations of walls), dissections (tearing of walls), and stenosis (narrowing of vessel). Medical imaging analysis The patient receives an intravenous injection of contrast and then the heart or the vessel is scanned using a high speed CT scanner. The contrast material is radiodense, which causes it to light up brightly within the blood vessels of interest. This method displays the anatomical detail of blood vessels more precisely than MRI or ultrasound. After the scan is completed the images are post-processed to better visualize the vessels and can even be created in the 3D images. 3

Lesson 12b Medical imaging analysis In our problem, each acquired CTA image is used to reconstruct a 3D model of the considered aortic segment through an image segmentation processing. The adopted imaging analysis develops in the following three steps: 1) acquisition of patient-specific medical images; 2) segmentation and anatomical reconstruction of the lumen profile; 3) virtual slicing of the 3-D reconstruction to assess cross-sectional contours at different sites. ACQUISITION Image are obtained via 4D electrocardiography (ECG)-gated CT. The ECG technique allows synchronizing a series of CT scans with the cardiac cycle through the ECG signal, to obtaining different snapshots of the vascular district as function of time (commonly expressed as percentage of the R-R interval). In our data, the 4D CT scans are acquired every 5% of the R–R interval ( 20 CTA images + last image at 100% assumed to be equal to the initial one at 0% ). Medical imaging analysis RECONSTRUCTION Medical images used are in DICOM format. In our data, the anatomical reconstruction of the descending aorta lumen profile is performed, for each CTA image, through a semiautomatic segmentation process [1], using the open source software ITK-Snap [2] based on a 3-D active contour segmentation method [3]. 1. F. Auricchio, M. Conti, S. Marconi, A. Reali, J. L. Tolenaar, and S. Trimarchi, “Patient-specific aortic endografting simulation: From diagnosis to prediction,” Comput. Biol. Med. , vol. 43, pp. 386–394, 2013. 2. www.itksnap.org 3. P. A. Yushkevich, J. Piven, H. Hazlett, R. Smith, J. G. S. Ho, and G. Gerig, “User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability,” Neuroimage , vol. 31, pp. 1116–1128, 2006. 4

Lesson 12b Medical imaging analysis SLICING The slicing procedure is performed through the following steps: 1. definition of the aortic centerline; 2. definition of n cutting planes normal to the centerline and equally spaced along the centerline; 3. detection of the cross-sectional contour points in each plane and spline interpolation; 4. 4) calculation of the center of mass for each cross-sectional contour and computation of the mean radius as the mean value of distances between the center of mass and contour points. In our data, silicing is implemented through a python-script exploring and combining modules of VTK library ( www.vtk.org ) and of VMTK library ( www.vmtk.org ). It is worth noting that the same cutting planes are usually kept for all images. Such an approximation is reasonable only under the working hypothesis that the movement of the considered tract of the aorta is negligible (descending tract of the thoracic aorta, which is bounded by the surrounding tissue). Medical imaging analysis Image processing will be addressed in one of the Labs you will follow. Here we focus on the decision support tool. 5

Lesson 12b Medical imaging analysis In this study, we consider the case of a 74-year-old female patient, who presents an asymptomatic 5.5-cm pseudo-aneurysm at the level of the distal anastomosis (eight years after ascending aortic repair for aneurysm) and whose medical history includes hypertension and atrial fibrillation. Because the patient declined a new sternotomy and the anatomy of the lesion was suitable, endovascular exclusion of the pseudo-aneurysm was planned using a custom-made stent graft (Bolton Medical, Inc., Sunrise, FL, USA). The core of the graft scaffold is composed of three self-expanding nitinol rings, having a 0.5 mm thickness, sutured on the polyester vascular fabric; a slender nitinol ring is sutured inside the fabric tube at each end. The proximal landing zone is composed of a surgical graft, 30 mm in diameter, which shows an elliptic shape with a maximum diameter of 37 mm. For such a reason, the part of ascending aorta targeted by the endovascular treatment is neglected in this study, so that the vascular region of our interest reduces to part of descending aorta. Medical imaging analysis The first figure shows the vascular region of our interest (i.e., descending aorta between the left subclavian and the diaphragm). The second figure shows different views of one of the 4D CTA scans: images highlight the region of interest clearly showing the contrasted descending aortic lumen. 6

Lesson 12b Medical imaging analysis The third figure shows the results of segmentation. The fourth figure shows the 3D reconstruction of the descending aorta lumen for one of the images, virtually sliced in eight sections. Medical imaging analysis Image processing will be addressed in one of the Labs you will follow. Here we focus on the decision support tool. Thus, I provide you the radii measured on the considered patient. 7

Lesson 12b Generation of aortic pressure waveforms The absence of noninvasive methods for directly measuring pressure waveforms in the aorta motivates their generation by means of a mathematical model. We refer to the lumped parameter model of the arterial tree already considered in one of the previous examples. The original model has been modified the existing lumped parameter model to increase the number segments in the observed aortic region. Hence, the same tract is modeled by a higher number of segments in series, each one with a reduced length. The aim is to obtain as many segments in the observed aortic piece as the number of the sections, to get an univocal correspondence between the segments of lumped parameter model and the considered sections. Generation of aortic pressure waveforms Lumped parameter model of the circulation: division of the segments to exactly match the position of the 8 sections. 8