Automated analysis of angiographic images for early diagnosis, monitoring and treatment of intracranial aneurysms (J2-8173)

General information

Title
Automated analysis of angiographic images for early diagnosis, monitoring and treatment of intracranial aneurysms
Period
May 1, 2017 -- Apr 30, 2020
Range
2.06 FTE
Activity
2.06 - Engineering Sciences and Technologies / Systems and Cybernetics / Biomedical Tehnics

Abstract

Cardiovascular and circulatory diseases are the world's leading cause of disability and mortality with their impact having increased at an alarming rate of 22.6% over the past two decades. The WHO statistics from 2008 show that cerebral vessels are among the most affected with 30% of all deaths caused by cerebrovascular pathologies. One such pathology is an intracranial aneurysm (IA), which is formed when a weakened part of cerebral arterial wall bulges into a balloon-like structure. The IA may eventually rupture and lead to subarachnoid hemorrhage, a serious health condition with a high mortality rate. Rupture is fatal in about 40% of cases, of those who survive about 66% suffer from permanent neurological deficit. Rupture is still rather rare as it is estimated that 50 to 80 percent of IAs do not rupture during a lifetime, but a staggering 3.2% prevalence of unruptured IAs (1 in 30 people) still leads 500,000 people to die worldwide each year due to rupture and half are younger than 50. The estimated overall direct and indirect costs of the treatment are 138 million USD per year. Clearly, there is a huge demand for constant improvement of tools and methods for clinical management of IAs. Although treatment options like neurosurgical clipping or endovascular coiling are well established for large (dome height>10 mm) and symptomatic IAs, there is an urgent need to improve clinical management of smaller IAs. These are often asymptomatic and discovered incidentally using 3D-DSA, CTA or MRA imaging, whereas corresponding treatment risk/benefit ratio is strongly in favor of the "no treatment" option, since small IAs rupture more frequently during treatment than larger ones. Similar considerations arise in the management of treated IAs irrespective of their size, where the prognostic factors and clinical guidelines to prevent a rare, but potentially fatal recurrence or rebleeding are yet to be established. Recent studies indicate that in-vivo 3D-DSA, CTA or MRA based morphologic measurements such as aneurysm size, aspect ratio (dome height/neck width), aneurysm-to-vessel size ratio and other shape indices are important independent factors contributing to high risk of rupture. Compared to hemodynamic indices like wall shear stress and pulsatility index, the morphologic indices proved more reliable for estimating rupture risk of large aneurysms. The morphologic indices mainly focus on large saccular IAs, while they are rather unspecific for small IAs due to their gross shape description. A recent study indicated that the risk is much higher for IAs that grow over time, irrespective of the initial size. Thus, novel and better risk factors may be established from longitudinal 3D-DSA, CTA or MRA images by quantifying subtle morphologic changes of the observed IA. The main goal of the proposed project is to develop innovative methods and systems based on in-vivo imaging in order to detect and diagnose IAs and perform pre- and post- treatment assessment and follow-up using quantitative morphologic descriptors. All the theoretical, computational and translational activities will be concentrated around the following themes: 1) develop accurate and reliable modality-independent (3D-DSA, CTA or MRA) detector using advanced convolutional neural networks so as to capture small IAs as early as possible; 2) develop novel methods for vasculature segmentation and IA isolation from parent vessels, and novel, more descriptive morphologic measures; 3) develop novel multi-modality deformable registration for normalization of follow-up images, and novel morphologic measures that quantify IA growth; 4) develop standardized validation datasets using real 3D-DSA, CTA or MRA images and perfom objective and rigorous validation of novel methods, and prospectively validate them in clinical screening studies; 5) translate the developed methods and systems into clinical environment and disseminate results into relevant scientific communities. Ultimately, the success of this project will have a tremendous impact on clinical management of intracranial aneurysms.

Researchers

Phases of the project and their realization

Work Packages (WPs)
WP I
Development of automated methods for detection of intracranial aneurysms
WP II
Development of automated methods for aneurysm morphology quantification
WP III
Development of automated methods for aneurysm growth quantification
WP IV
Validation of computer-aided aneurysm detection and quantification on clinical 3D-DSA, CTA and MRA images
WP V
Translation of methods and protocols into clinical environment and dissemination of research results
2017
Implementation of existing state of the art aneurysm detection methods (WP I)
Completed
Development of improved aneurysm detection based on convolutional neural networks (WP I)
Completed
Implementation of existing state of the art methods for aneurysm isolation (WP II)
Completed
Development of improved method for aneurysm isolation (WP II)
Completed
Creation of gold standard datasets with manually annotated and isolated aneurysms (WP IV)
Completed
2018
Reimplementation and development of novel quantitative morphologic measures (WP II)
Completed
Development of a deformable registration method to extract aneurysm growth patterns (WP III)
Completed
Reimplementation and development of novel morphologic measures to quantify aneurysm growth (WP III)
Completed
Validation of aneurysm detection and isolation using the gold standard datasets (WP IV)
Completed
Validation of the aneurysm detection in a clinical screening study (WP IV)
Completed
Design and implementation of a functional computer-aided system for aneurysm detection and quantification (WP V)
Completed
Dissemination of research results in peer reviewed scientific journals and at scientific meetings (WP V)
Completed
2019
Validation of the aneurysm detection in a clinical screening study (WP IV)
Completed
Validation of the morphologic measures for monitoring growth of unruptured aneurysms (WP IV)
Completed
Validation of the morphologic measures for monitoring aneurysm growth after endovascular coiling (WP IV)
Completed
Improvement of the computer-aided system based on clinical feedback (WP V)
Completed
Dissemination of research results in peer reviewed scientific journals and at scientific meetings (WP V)
Completed
2020
Validation of the aneurysm detection in a clinical screening study (WP IV)
Completed
Validation of the morphologic measures for monitoring growth of unruptured aneurysms (WP IV)
Completed
Validation of the morphologic measures for monitoring aneurysm growth after endovascular coiling (WP IV)
Completed
Improvement of the computer-aided system based on clinical feedback (WP V)
Completed
Dissemination of research results in peer reviewed scientific journals and at scientific meetings (WP V)
Completed

Bibliographics records

1.
Žiga Bizjak, Boštjan Likar, Franjo Pernuš, Žiga Špiclin: Modality agnostic intracranial aneurysm detection through supervised vascular surface classification. arXiv:2005.14467, 2020 [arXiv:2005.14467 ]
2.
Žiga Bizjak, Boštjan Likar, Franjo Pernuš, Žiga Špiclin: Vascular surface segmentation for intracranial aneurysm isolation and quantification. 23rd International Conference on Medical Image Computing and Computer Assisted Intervention - MICCAI 2020, A.L. Martel et al. (Eds.), Oct 4-8, Lima, Peru, Lecture Notes in Computer Science 12266:128-137, 2020 [COBISS-ID:33250307 ] [doi:10.1007/978-3-030-59725-2_13 ]
3.
Hennadii Madan, Rok Berlot, Nicola J. Ray, Franjo Pernuš, Žiga Špiclin: Practical priors for Bayesian inference of latent biomarkers. IEEE Journal of Biomedical and Health Informatics, 24(2):396-406, 2020 [COBISS-ID:12721492 ] [doi:10.1109/JBHI.2019.2945077 ]
4.
Tim Jerman, Aichi Chien, Franjo Pernuš, Boštjan Likar, Žiga Špiclin: Automated cutting plane positioning for intracranial aneurysm quantification. IEEE Transactions on Biomedical Engineering, 67(2):577-587, 2019 [COBISS-ID:12543316 ] [doi:10.1109/TBME.2019.2918921 ]
5.
Žiga Bizjak, Tim Jerman, Boštjan Likar, Franjo Pernuš, Aichi Chien, Žiga Špiclin: Registration based detection and quantification of intracranial aneurysm growth. SPIE Medical Imaging 2019: Computer-Aided Diagnosis, K. Mori, H.K. Hahn (Eds.), Feb 16-21, San Diego, CA, USA, Proc. SPIE 10950:1095007, 2019 [COBISS-ID:12500564 ] [doi:10.1117/12.2512781 ]
6.
Hennadii Madan, Franjo Pernuš, Žiga Špiclin: Reference-free error estimation for multiple measurement methods. Statistical Methods in Medical Research, 28(7):2196-2209, 2018 [COBISS-ID:11948116 ] [doi:10.1177/0962280217754231 ]
7.
Uroš Mitrović, Boštjan Likar, Franjo Pernuš, Žiga Špiclin: 3D-2D registration in endovascular image-guided surgery: evaluation of state-of-the-art methods on cerebral angiograms. International Journal of Computer Assisted Radiology and Surgery, 13(2):193-202, 2018 [COBISS-ID:11878228 ] [doi:10.1007/s11548-017-1678-2 ]
8.
Timur Aksoy, Žiga Špiclin, Franjo Pernuš, Gozde Unal: Monoplane 3D-2D registration of cerebral angiograms based on multi-objective stratified optimization. Physics in Medicine and Biology, 62(24):9377-9394, 2017 [COBISS-ID:11896660 ] [doi:10.1088/1361-6560/aa9474 ]
9.
Tim Jerman, Franjo Pernuš, Boštjan Likar, Žiga Špiclin: Aneurysm detection in 3D cerebral angiograms based on intra-vascular distance mapping and convolutional neural networks. 14th IEEE International Symposium on Biomedical Imaging - ISBI 2017, G. Egan, O. Salvado (Eds.), Apr 18-21, Melbourne, Australia, 612-615, 2017 [COBISS-ID:11774036 ] [doi:10.1109/ISBI.2017.7950595 ]