Automatic stent detection in intravascular OCT images using bagged decision trees

Lu, Hong; Gargesha, Madhusudhana; Wang, Zhao; Chamié, Daniel; Attizzani, Guilherme F.; Kanaya, Tomoaki; Ray, Soumya; Costa, Marco A.; Rollins, Andrew M.; Bezerra, Hiram G.; Wilson, David L.

doi:10.1364/boe.3.002809

Cited by 54 publications

(53 citation statements)

References 25 publications

(36 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several groups have developed computer algorithms tailored to coronary intravascular OCT images to perform tissue characterization,35 lumen segmentation,36 and identification of stent struts for detection of neointimal hyperplasia,37 38 quantification of in-stent restenosis,39 40 and measurement of vessel wall apposition 41–43. Although materials used to devise coronary and cerebrovascular stents may be similar, FDs are designed with more wires which have a strut thickness in the order of 30 µm compared with the 80–140 µm diameter strands typically used in coronary devices.…”

Section: Discussionmentioning

confidence: 99%

Grading of Regional Apposition after Flow-Diverter Treatment (GRAFT): a comparative evaluation of VasoCT and intravascular OCT

Marel

Gounis

Weaver

et al. 2015

J NeuroIntervent Surg

View full text Add to dashboard Cite

BACKGROUND: Poor vessel wall apposition of flow diverter (FD) stents poses risks for stroke-related complications when treating intracranial aneurysms, necessitating long-term surveillance imaging. To facilitate quantitative evaluation of deployed devices, a novel algorithm is presented that generates intuitive two-dimensional representations of wall apposition from either high-resolution contrast-enhanced cone-beam CT (VasoCT) or intravascular optical coherence tomography (OCT) images. METHODS: VasoCT and OCT images were obtained after FD implant (n=8 aneurysms) in an experimental sidewall aneurysm model in canines. Surface models of the vessel wall and FD device were extracted, and the distance between them was presented on a two-dimensional flattened map. Maps and cross-sections at potential locations of malapposition detected on VasoCT-based maps were compared. The performance of OCT-based apposition detection was evaluated on manually labeled cross-sections using logistic regression against a thresholded ( \u3e /=0.25 mm) apposition measure. RESULTS: VasoCT and OCT acquisitions yielded similar Grading of Regional Apposition after Flow-Diverter Treatment (GRAFT) apposition maps. GRAFT maps from VasoCT highlighted 16 potential locations of malapposition, of which two were found to represent malapposed device struts. Logistic regression showed that OCT could detect malapposition with a sensitivity of 98% and a specificity of 81%. CONCLUSIONS: GRAFT delivered quantitative and visually convenient representations of potential FD malapposition and occasional acute thrombus formation. A powerful combination for future neuroendovascular applications is foreseen with the superior resolution delivered by intravascular OCT

show abstract

Section: Discussionmentioning

confidence: 99%

Grading of Regional Apposition after Flow-Diverter Treatment (GRAFT): a comparative evaluation of VasoCT and intravascular OCT

Marel

Gounis

Weaver

et al. 2015

J NeuroIntervent Surg

View full text Add to dashboard Cite

show abstract

“…9 We have previously reported on IVOCT image analysis relating to segmentation, quantification, and visualization of plaques, stents, and other vessel wall components. [15][16][17][18][19][20][21][22][23][24] We and others are using IVOCT to assess viability of new coronary artery stent designs [18][19][20][21][22][23][24][25] and quantitative evaluation of atherosclerotic plaques. [15][16][17] The goal is automated classification and segmentation of plaque types from clinical IVOCT pullbacks.…”

Section: Introductionmentioning

confidence: 99%

Parameter estimation of atherosclerotic tissue optical properties from three-dimensional intravascular optical coherence tomography

et al. 2015

Self Cite

View full text Add to dashboard Cite

Abstract. We developed robust, three-dimensional methods, as opposed to traditional A-line analysis, for estimating the optical properties of calcified, fibrotic, and lipid atherosclerotic plaques from in vivo coronary artery intravascular optical coherence tomography clinical pullbacks. We estimated attenuation μ t and backscattered intensity I 0 from small volumes of interest annotated by experts in 35 pullbacks. Some results were as follows: noise reduction filtering was desirable, parallel line (PL) methods outperformed individual line methods, root mean square error was the best goodness-of-fit, and α-trimmed PL (α-T-PL) was the best overall method. Estimates of μ t were calcified (3.84 AE 0.95 mm −1 ), fibrotic (2.15 AE 1.08 mm −1 ), and lipid (9.99 AE 2.37 mm −1 ), similar to those in the literature, and tissue classification from optical properties alone was promising. © The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.

show abstract

“…The lumen segmentation method [16, 17] is in general very robust in images of varying quality and has been used in an earlier report for stent detection [10]. …”

Section: Discussionmentioning

confidence: 99%

“…However, the validation data size is small (4 pullbacks). Lu et al [10] applied bagging decision trees as the classifier on an initial screen of candidate struts, and achieved promising results in a moderately sized validation set. Such classification-based methods can take advantage of human expert knowledge, and can easily combine multiple features for decision-making, and are therefore potentially more robust.…”

Section: Introductionmentioning

confidence: 99%

3-D Stent Detection in Intravascular OCT Using a Bayesian Network and Graph Search

Wang

Jenkins

Linderman

et al. 2015

IEEE Trans. Med. Imaging

Self Cite

View full text Add to dashboard Cite

Worldwide, many hundreds of thousands of stents are implanted each year to revascularize occlusions in coronary arteries. Intravascular optical coherence tomography (OCT) is an important emerging imaging technique, which has the resolution and contrast necessary to quantitatively analyze stent deployment and tissue coverage following stent implantation. Automation is needed, as current, it takes up to 16 hours to manually analyze hundreds of images and thousands of stent struts from a single pullback. For automated strut detection, we used image formation physics and machine learning via a Bayesian network, and 3-D knowledge of stent structure via graph search. Graph search was done on en face projections using minimum spanning tree algorithms. Depths of all struts in a pullback were simultaneously determined using graph cut. To assess the method, we employed the largest validation data set used so far, involving more than 8,000 clinical images from 103 pullbacks from 72 patients. Automated strut detection achieved a 0.91±0.04 recall, and 0.84±0.08 precision. Performance was robust in images of varying quality. This method can improve the workflow for analysis of stent clinical trial data, and can potentially be used in the clinic to facilitate real-time stent analysis and visualization, aiding stent implantation.

show abstract

Automatic stent detection in intravascular OCT images using bagged decision trees

Cited by 54 publications

References 25 publications

Grading of Regional Apposition after Flow-Diverter Treatment (GRAFT): a comparative evaluation of VasoCT and intravascular OCT

Grading of Regional Apposition after Flow-Diverter Treatment (GRAFT): a comparative evaluation of VasoCT and intravascular OCT

Parameter estimation of atherosclerotic tissue optical properties from three-dimensional intravascular optical coherence tomography

3-D Stent Detection in Intravascular OCT Using a Bayesian Network and Graph Search

Contact Info

Product

Resources

About