Automated multi-atlas segmentation of cardiac 4D flow MRI

Bustamante, Mariana; Gupta, Vishu; Forsberg, Daniel; Carlhäll, Carl‐Johan; Engvall, Jan; Ebbers, Tino

doi:10.1016/j.media.2018.08.003

Cited by 38 publications

(32 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ref. 44) may increase accuracy in heart volume/mass estimation, even as investigators must remain cautious when comparing MRI with autopsy-derived measurements.…”

Section: Discussionmentioning

confidence: 99%

Metabolic rate of major organs and tissues in young adult South Asian women

et al. 2018

View full text Add to dashboard Cite

Background/Objectives Major organ- and tissue-specific metabolic rate (K i ) values were initially estimated using in vivo methods, and values reported by Elia 1 were subsequently supported by statistical analysis. However, the majority of work to date on this topic has addressed individuals of European descent, whereas population variability in resting energy metabolism has been reported. We aimed to estimate K i values in South Asian females. Subjects/Methods This cross-sectional study recruited 70 healthy young women of South Asian ancestry. Brain and organs were measured using magnetic resonance imaging, skeletal muscle mass by dual-energy X-ray absorptiometry, fat mass by the 4-component model, and whole-body resting energy expenditure by indirect calorimetry. Organ and tissue K i values were estimated indirectly using regression analysis through the origin. Preliminary analysis suggested overestimation of heart mass, hence the modeling was repeated with a literature-based 22.5% heart mass reduction. Results The pattern of derived K i values across organs and tissues matched that previously estimated in vivo , but the values were systematically lower. However, adjusting for the overestimation of heart mass markedly improved the agreement. Conclusions Our results support variability in K i values among organs and tissues, where some are more metabolically ‘expensive’ than others. Initial findings suggesting lower organ/tissue K i values in South Asian women were likely influenced by heart mass estimation bias. The question of potential ethnic variability in organ- and tissue-specific energy metabolism requires further investigation.

show abstract

“…ref. 44) may increase accuracy in heart volume/mass estimation, even as investigators must remain cautious when comparing MRI with autopsy-derived measurements.…”

Section: Discussionmentioning

confidence: 99%

Metabolic rate of major organs and tissues in young adult South Asian women

et al. 2018

View full text Add to dashboard Cite

show abstract

“…Reference [30] proposed a dual-pathway CNN for brain lesion segmentation. Reference [5] performed four-dimensional segmentation from cardiac 4D flow MRIs.…”

Section: Related Workmentioning

confidence: 99%

MaligNet: Semisupervised Learning for Bone Lesion Instance Segmentation Using Bone Scintigraphy

et al. 2020

View full text Add to dashboard Cite

One challenge in applying deep learning to medical imaging is the lack of labeled data. Although large amounts of clinical data are available, acquiring labeled image data is difficult, especially for bone scintigraphy (i.e., 2D bone imaging) images. Bone scintigraphy images are generally noisy, and ground-truth or gold standard information from surgical or pathological reports may not be available. We propose a novel neural network model that can segment abnormal hotspots and classify bone cancer metastases in the chest area in a semisupervised manner. Our proposed model, called MaligNet, is an instance segmentation model that incorporates ladder networks to harness both labeled and unlabeled data. Unlike deep learning segmentation models that classify each instance independently, MaligNet utilizes global information via an additional connection from the core network. To evaluate the performance of our model, we created a dataset for bone lesion instance segmentation using labeled and unlabeled example data from 544 and 9,280 patients, respectively. Our proposed model achieved mean precision, mean sensitivity, and mean F1-score of 0.852, 0.856, and 0.848, respectively, and outperformed the baseline mask region-based convolutional neural network (Mask R-CNN) by 3.92%. Further analysis showed that incorporating global information also helps the model classify specific instances that require information from other regions. On the metastasis classification task, our model achieves a sensitivity of 0.657 and a specificity of 0.857, demonstrating its great potential for automated diagnosis using bone scintigraphy in clinical practice.

show abstract

“…To extract meaningful data, the vessel or region of interest needs to be segmented or identified. Automated methods for placing flow-analysis planes [113] or segmenting the data [114] are crucial given the time-consuming nature of manual segmentation 2 and will drive large-scale analyses of these datasets. Accurate segmentations are also a prerequisite for analyses considering the vessel wall.…”

Section: Future Workmentioning

confidence: 99%

Improving Assessments of Hemodynamics and Vascular Disease

Ziegler

2019

Linköping University Medical Dissertations

View full text Add to dashboard Cite

No part of this publication may be reproduced, stored in a retrieval system, or be transmitted, in any form or by any means, electronic, mechanic, photocopying, recording, or otherwise, without prior permission of the author.Cover: Stylized streamline visualization of blood flow through an abdominal aortic aneurysm.We will never be here again. AbstractBlood vessels are more than simple pipes, passively enabling blood to pass through them. Their form and function are dynamic, changing with both aging and disease. This process involves a feedback loop wherein changes to the shape of a blood vessel affect the hemodynamics, causing yet more structural adaptation. This feedback loop is driven in part by the hemodynamic forces generated by the blood flow, and the distribution and strength of these forces appear to play a role in the initiation, progression, severity, and the outcome of vascular diseases.Magnetic Resonance Imaging (MRI) offers a unique platform for investigating both the form and function of the vascular system. The form of the vascular system can be examined using MR-based angiography, to generate detailed geometric analyses, or through quantitative techniques for measuring the composition of the vessel wall and atherosclerotic plaques. To complement these analyses, 4D Flow MRI can be used to quantify the functional aspect of the vascular system, by generating a full time-resolved three-dimensional velocity field that represents the blood flow.This thesis aims to develop and evaluate new methods for assessing vascular disease using novel hemodynamic markers generated from 4D Flow MRI and quantitative MRI data towards the larger goal of a more comprehensive non-invasive examination oriented towards vascular disease. In Paper I, we developed and evaluated techniques to quantify flow stasis in abdominal aortic aneurysms to measure this under-explored aspect of aneurysmal hemodynamics. In Paper II, the distribution and intensity of turbulence in the aorta was quantified in both younger and older men to understand how aging changes this aspect of hemodynamics. A method to quantify the stresses generated by turbulence that act on the vessel wall was developed and evaluated using simulated flow data in Paper III, and in Paper V this method was utilized to examine the wall stresses of the carotid artery. The hemodynamics of vascular disease cannot be uncoupled from the anatomical changes the vessel wall undergoes, and therefore Paper IV developed and evaluated a semi-automatic method for quantifying several aspects of vessel wall composition. These developments, taken together, help generate more valuable information from imaging data, and can be pooled together with other methods to form a more comprehensive non-invasive examination for vascular disease.v While my name stands alone on the cover, this thesis was undoubtably a team effort.I feel quite lucky to have had Petter Dyverfeldt as my main supervisor and mentor throughout this work, not only because of the freedom he entrusted me with, but also for his ...

show abstract

Automated multi-atlas segmentation of cardiac 4D flow MRI

Cited by 38 publications

References 51 publications

Metabolic rate of major organs and tissues in young adult South Asian women

Metabolic rate of major organs and tissues in young adult South Asian women

MaligNet: Semisupervised Learning for Bone Lesion Instance Segmentation Using Bone Scintigraphy

Improving Assessments of Hemodynamics and Vascular Disease

Contact Info

Product

Resources

About