giulia-berto.github.io/lap-project.html at main · giulia-berto/giulia-berto.github.io · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
<!DOCTYPE HTML>
<!--
	Phantom by HTML5 UP
	html5up.net | @ajlkn
	Free for personal and commercial use under the CCA 3.0 license (html5up.net/license)
-->
<html>
	<head>
		<title>Generic - Phantom by HTML5 UP</title>
		<meta charset="utf-8" />
		<meta name="viewport" content="width=device-width, initial-scale=1, user-scalable=no" />
		<link rel="stylesheet" href="./assets/css/main.css" />
		<noscript><link rel="stylesheet" href="./assets/css/noscript.css" /></noscript>
	</head>
	<body class="is-project">
		<!-- Wrapper -->
			<div id="wrapper">

				<!-- Header -->
					<header id="header">
						<div class="inner">

							<!-- Logo
								<a href="index.html" class="logo">
									<span class="symbol"><img src="images/logo.svg" alt="" /></span><span class="title">Phantom</span>
								</a>
							-->

						</div>
					</header>

				<!-- Menu -->
					<nav id="menu">
						<h2>Menu</h2>
						<ul>
							<li><a href="index.html">Home</a></li>
							<li><a href="generic.html">Ipsum veroeros</a></li>
							<li><a href="generic.html">Tempus etiam</a></li>
							<li><a href="generic.html">Consequat dolor</a></li>
							<li><a href="elements.html">Elements</a></li>
						</ul>
					</nav>

				<!-- Main -->
					<div id="main">
						<div class="inner">
							<h1>White Matter tract segmentation</h1>
							<p> This project aimed at <b>automatically segment</b> white matter tract in the human brain by considering both anatomy and
								geometry of the tracts.</p>
							<p>Diffusion Magnetic Resonance Imaging (dMRI) can provide a virtual representation of the paths of millions
								of fibers composing the brain white matter. The aim of white matter tract segmentation is to group together fibers that share a similar pathway
								into anatomically meaningful structures, called tracts, which are of interest for several applications, such as
								surgical planning and tractometry analyses.<br>
							Current methods for automatic tract segmentation consider either only prior knowledge about the
								geometrical properties of a tract, or only its relative anatomical
								position. To overcome this limitation, we extended a state-of-the-art
								fiber-based method based on the <b>Linear Assignment Problem</b> (LAP) by taking into account
								information about both the geometry of the tracts and the underlying anatomy at the same time.</p>
							<p>The LAP method performs automatic tract segmentation by
								exploiting the concept of fiber correspondence across subjects. Given a set of previously segmented
								tracts as examples, the method aims to find the same tract in a new target subject by solving a
								<b>combinatorial optimization problem</b>. To include prior anatomical
								information within the optimization process, we proposed to linearly combine the distance matrix (i.e. the original cost matrix) with other two
								anatomically-informed matrices: the endpoint-based distance matrix and the Region of Interest (ROI)-based distance matrix.</p>
								<span class="image main"><img src="images/lap-anat-image.png" alt="" /></span>
							<p>We ran multiple experiments with multiple examples using
								the LAP method (multi-LAP) and the proposed method (multi-LAP-anat) to segment 6 large tracts and 6 small tracts. We then compared their performances through the Dice
								Similarity Coefficient (DSC) score, which measures the degree of overlap with the ground truth tract (the higher the better). The proposed multi-LAP-anat method outperformed
								the multi-LAP method for all the bundles considered. Higher improvements were obtained in particular for smaller bundles.</p>
								<span class="image main"><img src="images/bar_plot_lap.png" alt="" /></span>
							<p>Read the full paper here: <a href=https://ieeexplore.ieee.org/document/8759174>https://ieeexplore.ieee.org/document/8759174</a>
								<br>Read the poster here: <a href=https://github.com/giulia-berto/app-multi-lap-anat/blob/master/isbi2019_poster.pdf>lap-anat-poster.pdf</a>
								<br>Try the brainlife App here: <a href=https://doi.org/10.25663/brainlife.app.227>https://doi.org/10.25663/brainlife.app.227</a>
								<br>Code: <a href=https://github.com/giulia-berto/app-multi-lap-anat>app-multi-lap-anat</a></p>
								<p>Citation: <i>Bertò, G., Avesani, P., Pestilli, F., Bullock, D., Caron, B. and Olivetti, E., 2019, April. Anatomically-Informed Multiple Linear Assignment Problems
									for White Matter Bundle Segmentation. In 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019) (pp. 135-138). IEEE.</i></p>
						</div>
					</div>
			</div>

		<!-- Scripts -->
			<script src="assets/js/jquery.min.js"></script>
			<script src="assets/js/browser.min.js"></script>
			<script src="assets/js/breakpoints.min.js"></script>
			<script src="assets/js/util.js"></script>
			<script src="assets/js/main.js"></script>

	</body>
</html>