-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathindex.html
More file actions
421 lines (351 loc) · 16.2 KB
/
Copy pathindex.html
File metadata and controls
421 lines (351 loc) · 16.2 KB
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
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description" content="VesselSim: learning 3D blood vessel segmentation without expert annotations">
<meta name="keywords" content="Vascular segmentation,synthetic data, test-time training">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>VesselSim</title>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/favicon.svg">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">VesselSim: learning 3D blood vessel segmentation
without expert annotations</h1>
<div class="is-size-6 publication-authors">
<span class="author-block">
<a href="https://www.linkedin.com/in/erin-rainville-01a41a306">Erin Rainville</a><sup>1</sup>,
<a href="https://www.linkedin.com/in/melissa-ananian-443a4b165">Melissa Ananian</a><sup>1</sup>,
<a href="https://www.linkedin.com/in/tristan-mirolla-2a698522a">Tristan Mirolla</a><sup>1</sup>,
</span>
<span class="author-block">
<a href="https://users.encs.concordia.ca/~hrivaz/">Hassan Rivaz</a><sup>2</sup>,
</span>
<span class="author-block">
<a href="https://yimingxiao.weebly.com/curriculum-vitae.html">Yiming Xiao </a><sup>1</sup>
</span>
</div>
<div class="is-size-6 publication-authors">
<span class="author-block"><sup>1</sup>Department of Computer Science and Software Engineering, Concordia University, Montreal, Canada,</span>
<span class="author-block"><sup>2</sup>Department of Electrical and Computer Engineering, Concordia University, Montreal, Canada</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<span class="link-block">
<a href="https://conferences.miccai.org/2026/en/default.asp" target="_blank">
<img src="./static/images/MICCAI2026-logo.png" alt="GitHub Link" width="500">
</a>
</span>
<!-- <span class="link-block">-->
<!-- <a href="[PAPER PDF LINK]" class="external-link button is-normal is-rounded is-dark">-->
<!-- <span class="icon"><i class="fas fa-file-pdf"></i></span>-->
<!-- <span>Paper</span>-->
<!-- </a>-->
<!-- </span>-->
<!-- <span class="link-block">-->
<!-- <a href="[ARXIV LINK]" class="external-link button is-normal is-rounded is-dark">-->
<!-- <span class="icon"><i class="ai ai-arxiv"></i></span>-->
<!-- <span>arXiv</span>-->
<!-- </a>-->
<!-- </span>-->
<!-- <span class="link-block">-->
<!-- <a href="[CODE LINK]" class="external-link button is-normal is-rounded is-dark">-->
<!-- <span class="icon"><i class="fab fa-github"></i></span>-->
<!-- <span>Code</span>-->
<!-- </a>-->
<!-- </span>-->
<!-- <span class="link-block">-->
<!-- <a href="[DATA LINK]" class="external-link button is-normal is-rounded is-dark">-->
<!-- <span class="icon"><i class="far fa-images"></i></span>-->
<!-- <span>Data</span>-->
<!-- </a>-->
<!-- </span>-->
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser">
<div class="container is-max-desktop has-text-centered">
<div class="hero-body">
<!-- Use either an image or a video teaser. Delete the one you do not need. -->
<img src="./static/images/dataCompare.png" alt="Project teaser figure">
<!--
<video id="teaser" autoplay muted loop playsinline height="100%">
<source src="./static/videos/teaser.mp4" type="video/mp4">
</video>
-->
<h2 class="subtitle has-text-centered">
VesselSim enables zero-shot blood vessel segmentation across modalities and organs without any real training images. The full generated data is available for use
<a href="https://osf.io/8n5wu/overview">here</a>, the generation code is available <a href="https://github.com/HealthX-Lab/VesselSim">here</a> and the
manuscript is available <a href="https://arxiv.org/abs/2605.26277">here</a>.
</h2>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
Blood vessel segmentation is a core task in medical image analysis for the care of vascular diseases and
surgical planning, yet the challenges of providing expert vascular annotations pose a major obstacle for the
progress of related deep learning techniques. To address this, we propose <kbd>VesselSim</kbd>, a two-stage
framework for universal 3D blood vessel segmentation that eliminates the need for real annotated data during
training. First, we introduce a stochastic, geometry-driven vascular simulation framework that models
recursive branching, curvature-controlled growth, and collision-aware topology, followed by
domain-randomized intensity synthesis to generate 16,500 anatomically plausible 3D angiographic volumes.
Second, a 3D U-Net is trained solely on this synthetic data. To bridge the domain gap from synthetic to
real images at inference time, we introduce a test-time adaptation strategy via a self-supervised mask
reconstruction decoder, enabling adaptation to unseen clinical scans without prior domain knowledge.
We evaluate <kbd>VesselSim</kbd> in a zero-shot setting on multiple real-world datasets spanning MR and CT
across several anatomical regions, including the brain and kidneys. Despite being trained exclusively on
synthetic data, <kbd>VesselSim</kbd> achieves performance competitive with state-of-the-art vascular
segmentation foundation models. These findings suggest that learning vessel geometry from synthetic tubular
structures is effective for robust cross-domain generalization, substantially reducing the reliance on
acquired medical imaging data and more importantly, expert annotations.
</p>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Method Overview</h2>
<div class="content has-text-justified">
<ul>
<li>
<b>Synthetic Vascular Training Data:</b>
A stochastic vessel generation framework produces diverse 3D vascular networks and domain-randomized angiographic images without requiring real annotated data.
</li>
<li>
<b>3D U-Net Vessel Segmentation:</b>
A 3D U-Net is trained entirely on synthetic vascular volumes using Dice, Cross-Entropy, and centerline-aware losses to preserve thin vessel structures and connectivity.
</li>
<li>
<b>Self-Supervised Reconstruction Branch:</b>
A masked in-painting decoder reconstructs randomly corrupted image regions, providing auxiliary self-supervision and improving domain robustness.
</li>
<li>
<b>Test-Time Adaptation (TTA):</b>
During inference, reconstruction-based self-supervision adapts encoder normalization parameters to unseen clinical scans without requiring annotations.
</li>
<li>
<b>Cross-Modality Generalization:</b>
The framework is evaluated in a zero-shot setting across CTA, MRA, and HiP-CT datasets spanning different anatomies and imaging domains.
</li>
</ul>
</div>
<div class="content has-text-centered">
<img src="./static/images/method.png" alt="Method overview figure" width="90%">
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Results</h2>
<div class="content has-text-justified">
<p>
We evaluate on two publicly available medical image repositories,
<a href="https://kaggle.com/competitions/blood-vessel-segmentation">HiP-CT</a> and
<a href="https://topcow24.grand-challenge.org/data/">TopCoW</a>. <i>HiP-CT</i> consists of three
ultra–high-resolution 3D human kidney volumes acquired using Hierarchical Phase-Contrast Tomography.
<i>TopCoW</i> includes 125 computed tomography angiographies (CTA) and 125 magnetic resonance angiographies
(MRA) of the human brain. For <i>TopCow CTA</i>, manual annotations are provided for the Circle
of Willis. Volumes are cropped to the bounding box of this region to avoid penalizing predictions outside
the territory. For <i>TopCoW MRA</i> , we use the <a href="https://i-vesseg.github.io/vesselverse/">VesselVerse</a>
extension, which expands the vessel annotations to the full brain. Thus allowing us to assess robustness
across organ systems (kidney, Circle of Willis, full brain) and across imaging contrasts (HiP-CT, CTA and MRA).
</p>
<p>
Comparison of segmentation performance across datasets.
<b>VesselSim</b> denotes our final model with
<i>CE + Dice + cbDice + Reconstruction TTA</i>.
Best results are shown in <b>bold</b>,
second best results are <u>underlined</u>.
</p>
</div>
<div style="overflow-x:auto;">
<table class="table is-striped is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="2">TopCoW CTA</th>
<th colspan="2">TopCoW MRA</th>
<th colspan="2">HiP-CT</th>
</tr>
<tr>
<th>Dice</th>
<th>clDice</th>
<th>Dice</th>
<th>clDice</th>
<th>Dice</th>
<th>clDice</th>
</tr>
</thead>
<tbody>
<tr class="table-section">
<td colspan="7"><b>Foundation Model Baselines</b></td>
</tr>
<tr>
<td>VesselFM</td>
<td><b>52.9±9.5</b></td>
<td>58.9±10.2</td>
<td>48.3±8.1</td>
<td>46.9±9.1</td>
<td><u>35.1±24.5</u></td>
<td>25.6±19.5</td>
</tr>
<tr>
<td>SAM-Med3D</td>
<td>6.6±4.4</td>
<td>9.1±6.0</td>
<td>3.1±2.6</td>
<td>2.6±2.2</td>
<td>8.0±21.6</td>
<td>3.1±8.9</td>
</tr>
<tr>
<td>UniverSeg</td>
<td>19.7±8.6</td>
<td>29.4±12.3</td>
<td>48.2±5.1</td>
<td>47.0±6.1</td>
<td>4.7±21.1</td>
<td>0.0±0.0</td>
</tr>
<tr class="table-section">
<td colspan="7"><b>Ablation Studies</b></td>
</tr>
<tr>
<td>Base UNet</td>
<td>42.7±12.2</td>
<td>60.6±12.7</td>
<td>70.1±3.9</td>
<td>73.1±5.1</td>
<td>34.3±23.0</td>
<td>23.0±42.7</td>
</tr>
<tr>
<td>+ cbDice</td>
<td>46.8±12.9</td>
<td><u>62.9±13.6</u></td>
<td><b>72.6±3.5</b></td>
<td><u>75.7±4.4</u></td>
<td>31.1±23.7</td>
<td>41.2±25.1</td>
</tr>
<tr>
<td>+ recon</td>
<td>43.1±14.1</td>
<td>60.6±13.3</td>
<td><u>71.9±3.9</u></td>
<td>74.6±5.2</td>
<td><b>36.0±22.7</b></td>
<td><u>41.3±23.1</u></td>
</tr>
<tr style="background-color:#f5e1ff;">
<td><b>VesselSim</b></td>
<td><u>48.7±11.9</u></td>
<td><b>64.0±13.0</b></td>
<td>71.8±3.3</td>
<td><b>76.2±4.1</b></td>
<td>31.6±21.5</td>
<td><b>42.3±23.8</b></td>
</tr>
<tr style="background-color:#fffacd;">
<td>+ finetuned</td>
<td>59.5±7.7</td>
<td>70.4±7.3</td>
<td>77.3±3.3</td>
<td>83.2±3.6</td>
<td>46.8±33.2</td>
<td>50.7±27.8</td>
</tr>
</tbody>
</table>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Segmentation Visualization</h2>
<div class="content has-text-justified">
<p>
Maximum intensity projections of the input scan and different model predictions.
From top to bottom: HiP-CT, TopCoW CTA, TopCoW MRA. For the predictions,
true positives are shown in blue, false positives in red,
and false negatives in green. SAM-Med3D visualizations are omitted due to consistently low segmentation
performance across datasets.
</p>
</div>
<div class="content has-text-centered">
<img src="./static/images/results.png" alt="Results figure" width="90%">
</div>
</div>
</div>
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
The VesselSim paper has been accepted into the MICCAI 2026 conference. The peer reviewed, finalized copy will be available
in October 2026. The unofficial version is currently available on <a href="https://arxiv.org/abs/2605.26277">arXiv</a>.
<pre><code> @misc{rainville2026vesselsim,
title={VesselSim: learning 3D blood vessel segmentation without expert annotations},
author={Erin Rainville and Melissa Ananian and Tristan Mirolla and Hassan Rivaz and Yiming Xiao},
year={2026},
eprint={2605.26277},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2605.26277},
}</code></pre>
</div>
</section>
<footer class="footer">
<div class="container">
<div class="columns is-centered has-text-centered">
<div class="column is-8">
<div class="content">
<p>
Website template adapted from
<a href="https://nerfies.github.io/">Nerfies</a>.
</p>
</div>
</div>
</div>
</div>
</footer>
</body>
</html>