=Paper=
{{Paper
|id=Vol-1476/paper9
|storemode=property
|title=Phantom-based evaluation of a semi-automatic segmentation algorithm for cerebral vascular structures in 3D ultrasound angiography (3D USA)
|pdfUrl=https://ceur-ws.org/Vol-1476/Proceedings_CURAC_2011_Paper_9.pdf
|volume=Vol-1476
|dblpUrl=https://dblp.org/rec/conf/curac/ChalopinKMAML11
}}
==Phantom-based evaluation of a semi-automatic segmentation algorithm for cerebral vascular structures in 3D ultrasound angiography (3D USA) ==
https://ceur-ws.org/Vol-1476/Proceedings_CURAC_2011_Paper_9.pdf
10. CURAC-Jahrestagung, 15. - 16. September 2011, Magdeburg
Phantom-based evaluation of a semi-automatic segmentation algo-
rithm for cerebral vascular structures in 3D ultrasound angiography
(3D USA)
C. Chalopin¹, K. Krissian², A. Müns3, F. Arlt3, J. Meixensberger³, D. Lindner3
¹ Universität Leipzig, ICCAS, Leipzig, Germany
² Universidad de Las Palmas de Gran Canaria, GIMET, Las Palmas, Spain
³ Universität Leipzig, Klinik für Neurochirurgie, Leipzig, Germany
Kontakt: claire.chalopin@iccas.de
Abstract:
Intraoperative ultrasound angiography (USA) provides to the neurosurgeon real-time information about the cerebral
vascular network but is difficult to interpret due to the presence of noise and artifacts. A segmentation algorithm may
improve the visualization of data by extracting the vascular structures only. We propose to adapt and test an existing
model-based segmentation method on 3D USA data of a vascular phantom with 4 mm tube radii. The performance of
the algorithm is evaluated by comparison with a gold standard (CT data) and with manual delineations. The algorithm
generated a segmentation model whose radii values are overestimated of more than half of one mm in comparison with
the gold standard but with more realistic geometrical features than the manual delineations.
Keywords: 3D ultrasound angiography, vascular segmentation, physical phantom
1 Problem
Intraoperative ultrasound angiography (iUSA) is an imaging modality which, in neurosurgery, enables the surgeon to
visualize the real-time information of the anatomy and function of the cerebral vascular network during the intervention
[1]. With the development of new ultrasound contrast agents, contrast harmonic imaging (CHI) is becoming an emerg-
ing modality, which enables enhancing the main cerebral vascular structures in the images. However, the interpretation
of iUSA data may be complex. The image quality is reduced by the speckle, but also by the presence of blood and cere-
brospinal liquids which occur during the surgical intervention. The contrast agent induces artifacts as well, called bub-
ble noise and blooming effect. Boundaries of the vascular structures are therefore unclearly defined in the USA data.
The extraction of the vascular structures would improve their visualization by keeping the object of interest only and by
eliminating noise and artifacts.
Segmentation of vascular structures has already been extensively studied and validated with success on patient data but
mainly focuses on good quality images [2]. Some methods have been however adapted on US data. In 3D power Dopp-
ler data thresholding techniques ([3]) or region growing algorithms ([4]) are the most common methods used. In 3D B-
mode volumes model-based techniques are required to overcome the problem of unclear borders. Two dimensional ac-
tive contours performing slice by slice ([5]) and a dynamic balloon represented by a triangular mesh ([6]) have been
used to segment the carotid contours. Furthermore, Krissian et al [6] proposed a model-based multiscale scheme that
computes a vesselness measure to segment the aorta artery. The model is represented by a circular cross-section cylind-
er. The technique aims estimating the centerline position and radii values of elongated structures in the data. A multi-
scale implementation allows extracting vascular structures of different sizes. It is then possible to generate a segmenta-
tion model from the estimated centerlines and radii values. None of these methods have been validated so far on 3D
USA data of the brain.
In this work we aim to quantitatively estimate on a physical vascular phantom the performance of an adapted version of
the semi-automatic segmentation method proposed in [6].
75
10. CURAC-Jahrestagung, 15. - 16. September 2011, Magdeburg
2 Methods
Semi-automatic segmentation method
The segmentation method is based on a model-based multiscale detection of the vessel centerlines using a cylindrical
model with circular cross-section. Briefly, a vesselness measure is computed for each voxel of the USA volume which
represents the probability that a voxel belongs to the centerline of an elongated structure. The vesselness measure at a
voxel position is computed based on the image gradient information along a circle C which centre is the voxel itself. Its
orientation is defined based on the computation of the eigenvectors of the structure tensor which represent the direc-
tions of axis and cross-section of the elongated structure. The structure tensor is computed for a sigma value σ propor-
tional to the radius value r of the elongated structure, representing the radius value of the circle C. Thus, a multi-scale
implementation consisting in computing the vesselness measures for Nscales different radii values rlower ≤ r ≤ rupper is used
to extract vascular structures of various sizes. For each voxel, the maximum vesselness response is kept in the multis-
cale space and the corresponding rmax value represents an estimation of the radius. In the original method, the user ma-
nually selects the vascular structure centerlines in the volume of maximum vesselness responses. In order to reduce the
interaction, the maximum vesselness responses are here thresholded with a value Tmaxvess provided by the user. The cen-
terline segments whose sizes are shorter than a given pruning size Spruning are considered as noise and are automatically
eliminated. A surface reconstruction of the vascular structures, called here segmentation model, is then generated based
on the extracted centerlines and estimated radii information. The surface is obtained in two steps: i) creation of a vo-
lume data representing the distance transform to the estimated tubular structures processing each segment as a circular
cylinder, ii) iso-surface generation based on the marching cubes algorithm (Figure 2a).
Physical vascular phantom and 3D US acquisition
The physical phantom includes two silicon tubes mimicking blood vessels whose inside diameter is 4 mm and wall
thickness is 1 mm (Figure 1a). The silicon tubes have been laid down into a plastic container filled with gelatine. The
acquisition system of the 3D USA data includes a common US device (Sonoline Elegra, Siemens) with a 2D free-hand
2.5 MHz phased array probe, an optical tracking system (NDI, Polaris) and a navigation system (SonoNavigator, Loca-
lite). The optical tracking system aims at estimating the position of the US probe in the room. The navigation system is
used to compound the set of 2D US images acquired with the US probe within an US volume. The vascular phantom is
linked to a pump which simulates a laminar blood flow within the tubes, filled with water. Short before the acquisition,
an US contrast agent (SonoVue, Bracco) is injected into the phantom tubes. The operator scans then the phantom sur-
face with the US probe positioned perpendicular to the tube lengths and moved parallel to the tubes. A set of 2D en-
hanced images is obtained and sent to the navigation system through a S-video connection. A 3D USA volume of voxel
size 1×1×1 mm3 is eventually reconstructed (Figure 1 b and c).
Figure 1: The physical vascular phantom includes two silicon tubes mimicking the blood vessels (a). (b) and (c) are
two slices of the 3D USA data acquired with an US contrast agent. (d) and (e) are two slices of the CT vo-
lume.
76
10. CURAC-Jahrestagung, 15. - 16. September 2011, Magdeburg
Evaluation of the segmentation algorithm by comparison with a gold standard
We assume that the phantom tube geometrical features may deform during the building (weight of the gelatine) and
through the image acquisition process (pressure of the US probe on the phantom surface). We estimate therefore the
performance of the segmentation algorithm operating on the 3D USA data by comparison with CT data of the phantom,
considered as gold standard. A CT scanner (Philips) is used here in helical mode with 0.33 mm spacing between the
slices. The pixel size in the slices is 0.20x0.20 mm². The phantom tubes are filled with water and the scanning per-
formed without the pump for easier practical reasons (Figure 1 d and e).
The extraction of tube lumens in the CT data is performed by a region growing algorithm with upper threshold value set
to zero representing the interface between the silicon wall and the water (Figure 2b). Evaluation of the segmentation
algorithm performance is done by comparing the tube lumen radii values. As it was already described above, the model-
based segmentation algorithm provides an estimate of the radii values for each point of the extracted centerlines. The
radii values in the CT data are computed as following. We assume, based on visual observation, that the tube lumens
are perpendicular to volume cross-sections, and that therefore the lumen cross-sections are disks. The number of voxels
included in the lumen cross-sections is counted in each volume cross-section. It represents the surface SCT and the radii
values rCT are then deduced.
Comparison of the segmentation algorithm and manual delineations
Result of the segmentation algorithm is then compared to manual delineations. Seven observers manually delineated the
tube lumen borders in the 3D USA data of the phantom using the free ITK-SNAP segmentation tool. The observers
needed between 15 to 30 minutes to perform the task. High differences between the manual delineations are observed
and the delineation of two observers has been removed due to a too large overestimation of the tube lumens. An aver-
age volume has been then computed from the manual delineations of the five remaining observers (Figure 2c).
The geometrical features defined for the comparison are the tube lumen cross-section area and the centerline distance,
since the lumen contours in the average delineation are rather elliptic. The cross-section areas are computed for each
volume cross-section as the number of lumen voxels. The average delineation has been thinned to extract the lumen
centerline and the distance to the centerline of the segmentation model computed for each volume cross-section.
Figure 2: Phantom tube lumens extracted by (a) the segmentation algorithm in the 3D USA data, (b) a region growing
method in the CT data and (c) manual delineations performed in the 3D USA data by observers and averaged.
3 Results
Segmentation model generation
The tube lumens have been segmented in the 3D USA data using the segmentation method previously described. Al-
though the known dimensions of the silicon tubes in the phantom are constant, the tube lumen diameters do not look
homogeneous in the 3D USA data. The segmentation algorithm has been therefore applied in a multi-scale manner with
77
10. CURAC-Jahrestagung, 15. - 16. September 2011, Magdeburg
the following values: rlower=0.5, rupper=5.0 and Nscales=10. Thus, a large set of radii values are tested by the algorithm.
The segmentation model has been generated with Tmaxvess=15.0 and Spruning=5 (Figure 2a).
Comparison of the segmentation model with the gold standard and the average delineation
Geometrical features have been calculated on the segmentation model, the gold standard and the average delineation as
previously explained. Since the voxel size is different in both volumes, the mean values and standard deviations have
been used for the comparison (Table 1). Values show that the mean radii values in the segmentation model are larger
than nearly one millimeter in comparison to the real tube lumen size and larger than more than half of a millimeter in
comparison to the gold standard. The mean cross-section areas estimated by the observers are twice larger than those
values in the gold standard although this report is smaller than two in the comparison between the segmentation model
and the gold standard. The mean centerline distance values calculated between the segmentation model and the mean
delineation is less than one voxel (0.8 ± 0.6 mm for tube 1 and 0.6 ± 0.6 mm for tube 2). The tube centerlines are there-
fore estimated nearly at the same position by the algorithm and the observers.
radii values (mm) cross-section area (mm2)
tube 1 tube 2 tube 1 tube 2
phantom 2.0 2.0 12.6 12.6
gold standard 2.24 ± 0.04 2.30 ± 0.05 15.8 ± 0.6 16.5 ± 0.7
average delineation - - 41.6 ± 6.8 38.6 ± 5.2
segmentation model 2.80 ± 0.43 3.03 ± 0.38 30.0 ± 5.8 25.9 ± 7.2
Table 1: Comparison of the segmentation model with the gold standard and the average delineation using the
radii and cross-section area features.
4 Discussion
Comparison results showed that the mean radius value of the segmentation model generated by the algorithm is overes-
timated of more than half of a millimeter in comparison with the gold standard. Two main reasons may explain this dif-
ference. First, the image resolution in the US data is lower, increasing the partial volume effect. The tube lumen diame-
ters look visually larger than 4 voxels. Second, the tube lumen cross-sections in the US data look rather like an ellipse,
due to the US probe pressure during the data acquisition. Visually, the circular cross-sections of the segmentation model
have for radii values the largest ellipse axis. Moreover the geometrical features calculated in the gold standard are
slightly larger than the real tube sizes meaning that the phantom deformed. We showed moreover that the observers still
more overestimated the tube lumen cross-section area than the segmentation algorithm did it. They have been more
hindered by noise and artifacts in the data to correctly delineate the unclear tube borders. We conclude that the segmen-
tation algorithm succeeded therefore, on our vascular phantom, in providing a model with realistic geometrical features
regarding the low image resolution of the 3D USA data and in eliminating noise and artifacts. Moreover, processing
time for the multi-scale scheme was less than one minute for a volume of 62x71x145 voxels. The algorithm parameters
rlower, rupper, Nscales and Spruning may be set fixed, also for patient data. Only the threshold value Tmaxvess for extracting the
vascular structures has to be tuned since its value is a compromise between the among of information and the among of
noise in the segmentation model. However, the segmentation tool is suitable for the operating room.
Next step will consist in evaluating the segmentation algorithm on a more realistic vascular phantom including tubes of
different radii values and close to the cerebral vascular anatomy and with bifurcations. It should be interesting to check
the behavior of the segmentation algorithm on thinner tubes and bifurcations. Tests on intraoperative 3D USA data of
patients is planed as well since the goal of the project is to integrate the segmentation model of the intraoperative cere-
bral vascular network into a navigation system. Applications might be the guidance of the neurosurgeon to reach the
tumor without damaging the surrounding blood vessels or the check of the success of aneurysm clipping surgeries.
5 References
[1] Unsgaard G, Rygh OM, Selbekk T, Müller TB, Kolstad F, Lindseth F, Nagelhus Hernes TA. Intra-operative 3D
ultrasound in neurosurgery. Acta Neurochir 2006;148:235-253.
[2] Kirbas C, Quek Francis. A review of vessel extraction techniques and algorithms. ACM Computing Surveys
2004;36(2):81-121.
78
10. CURAC-Jahrestagung, 15. - 16. September 2011, Magdeburg
[3] Reinertsen I, Lindseth F, Unsgaard G, Collins DL. Clinical validation of vessel-based registration for correc-
tion of brain-shift. Medical Image Analysis 2007;11:673-684.
[4] Hold S, Hensel K, Winter S, Dekomien C, Schmitz G. Segmentation of blood vessels in 3D ultrasound-datasets
by a model-based region growing algorithm In Proceedings of Computer Assisted Orthopaedic Surgery
(CAOS) 2007:610-603.
[5] Gill J, Ladak H, Steinman D, Fenster A. Accuracy and Variability Assessment of Semi-Automatic Technique
for Segmentation of the Carotid Arteries from 3D Ultrasound Images. Medical Physics 2000;27(6):1333-1342.
[6] A Zahalka, A Fenster. An automated segmentation method for three-dimensional carotid ultrasound images.
Phys Med Biol 2001;46:1321-1342.
[7] Krissian K, Ellsmere J, Vosburgh K, Kikinis R, Westin CF. Multiscale segmentation of the aorta in 3D ultra-
sound images. Proceedings of the 25th Annual Int. Conf. on the IEEE Engineering in Medicine and Biology
Society, EMBS, Cancun Mexico 2003:638-641.
79