Biomedical Engineering Reference
In-Depth Information
z
g
Top (moving) frame
a
b
y
x
Base (fixed) frame
Fig. 2.3 External view of robotic couch with six degrees of freedom. The couch system consists
of top (moving) frame linked with a fixed base frame using independent mechanical legs. Here
the top platform is defined by six independent position-orientation variables—coordinates (x, y, z,
a, b, c)[
10
]
# of Cited
Model-based
Model-free
Hybrid
Enhanced
Adaptive & Sinusoidal [37]
110
Deformation from orbiting views [14]
Vector based [126]
Local regression [87]
70
Adaptive motion [100]
Finite state [53]
Support vector regression [33]
60
Quaternion-based [115]
50
Adaptive NN [35-36]
36]
Optical flow deformable [38]
Adaptive NN [8]
Kernel density [32]
40
Adaptive tumor
tracking system [41]
Single-imager
DMLC [85]
Local circular
motion [12]
Patient-specific [18]
30
ARMA [47-
48]
Hidden Markov [31]
Adaptive neuro fuzzy [40]
20
MLC-based [86]
Linear & Kalman filter
& Artificial NN [46]
Finite element [34]
Hybrid Extended
Kalman filter [7]
Diaphragm-based [82]
Surrogate-based [83]
10
IMM [45]
0
2004
2005
2006
2007
2008
2009
2010
2011
Year
Fig. 2.4 Variable prediction algorithms for respiratory motion. This figure shows the key
studies, which have more than 30 references in the last 10 years, representing to the salient
algorithms covered
2.3 Prediction Algorithms for Respiratory Motion
A number of prediction methods for respiratory motion have been investigated
based on surrogate markers and tomography images [
5
,
16
,
38
,
55
,
56
,
72
-
78
]. The
previous methods can be categorized into three approaches: (1) model-based
approach [
38
,
55
,
56
,
73
,
75
,
77
] which uses a specific biomechanical or mathe-
matical model for respiratory motion functions or models; (2) model-free approach
[
57
,
68
,
74
,
75
,
79
] heuristic learning algorithms that are trained based on the
observed respiratory patterns; (3) hybrid approach [
80
,
81
], which uses united
Search WWH ::
Custom Search