lecture 10: medical image segmentation as an …bagci/teaching/mic16/lec10.pdfmedical image...
TRANSCRIPT
![Page 1: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/1.jpg)
MEDICAL IMAGE COMPUTING (CAP 5937)
LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem
Dr. Ulas BagciHEC 221, Center for Research in Computer Vision (CRCV), University of Central Florida (UCF), Orlando, FL [email protected] or [email protected]
1SPRING 2016
![Page 2: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/2.jpg)
Outline• Energy functional
– Data and Smoothness terms• Graph Cut
– Min cut– Max Flow
• Applications in Radiology Images
2
![Page 3: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/3.jpg)
Labeling & Segmentation• Labeling is a common way for modeling various computer
vision problems (e.g. optical flow, image segmentation, stereo matching, etc)
3
![Page 4: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/4.jpg)
Labeling & Segmentation• Labeling is a common way for modeling various computer
vision problems (e.g. optical flow, image segmentation, stereo matching, etc)
• The set of labels can be discrete (as in image segmentation)
4
L = {l1, . . . , lm} with L = m
![Page 5: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/5.jpg)
Labeling & Segmentation• Labeling is a common way for modeling various computer
vision problems (e.g. optical flow, image segmentation, stereo matching, etc)
• The set of labels can be discrete (as in image segmentation)
• Or continuous (tracking, etc.)
5
L = {l1, . . . , lm} with L = m
L ⇢ Rnfor n � 1
![Page 6: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/6.jpg)
Labeling is a function• Labels are assigned to sites (pixel locations)
6
![Page 7: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/7.jpg)
Labeling is a function• Labels are assigned to sites (pixel locations)• For a given image, we have
7
|⌦| = Ncols
.Nrows
![Page 8: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/8.jpg)
Labeling is a function• Labels are assigned to sites (pixel locations)• For a given image, we have• Identifying a labeling function (with segmentation) is
8
|⌦| = Ncols
.Nrows
f : ⌦ ! L
![Page 9: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/9.jpg)
Labeling is a function• Labels are assigned to sites (pixel locations)• For a given image, we have• Identifying a labeling function (with segmentation) is
We aim at calculating a labeling function that minimizes a given (total) error or energy
9
|⌦| = Ncols
.Nrows
f : ⌦ ! L
![Page 10: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/10.jpg)
Labeling is a function• Labels are assigned to sites (pixel locations)• For a given image, we have• Identifying a labeling function (with segmentation) is
We aim at calculating a labeling function that minimizes a given (total) error or energy
10
|⌦| = Ncols
.Nrows
f : ⌦ ! L
E(f) =X
p2⌦
[Edata
(p, fp
) +X
q2A(p)
Esmooth
(fp
, fq
)]
*A is an adjacency relation between pixel locations
![Page 11: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/11.jpg)
Example of Energy Minimization Methods
11
Credit: Zhao et al., IJNMBE 15
![Page 12: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/12.jpg)
Energy Function?• Penalizing results which are not compatible with the
observed images/volumes
12
![Page 13: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/13.jpg)
Energy Function?• Penalizing results which are not compatible with the
observed images/volumes
13
Unary (data) cost (inverted)
![Page 14: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/14.jpg)
Energy Function?• Penalizing results which are not compatible with the
observed images/volumes• Enforcing spatial coherence.
14
Pairwise (boundary) cost (inverted)
![Page 15: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/15.jpg)
Energy Function?• Penalizing results which are not compatible with the
observed images/volumes• Enforcing spatial coherence.
15
Pairwise (boundary) cost (inverted)
p q
![Page 16: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/16.jpg)
Segmentation as an Energy Minimization Problem
• Edata assigns non-negative penalties to a pixel location p when assigning a label to this location.
10/1/15
16
![Page 17: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/17.jpg)
Segmentation as an Energy Minimization Problem
• Edata assigns non-negative penalties to a pixel location p when assigning a label to this location.
• Esmooth assigns non-negative penalties by comparing the assigned labels fp and fq at adjacent positions p and q.
17
![Page 18: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/18.jpg)
Segmentation as an Energy Minimization Problem
• Edata assigns non-negative penalties to a pixel location p when assigning a label to this location.
• Esmooth assigns non-negative penalties by comparing the assigned labels fp and fq at adjacent positions p and q.
18
Node/vertexEdge
pixelp q
Image as a graphrepresentation
![Page 19: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/19.jpg)
Segmentation as an Energy Minimization Problem
• Edata assigns non-negative penalties to a pixel location p when assigning a label to this location.
• Esmooth assigns non-negative penalties by comparing the assigned labels fp and fq at adjacent positions p and q.
19
This optimization model is characterized by local interactions along edges between adjacent pixels, and often called MRF (Markov Random Field) model.
![Page 20: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/20.jpg)
Energy Function-Details
20
E(f) =X
p2⌦
[Edata
(p, fp
) +X
q2A(p)
Esmooth
(fp
, fq
)]
![Page 21: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/21.jpg)
Energy Function-Details
21
E(f) =X
p2⌦
[Edata
(p, fp
) +X
q2A(p)
Esmooth
(fp
, fq
)]
Edata(p, fp) = (p)
(p = 0) = �logP (p 2 BG)
(p = 1) = �logP (p 2 FG)
Example Data Term:
![Page 22: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/22.jpg)
Energy Function-Details
22
E(f) =X
p2⌦
[Edata
(p, fp
) +X
q2A(p)
Esmooth
(fp
, fq
)]
Edata(p, fp) = (p)
(p = 0) = �logP (p 2 BG)
(p = 1) = �logP (p 2 FG)
Example Data Term:
(p, q) = Kpq�(p 6= q) where
Kpq =exp(��(Ip � Iq)2/(2�2))
||p, q||
Example Smoothness Term:
![Page 23: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/23.jpg)
Energy Function-Details
23
E(f) =X
p2⌦
[Edata
(p, fp
) +X
q2A(p)
Esmooth
(fp
, fq
)]
E(p) =X
p2⌦
p(p) +X
p2⌦
X
q2A(p)
pq(p, q)
p⇤ = argminp2L
E(p)
![Page 24: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/24.jpg)
Energy Function-Details
24
E(f) =X
p2⌦
[Edata
(p, fp
) +X
q2A(p)
Esmooth
(fp
, fq
)]
E(p) =X
p2⌦
p(p) +X
p2⌦
X
q2A(p)
pq(p, q)
p⇤ = argminp2L
E(p)
To solve this problem, transform the energy functional into min-cut/max-flow problem and solve it!
![Page 25: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/25.jpg)
Graph Cuts for Optimal Boundary Detection (Boykov ICCV 2001)
25
23
F
B
F
B
F
F F
F B
B
B
• Binary label: foreground vs. background• User labels some pixels • Exploit
– Statistics of known Fg & Bg– Smoothness of label
• Turn into discrete graph optimization– Graph cut (min cut / max flow)
![Page 26: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/26.jpg)
Graph Cut26
Each pixel is connected to its neighbors in an undirected graphGoal: split nodes into two sets A and B based on pixel values, and try to classify Neighbors in the same way too!
![Page 27: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/27.jpg)
Graph Cut27
Each pixel is connected to its neighbors in an undirected graphGoal: split nodes into two sets A and B based on pixel values, and try to classify Neighbors in the same way too!
![Page 28: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/28.jpg)
Graph-Cut28
• Each pixel = node• Add two nodes F & B• Labeling: link each pixel to either F or B
F
B
F
B
F
F F
F B
B
B
Desired result
![Page 29: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/29.jpg)
Cost Function: Data term
29
• Put one edge between each pixel and both F & G• Weight of edge = minus data term
B
F
![Page 30: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/30.jpg)
Cost Function: Smoothness term30
• Add an edge between each neighbor pair• Weight = smoothness term
B
F
![Page 31: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/31.jpg)
Min-Cut31
• Energy optimization equivalent to graph min cut• Cut: remove edges to disconnect F from B• Minimum: minimize sum of cut edge weight
B
Fcut
![Page 32: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/32.jpg)
Min-Cut32
Source
Sink
v1 v2
2
5
9
42
1Graph (V, E, C)
Vertices V = {v1, v2 ... vn}Edges E = {(v1, v2) ....}Costs C = {c(1, 2) ....}
![Page 33: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/33.jpg)
Min-Cut33
Source
Sink
v1 v2
2
5
9
42
1
What is a st-cut?
An st-cut (S,T) divides the nodes between source and sink.
What is the cost of a st-cut?
Sum of cost of all edges going from S to T
5 + 2 + 9 = 16
![Page 34: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/34.jpg)
Min-Cut34
What is a st-cut?
An st-cut (S,T) divides the nodes between source and sink.
What is the cost of a st-cut?
Sum of cost of all edges going from S to T
Source
Sink
v1 v2
2
5
9
42
1
2 + 1 + 4 = 7
What is the st-mincut?
st-cut with the minimum cost
![Page 35: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/35.jpg)
How to compute min-cut?35
Source
Sink
v1 v2
2
5
9
42
1
Solve the dual maximum flow problem
In every network, the maximum flow equals the cost of the st-mincut
Min-cut\Max-flow Theorem
Compute the maximum flow between Source and Sink
Constraints
Edges: Flow < CapacityNodes: Flow in & Flow out
![Page 36: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/36.jpg)
Max-Flow Algorithms
36
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Source
Sink
v1 v2
2
5
9
42
1
Algorithms assume non-negative capacity
Flow=0
![Page 37: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/37.jpg)
Max-Flow Algorithms37
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
2
5
9
42
1
Flow=0
![Page 38: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/38.jpg)
Max-Flow Algorithms
38
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
2-2
5-2
9
42
1
Flow=0+2
![Page 39: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/39.jpg)
Max-Flow Algorithms
39
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
3
9
42
1
Flow=2
![Page 40: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/40.jpg)
Max-Flow Algorithms
40
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
3
9
42
1
Flow=2
![Page 41: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/41.jpg)
Max-Flow Algorithms
41
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
3
9
42
1
Flow=2
![Page 42: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/42.jpg)
Max-Flow Algorithms
42
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
3
5
02
1
Flow=2+4
![Page 43: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/43.jpg)
Max-Flow Algorithms
43
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
3
5
02
1
Flow=6
![Page 44: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/44.jpg)
Max-Flow Algorithms
44
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
3
5
02
1
Flow=6
![Page 45: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/45.jpg)
Max-Flow Algorithms
45
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
2
4
02+1
1-1
Flow=6+1
![Page 46: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/46.jpg)
Max-Flow Algorithms
46
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
2
4
03
0
Flow=7
![Page 47: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/47.jpg)
Max-Flow Algorithms
47
Augmenting Path Based Algorithms
1. Find path from source to sink with positive capacity
2. Push maximum possible flow through this path
3. Repeat until no path can be found
Algorithms assume non-negative capacity
Source
Sink
v1 v2
0
2
4
03
0
Flow=7
![Page 48: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/48.jpg)
Another Example-Max Flow
48
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
![Page 49: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/49.jpg)
Another Example-Max Flow
49
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
![Page 50: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/50.jpg)
Another Example-Max Flow
50
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 3
![Page 51: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/51.jpg)
Another Example-Max Flow
51
source
sink
9
5
6-3
8-3
42+3
2
2
5-3
3-3
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 3
+3
![Page 52: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/52.jpg)
Another Example-Max Flow
52
source
sink
9
5
3
5
45
2
2
2
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 3
3
![Page 53: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/53.jpg)
Another Example-Max Flow
53
source
sink
9
5
3
5
45
2
2
2
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 3
3
![Page 54: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/54.jpg)
Another Example-Max Flow
54
source
sink
9
5
3
5
45
2
2
2
5
5
311
6
2
3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 6
3
![Page 55: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/55.jpg)
Another Example-Max Flow
55
source
sink
9-3
5
3
5-3
45
2
2
2
5
5
3-311
6
2
3-3
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 6
3
+3
+3
![Page 56: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/56.jpg)
Another Example-Max Flow
56
source
sink
6
5
3
2
45
2
2
2
5
5
11
6
2
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 6
3
3
3
![Page 57: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/57.jpg)
Another Example-Max Flow
57
source
sink
6
5
3
2
45
2
2
2
5
5
11
6
2
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 6
3
3
3
![Page 58: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/58.jpg)
Another Example-Max Flow
58
source
sink
6
5
3
2
45
2
2
2
5
5
11
6
2
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 11
3
3
3
![Page 59: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/59.jpg)
Another Example-Max Flow
59
source
sink
6-5
5-5
3
2
45
2
2
2
5-5
5-5
1+51
6
2+5
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 11
3
3
3
![Page 60: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/60.jpg)
Another Example-Max Flow
60
source
sink
1
3
2
45
2
2
2
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 11
3
33
![Page 61: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/61.jpg)
Another Example-Max Flow
61
source
sink
1
3
2
45
2
2
2
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 11
3
33
![Page 62: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/62.jpg)
Another Example-Max Flow
62
source
sink
1
3
2
45
2
2
2
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 13
3
33
![Page 63: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/63.jpg)
Another Example-Max Flow
63
source
sink
1
3
2-2
45
2-2
2-2
2-2
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 13
3
33
+2
+2
![Page 64: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/64.jpg)
Another Example-Max Flow
64
source
sink
1
3
45
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 13
3
33
2
2
![Page 65: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/65.jpg)
Another Example-Max Flow
65
source
sink
1
3
45
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 13
3
33
2
2
![Page 66: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/66.jpg)
Another Example-Max Flow
66
source
sink
1
3
45
61
6
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 15
3
33
2
2
![Page 67: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/67.jpg)
Another Example-Max Flow
67
source
sink
1
3-2
4-25
61
6-2
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 15
3
33+2
2
2-2
+2
![Page 68: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/68.jpg)
Another Example-Max Flow
68
source
sink
1
1
5
61
4
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 15
3
35
2
2
2
![Page 69: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/69.jpg)
Another Example-Max Flow
69
source
sink
1
1
5
61
4
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 15
3
35
2
2
2
![Page 70: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/70.jpg)
Another Example-Max Flow
70
source
sink
1
1
5
61
4
7
Ford & Fulkerson algorithm (1956)
Find the path from source to sink
While (path exists)
flow += maximum capacity in the path
Build the residual graph (“subtract” the flow)
Find the path in the residual graph
End
flow = 15
3
35
2
2
2
![Page 71: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/71.jpg)
Another Example-Max Flow
71
source
sink
1
1
5
61
4
7
Ford & Fulkerson algorithm (1956)
Why is the solution globally optimal ?
flow = 15
3
35
2
2
2
![Page 72: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/72.jpg)
Another Example-Max Flow
72
source
sink
1
1
5
61
4
7
Ford & Fulkerson algorithm (1956)
Why is the solution globally optimal ?
1. Let S be the set of reachable nodes in the
residual graph
flow = 15
3
35
2
2
2
S
![Page 73: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/73.jpg)
Another Example-Max Flow
73
Ford & Fulkerson algorithm (1956)
Why is the solution globally optimal ?
1. Let S be the set of reachable nodes in the
residual graph
2. The flow from S to V - S equals to the sum
of capacities from S to V – S
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
S
![Page 74: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/74.jpg)
Another Example-Max Flow
74
flow = 15
Ford & Fulkerson algorithm (1956)
Why is the solution globally optimal ?
1. Let S be the set of reachable nodes in the residual graph
2. The flow from S to V - S equals to the sum of capacities from S to V – S
3. The flow from any A to V - A is upper bounded by the sum of capacities from A to V – A
4. The solution is globally optimal
source
sink
8/9
5/5
5/6
8/8
2/40/2
2/2
0/2
5/5
3/3
5/5
5/5
3/30/10/1
2/6
0/2
3/3
Individual flows obtained by summing up all paths
![Page 75: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/75.jpg)
Another Example-Max Flow
75
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
S
T
cost = 18
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
S
T
![Page 76: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/76.jpg)
Another Example-Max Flow
76
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
S
T
cost = 23
![Page 77: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/77.jpg)
Another Example-Max Flow
77
C(x) = 5x1 + 9x2 + 4x3 + 3x3(1-x1) + 2x1(1-x3)
+ 3x3(1-x1) + 2x2(1-x3) + 5x3(1-x2) + 2x4(1-x1)
+ 1x5(1-x1) + 6x5(1-x3) + 5x6(1-x3) + 1x3(1-x6)
+ 3x6(1-x2) + 2x4(1-x5) + 3x6(1-x5) + 6(1-x4)
+ 8(1-x5) + 5(1-x6)
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
x1 x2
x3
x4
x5
x6
![Page 78: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/78.jpg)
Another Example-Max Flow
78
C(x) = 2x1 + 9x2 + 4x3 + 2x1(1-x3)
+ 3x3(1-x1) + 2x2(1-x3) + 5x3(1-x2) + 2x4(1-x1)
+ 1x5(1-x1) + 3x5(1-x3) + 5x6(1-x3) + 1x3(1-x6)
+ 3x6(1-x2) + 2x4(1-x5) + 3x6(1-x5) + 6(1-x4)
+ 5(1-x5) + 5(1-x6)
+ 3x1 + 3x3(1-x1) + 3x5(1-x3) + 3(1-x5)
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
x1 x2
x3
x4
x5
x6
![Page 79: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/79.jpg)
Another Example-Max Flow
79
C(x) = 2x1 + 9x2 + 4x3 + 2x1(1-x3)
+ 3x3(1-x1) + 2x2(1-x3) + 5x3(1-x2) + 2x4(1-x1)
+ 1x5(1-x1) + 3x5(1-x3) + 5x6(1-x3) + 1x3(1-x6)
+ 3x6(1-x2) + 2x4(1-x5) + 3x6(1-x5) + 6(1-x4)
+ 5(1-x5) + 5(1-x6)
+ 3x1 + 3x3(1-x1) + 3x5(1-x3) + 3(1-x5)
3x1 + 3x3(1-x1) + 3x5(1-x3) + 3(1-x5)
=
3 + 3x1(1-x3) + 3x3(1-x5)
source
sink
9
5
6
8
42
2
2
5
3
5
5
311
6
2
3
x1 x2
x3
x4
x5
x6
![Page 80: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/80.jpg)
Another Example-Max Flow
80
C(x) = 3 + 2x1 + 6x2 + 4x3 + 5x1(1-x3)
+ 3x3(1-x1) + 2x2(1-x3) + 5x3(1-x2) + 2x4(1-x1)
+ 1x5(1-x1) + 3x5(1-x3) + 5x6(1-x3) + 1x3(1-x6)
+ 2x5(1-x4) + 6(1-x4)
+ 2(1-x5) + 5(1-x6) + 3x3(1-x5)
+ 3x2 + 3x6(1-x2) + 3x5(1-x6) + 3(1-x5)
3x2 + 3x6(1-x2) + 3x5(1-x6) + 3(1-x5)
=
3 + 3x2(1-x6) + 3x6(1-x5)
source
sink
9
5
3
5
45
2
2
2
5
5
311
6
2
33
x1 x2
x3
x4
x5
x6
![Page 81: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/81.jpg)
81
C(x) = 6 + 2x1 + 6x2 + 4x3 + 5x1(1-x3)
+ 3x3(1-x1) + 2x2(1-x3) + 5x3(1-x2) + 2x4(1-x1)
+ 1x5(1-x1) + 3x5(1-x3) + 5x6(1-x3) + 1x3(1-x6)
+ 2x5(1-x4) + 6(1-x4) + 3x2(1-x6) + 3x6(1-x5)
+ 2(1-x5) + 5(1-x6) + 3x3(1-x5)
source
sink
6
5
3
2
45
2
2
2
5
5
11
6
2
3
3
3
x1 x2
x3
x4
x5
x6
Another Example-Max Flow
![Page 82: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/82.jpg)
Another Example-Max Flow82
C(x) = 15 + 1x2 + 4x3 + 5x1(1-x3)
+ 3x3(1-x1) + 7x2(1-x3) + 2x1(1-x4)
+ 1x5(1-x1) + 6x3(1-x6) + 6x3(1-x5)
+ 2x5(1-x4) + 4(1-x4) + 3x2(1-x6) + 3x6(1-x5)
source
sink
1
1
5
61
4
7
3
35
2
2
2x1 x2
x3
x4
x5
x6 cost = 0
min cut
S
T
cost = 0
• All coefficients positive
• Must be global minimum
S – set of reachable nodes from s
![Page 83: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/83.jpg)
History of Max-Flow Algorithms83
Augmenting Path and Push-Relabeln: #nodesm: #edgesU: maximum edge weight
Algorithms assume non-negative edge
weights
![Page 84: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/84.jpg)
Software Packages for Optimization
84
![Page 85: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/85.jpg)
Applications Used in Energy Minimization Based Segmentation Methods
85
![Page 86: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/86.jpg)
Applications
86
![Page 87: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/87.jpg)
Interactive Organ Segmentation (Boykov and Jolly, MICCAI 2000)
87
Segmentation of multiple objects. (a-c): Cardiac MRI. (d): Kidney CE-MR angiography
![Page 88: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/88.jpg)
Interactive Organ Segmentation (Boykov and Jolly, MICCAI 2000)
88
Segmentation of the right lung in CT. (a): representative 2D slice of original 3D data. (b): segmentation results on the slice in (a). (c-d) 3D visualization of segmentation results.
![Page 89: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/89.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
• Isolating the entire heart allows the coronary vessels on the surface of the heart to be easily visualized despite the proximity of surrounding organs such as the ribs and pulmonary blood vessels.
89
![Page 90: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/90.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
• Isolating the entire heart allows the coronary vessels on the surface of the heart to be easily visualized despite the proximity of surrounding organs such as the ribs and pulmonary blood vessels.
• Numerous techniques have been described for segmenting the left ventricle of the heart in images from various types of medical scanners but rarely has the entire heart been segmented.
90
![Page 91: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/91.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
• Isolating the entire heart allows the coronary vessels on the surface of the heart to be easily visualized despite the proximity of surrounding organs such as the ribs and pulmonary blood vessels.
• Numerous techniques have been described for segmenting the left ventricle of the heart in images from various types of medical scanners but rarely has the entire heart been segmented.
• Graph-cut formation:
91
![Page 92: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/92.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
• Isolating the entire heart allows the coronary vessels on the surface of the heart to be easily visualized despite the proximity of surrounding organs such as the ribs and pulmonary blood vessels.
• Numerous techniques have been described for segmenting the left ventricle of the heart in images from various types of medical scanners but rarely has the entire heart been segmented.
• Graph-cut formation:
92
![Page 93: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/93.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
• Isolating the entire heart allows the coronary vessels on the surface of the heart to be easily visualized despite the proximity of surrounding organs such as the ribs and pulmonary blood vessels.
• Numerous techniques have been described for segmenting the left ventricle of the heart in images from various types of medical scanners but rarely has the entire heart been segmented.
• Graph-cut formation:
93
(C is center)
![Page 94: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/94.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
• Isolating the entire heart allows the coronary vessels on the surface of the heart to be easily visualized despite the proximity of surrounding organs such as the ribs and pulmonary blood vessels.
• Numerous techniques have been described for segmenting the left ventricle of the heart in images from various types of medical scanners but rarely has the entire heart been segmented.
• Graph-cut formation:
94
If cos(.) <0, (0 otherwise.)
(C is center)
![Page 95: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/95.jpg)
AUTOMATIC HEART ISOLATION FOR CT CORONARY VISUALIZATION USING GRAPH-CUTS (Funka-Lea et al, ISBI 2006)
95
Top Left:A balloon is expanded within the heart. The heart wall pushes the balloon toward the heart center as the balloon grows.
Top right: volume rendering of original heart volume.
Bottom left: heart cropped based on segmentation mask.
Bottom right:volume rendering after automatic heart isolation algorithm.
![Page 96: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/96.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
• Proposed a method for simultaneously segmenting longitudinal magnetic resonance (MR) images
96
![Page 97: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/97.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
• Proposed a method for simultaneously segmenting longitudinal magnetic resonance (MR) images– 3D MRI + time component (longitudinal)
97
![Page 98: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/98.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
• Proposed a method for simultaneously segmenting longitudinal magnetic resonance (MR) images– 3D MRI + time component (longitudinal)
• A 4D graph is used to represent the longitudinal data: – edges are weighted based on spatial and intensity priors and connect
spatially and temporally neighboring voxels represented by vertices in the graph.
98
![Page 99: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/99.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
• Proposed a method for simultaneously segmenting longitudinal magnetic resonance (MR) images– 3D MRI + time component (longitudinal)
• A 4D graph is used to represent the longitudinal data: – edges are weighted based on spatial and intensity priors and connect
spatially and temporally neighboring voxels represented by vertices in the graph.
– Solving the min-cut/max-flow problem on this graph yields the segmentation for all time-points in a single step
99
![Page 100: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/100.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
• Proposed a method for simultaneously segmenting longitudinal magnetic resonance (MR) images– 3D MRI + time component (longitudinal)
• A 4D graph is used to represent the longitudinal data: – edges are weighted based on spatial and intensity priors and connect
spatially and temporally neighboring voxels represented by vertices in the graph.
– Solving the min-cut/max-flow problem on this graph yields the segmentation for all time-points in a single step
• Time-series image can be considered as a single (4D) image, with the following dimensions: x,y,z and t
100
![Page 101: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/101.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
• Proposed a method for simultaneously segmenting longitudinal magnetic resonance (MR) images– 3D MRI + time component (longitudinal)
• A 4D graph is used to represent the longitudinal data: – edges are weighted based on spatial and intensity priors and connect
spatially and temporally neighboring voxels represented by vertices in the graph.
– Solving the min-cut/max-flow problem on this graph yields the segmentation for all time-points in a single step
• Time-series image can be considered as a single (4D) image, with the following dimensions: x,y,z and t
101
6 spatial neighborsin 3DAnd 2 temporal neighbors
and
![Page 102: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/102.jpg)
Measurement of hippocampal atrophy using 4D graph-cut segmentation: Application to ADNI (Wolz, et al. NeuroImage 10)
102
a. Right hippocampus segmentation (baseline), b. follow up segmentation (12 months)
![Page 103: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/103.jpg)
Integrated Graph Cuts for Brain Image Segmentation (Song et al, MICCAI 2006)
• In addition to image intensity, tissue priors and local boundary information are integrated into the edge weight metrics in the graph.
103
![Page 104: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/104.jpg)
Integrated Graph Cuts for Brain Image Segmentation (Song et al, MICCAI 2006)
• In addition to image intensity, tissue priors and local boundary information are integrated into the edge weight metrics in the graph.
104
Example of the graph with three terminals for brain MRI tissue segmentation of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). The set of nodes V includes all voxels and terminals. The set of edges E includes all n-links and t-links.
n-links: voxel-to-voxel edgest-links: voxel-to-terminal edges
![Page 105: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/105.jpg)
Integrated Graph Cuts for Brain Image Segmentation (Song et al, MICCAI 2006)
105
![Page 106: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/106.jpg)
Integrated Graph Cuts for Brain Image Segmentation (Song et al, MICCAI 2006)
106
Pairwise termAtlas termData termRegularizationterm
![Page 107: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/107.jpg)
Integrated Graph Cuts for Brain Image Segmentation (Song et al, MICCAI 2006)
107
Atlas termData termRegularizationterm
(T2, segmentation results)
![Page 108: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/108.jpg)
Summary– Data and Smoothness Terms -> Graph based segmentation methods– Additional terms can(should) be added into segmentation formulation
based observation/need and problem definition– Problems formulated as a MRF task can be solved by max-flow/min-
cut
108
![Page 109: LECTURE 10: Medical Image Segmentation as an …bagci/teaching/mic16/lec10.pdfMEDICAL IMAGE COMPUTING (CAP 5937) LECTURE 10: Medical Image Segmentation as an Energy Minimization Problem](https://reader035.vdocument.in/reader035/viewer/2022070608/5ac137c57f8b9ad73f8ca379/html5/thumbnails/109.jpg)
Slide Credits and References• Fredo Durand• M. Tappen• R. Szelisky• http://www.csd.uwo.ca/faculty/yuri/Abstracts/eccv06-tutorial.html• J.Malcolm, Graph Cut in Tensor Scale• Interactive Graph Cuts for Optimal Boundary & Region Segmentation of
Objects in N-D images.Yuri Boykov and Marie-Pierre Jolly.In International Conference on Computer Vision, (ICCV), vol. I, 2001.http://www.csd.uwo.ca/~yuri/Abstracts/iccv01-abs.html
• http://www.cse.yorku.ca/~aaw/Wang/MaxFlowStart.htm• http://research.microsoft.com/vision/cambridge/i3l/segmentation/GrabCut.htm• http://www.cc.gatech.edu/cpl/projects/graphcuttextures/• A Comparative Study of Energy Minimization Methods for Markov Random
Fields. Rick Szeliski, Ramin Zabih, Daniel Scharstein, Olga Veksler, Vladimir Kolmogorov, Aseem Agarwala, Marshall Tappen, Carsten Rother. ECCV 2006www.cs.cornell.edu/~rdz/Papers/SZSVKATR.pdf
• P. Kumar, Oxford University.
109