Method for iteratively extracting motion parameters from angiography images

Patent metadata

Field	Value
Publication number	US-2016189394-A1
Application number	US-201514960461-A
Country	US
Kind code	A1
Filing date	Dec 7, 2015
Priority date	Dec 30, 2014
Publication date	Jun 30, 2016
Grant date	—

Abstract

A method for extracting motion parameters from angiography images using a multi-parameter model. The method includes: 1) extracting I vascular structural feature points automatically from a medical image of an angiography image sequence, and auto-tracking the feature points respectively in the angiography image sequence to obtain a tracking sequence of each feature point; 2) performing a discrete Fourier transformation on the tracking sequence of each feature point to obtain a discrete Fourier transformation result; initializing an iterative parameter, and obtaining amplitude range and frequency range of each frequency point of the discrete Fourier transformation result; 3) performing a Fourier transformation on a tracking sequence of each frequency point in the amplitude range and the frequency range thereof to obtain Fourier transformation results; and 4) performing an inverse Fourier transformation on the Fourier transformation results, and obtaining an estimated minimum mean square error of each frequency point.

First claim

The invention claimed is:

1 . A method for extracting motion parameters from angiography images, the method comprising:
(1) extracting I vascular structural feature points from a medical image of an angiography image sequence, and tracking the feature points respectively in the angiography image sequence to obtain a tracking sequence {s i (n), i=1, . . . , I} of each feature point, where n is frame number of the medical image in the angiography image sequence; (2) performing a discrete Fourier transformation on the tracking sequence {s i (n), i=1, . . . , I} of each feature point in (1) to obtain a discrete Fourier transformation result S i (k); (3) initializing an iterative parameter j=0, and obtaining an amplitude range and a frequency range of each frequency point of the discrete Fourier transformation result S i (k) in (2); (4) performing a Fourier transformation on a tracking sequence of each frequency point in the amplitude range and the frequency range thereof to obtain Fourier transformation results; (5) performing an inverse Fourier transformation on the Fourier transformation results in (4), and obtaining an estimated minimum mean square error of each frequency point; (6) determining whether the estimated minimum mean square error is greater than a predetermined threshold, and proceeding to (7) if yes, otherwise ending the process; (7) processing spectrums of each frequency point by a multi-parameter iterative optimizing algori th m to obtain (j+1) th iterated time-domain signals; (8) processing a residual signal by a translation model to obtain a (j+1) th iterated translation signal; (9) adding the (j+1) th iterated time-domain signals to the (j+1) th iterated translation signal to obtain an (j+1) th iterated estimated mixed signal, and calculating a (j+1) th iterated minimum mean square error; and (10) determining whether the (j+1) th iterated minimum mean square error is greater than the threshold in (6), and returning to (7) if yes, otherwise ending the process.

2 . The method of claim 1 , wherein in (1), s i (n) is expressed by the following equation:
s i ( n )= L ( n )+ r i ( n )+ c i ( n )+ h i ( n )+ t i ( n ), i∈[ 1, I],
where L(n) represents translational movement, r i (n) represents breathing movement of an i th feature point, c i (n) represents cardiac movement of the i th feature point, h i (n) represents tremor movement of the i th feature point, and t i (n) represents other movements of the i th feature point.

3 . The method of claim 2 , wherein in (2), S i ( k ) is expressed by the following equation:
S i ( k )= L ( k )+ R i ( k )+ C i ( k )+ H i ( k ),
where k represents a frequency point, and L(k), C(k), R(k) and H(k) represent harmonic coefficients of L(n), c(n), r(n) and h(n) correspondingly and respectively.

4 . The method of claim 3 , wherein in (5), the estimated minimum mean square error {circumflex over (ε)} i j of the frequency point is expressed by the following equation:

ɛ
^

i
j

=

min
(

1
N



∑
n





(

s
i



(
n
)

-

s
^

i
j



(
n
)

)

2

)

,

where ŝ i j (n)=L j (n)+r i j (n)+c i j (n)+h i j (n).

5 . The method of claim 4 , wherein (7) further comprises the following sub-steps of:
(7.1) calculating values L j (k ic ), R i j (k ic ) and H i j (k ic IC) of L j (k), R i j (k) and H i j (k) near a frequency point k ic in the frequency range respectively by the following equation while keeping L j (k), R i j (k) and H i j (k) constant:

X
p



(
k
)

=

∑

n
=
0

N
-
1





x
p



(
n
)

·



-

j


(

2

π

N

)


nk

,

calculating a (j+1) th iterated cardiac signal spectrum by an equation C i j+1 (k ic )=C i 0 (k ic )−R i j (k ic )−H i j (k ic ), performin

Assignees

• Univ Huazhong Science Tech

Inventors

• Zhang Tianxu
• Huang Zhenghua
• Deng Lihua
• Zhang Yaozong
• Song Qiong

Classifications

• G06T7/246

  using feature-based methods, e.g. the tracking of corners or segments · CPC title

• A61B6/5217

  extracting a diagnostic or physiological parameter from medical diagnostic data · CPC title

• G06T2207/30048

  Heart; Cardiac · CPC title

• G06T2207/20056

  Discrete and fast Fourier transform, [DFT, FFT] · CPC title

• G06T7/262

  using transform domain methods, e.g. Fourier domain methods · CPC title