CN105030279B - A kind of tissue characterization method based on ultrasonic radio frequency time series - Google Patents
A kind of tissue characterization method based on ultrasonic radio frequency time series Download PDFInfo
- Publication number
- CN105030279B CN105030279B CN201510359391.0A CN201510359391A CN105030279B CN 105030279 B CN105030279 B CN 105030279B CN 201510359391 A CN201510359391 A CN 201510359391A CN 105030279 B CN105030279 B CN 105030279B
- Authority
- CN
- China
- Prior art keywords
- ultrasonic
- time serieses
- formula
- ultrasound
- tissue
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The invention discloses a kind of tissue characterization methods based on ultrasonic radio frequency time series, include the following steps:1st, feature extraction:Each tissue samples are proceeded as follows, obtain corresponding feature vector;2nd, tissue identification;The step 1 includes the following steps:11, tissue is scanned, obtains its ultrasonic echo RF signal multiframes;12, it demodulates certain frame ultrasonic echo RF signal and shows Type B figure;13, the ROI that size is M × N is chosen on Type B figure;14, its preceding L frames RF signal is taken to construct the ultrasonic RF time serieses that M × N number of length is L to every in ROI;15, fractal characteristic, frequency domain and the temporal signatures of extraction ultrasound RF time serieses are to get to its feature vector;The present invention is based on the features that the extraction of ultrasonic RF time serieses divides three shape, frequency domain, time domain angles, and tissue is identified using grader, have many advantages, such as to be widely used in ultrasound medicine detection, Ultrasonic tissue characterization field.
Description
Technical field
The present invention relates to ultrasound medicine, Ultrasonic tissue characterization fields, more particularly to a kind of to be based on ultrasonic radio frequency time series
Tissue characterization method.
Background technology
Since ultrasound is by the transmission of tissue and the complexity of reflection characteristic, ultrasonic and tissue interaction mechanism is not yet
It is apparent from, people can only differentiate normal and pathological tissues by obtaining the useful information of tissue and making explanations to reach indirectly
Purpose, simultaneously because other inspection examine means there are it is invasive, not reproducible, easily cause complication the problems such as, so as to promote people
Carry out the research of Ultrasonic tissue characterization.
There are mainly two types of traditional Ultrasonic tissue characterization methods:Ultrasonic B figure and ultrasonic echo RF signals.Based on ultrasonic B
The tissue characterization of type image mainly extracts the textural characteristics of image, and then the classification with graders such as SVM, neural networks is accurate
Rate, susceptibility, specificity etc. assess classifying quality.There is following defects by ultrasonic B Tu Dingzhengfa:(1) it adopts
It is modulated with briliancy and shows amplitude information, abandoned abundant phase information so that portion of tissue information is lost;(2) this method makes
With ultrasound image grayscale, inspection result is influenced by instrument model, instrument setting etc., and consistency is poor.It is returned based on ultrasound
Wave RF signals mainly carry out tissue characterization to same frame back scattering echo RF signals using ultrasonic spectrum parametric method, assume in ROI
Tissue is uniform, and specific practice is to obtain ultrasonic echo RF signals using broad-band ultrasonic scanning probe tissue, and demodulation shows certain
One frame intercepts ROI.It carries out spectrum analysis one by one to the acoustic beam in ROI, extracts spectrum signature parameter, finding out can effectively distinguish normally
With the characteristic parameter of pathological tissues.The method needs depth attenuation to compensate, and the tissue attenuation ability of different lesion degrees is that have
Difference, while individual difference can also so that acoustic wave propagation path is different, and then seriously affects its nicety of grading.
Above-mentioned Ultrasonic tissue characterization method to noise-sensitive, is influenced greatly, it is only often line to put forward feature by individual difference
Feature or spectrum signature are managed, the angle for extracting feature is excessively single, and information content is not abundant enough.
Invention content
The shortcomings that it is a primary object of the present invention to overcome the prior art and deficiency provide a kind of based on the ultrasonic radio frequency time
The tissue characterization method of sequence, this method are based on ultrasound RF time serieses, and feature is carried out from three shape, frequency domain, time domain angles are divided
Extraction is trained grader using training set feature vector, test set then is inputted grader, made with the nicety of grading of grader
Evaluation index for tissue identification result.
The purpose of the present invention is realized by following technical solution:A kind of tissue characterization based on ultrasonic radio frequency time series
Method includes the following steps:
Step 1, feature extraction:Each tissue samples are proceeded as follows, obtain corresponding feature vector;
Step 2, tissue identification;
The step 1 includes the following steps:
(1-1) scans tissue, obtains its ultrasonic echo RF signal multiframes;
(1-2) demodulates certain frame ultrasonic echo RF signal and shows Type B figure;
(1-3) chooses the ROI that size is M × N on Type B figure;
(1-4) takes every in ROI its preceding L frames RF signal to construct the ultrasonic RF time serieses that M × N number of length is L;
Fractal characteristic, frequency domain and the temporal signatures of (1-5) extraction ultrasound RF time serieses are to get to its feature vector;
The step 2 includes the following steps:
Tissue samples are divided into test set and training set by (2-1) by a certain percentage;
(2-2) trains grader using the feature vector of training set;
(2-3) is with the assessment recognition result of grader under test set.
In the step (1-5), the method for extracting ultrasound RF time series feature vectors is as follows:
(1-5-1) fractal characteristic:The ultrasonic RF times sequence that the M × N number of length constructed in step (1-4) is L is calculated successively
It is as follows to calculate step for the fractal dimension of row:
(1-5-1-1) note time series is { x (l):1≤l≤L }, x (l) represents the amplitude of l frame ultrasound RF time serieses
Value;
(1-5-1-2) is built new time series by given time series { x (l) }
(1-5-1-3) is calculatedLength Lm(k);
In formula, Lm(k) it is newly-built time seriesLength of curve;
(1-5-1-4) seeks Lm(k) mean value
In formula, L (k) is length of curve mean value;
(1-5-1-5) fitting a straight lineSlope is the FRACTAL DIMENSION of ultrasound RF time serieses
Number;
(1-5-1-6) asks for the fractal dimension mean value of M × N number of ultrasound RF time serieses up to the fractal characteristic of the sample;
(1-5-2) frequency domain character:Be extracted frequency spectrum fitting a straight line slope S lope, intercept Intercept and 4 sections of frequency spectrums it
And S1,S2,S3,S4Totally 6 frequency domain characters, calculating step is:
(1-5-2-1) obtains frequency spectrum X (w) to the ultrasonic RF time serieses { x (l) } that length is L as FFT transform;
(1-5-2-2) averages to frequency spectrum X (w) amplitudes under identical frequency Xave(w);
(1-5-2-3) is to Xave(w) normalized spatial spectrum is obtained as normalized
In formula, | Xave(k) | it is Xave(k) absolute value, w are signal frequency;
(1-5-2-4) is rightFitting a straight line is carried out, obtains spectrum slope S lope, intercept Intercept;Root
S is sought according to formula1,S2,S3,S4:
It is signal length in formula,For normalized spatial spectrum;
(1-5-3) temporal signatures:It is extracted kurtosis Kurtosis, peak value Peak, fuzzy entropy, zero crossing mZCI and zero crossing
Standard deviation nsZCI calculates this 5 spies for the ultrasonic RF time serieses that the M × N number of length constructed in step (1-4) is L successively
Sign, step are as follows:
In formula, L is signal length,For ultrasonic RF time serieses amplitude mean value, x (l) represents l frame ultrasound RF time sequences
The range value of row;
(1-5-3-2) peak value Peak is the mean value of a absolute values by a relatively large margin of p before ultrasound RF time serieses;
(1-5-3-3) calculates fuzzy entropy, and the calculating step of the fuzzy entropy is as follows:
It is { x (l) that (1-5-3-3-1), which sets length as the ultrasonic RF time serieses of L,:1≤l≤L }, wherein, x (l) represents l
The range value of frame ultrasound RF time serieses;
(1-5-3-3-2) regenerates the vector of one group of m dimension according to sequence order
In formula, i=1,2 ..., L-m,It represents by starting point of i using m to be long
The vector of degree, x0(i) it is its amplitude mean value;
(1-5-3-3-3) seeks vectorWithBetween distance
In formula, i, j=1,2 ..., L-m, j ≠ i,For vectorWithBetween distance, k is the sigtnal interval;
In formula, n is exponential function boundary gradient, and r is border width;
(1-5-3-3-5) function
(1-5-3-3-6) repeats the new vector that b~e generates one group of m+1 dimension according to sequence order reconstruct:
(1-5-3-3-7) is calculated
(1-5-3-3-8) calculates valuation fuzzy entropy because L is limited:
In formula, E (m, n, r, L) is valuation fuzzy entropy, and m is generation vector dimension, and n is exponential function boundary gradient, and r is side
Boundary's width;
(1-5-3-4) calculates zero crossing mZCI and zero crossing standard deviation nsZCI, the zero crossing mZCI and zero crossing standard
The calculating step of poor nsZCI is as follows:
It is { x (l) that (1-5-3-4-1), which sets length as the ultrasonic RF time serieses of L,:1≤l≤L};
(1-5-3-4-2) goes equalization to handle
(1-5-3-4-3) searches the zero passage points in { y (l) }, calculates the points between two adjacent zero crossings, is denoted as
d1,d2,…,dZ;
(1-5-3-4-4) sequence of calculation d1,d2,…,dZMean value mZCI and Normalized standard deviation nsZCI, calculation formula is such as
Under:
In formula,For ultrasonic RF time serieses amplitude mean value, Z is zero crossing interval maximum number;
(1-5-3-5) asks for M × kurtosis Kurtosis of N number of ultrasound RF time serieses, peak value Peak, fuzzy entropy, zero passage
The mean value of this 5 features of point mZCI and zero crossing standard deviation nsZCI obtains the temporal signatures of the sample.
In the step (2-3), using nicety of grading as the evaluation index of assessment recognition result.
The principle of the present invention:The present invention acquires ultrasonic echo radio frequency (Radio Frequency, RF) signal multiframe first;
It demodulates certain frame RF signals and shows Type B figure, region of interest (Region Of Interest, ROI) is chosen, to every point in ROI
Its preceding L frames echo RF signal is taken to form ultrasound RF time serieses;Ultrasonic RF time serieses extraction fractal characteristic, frequency based on structure
Sample set is divided into training set and test set by domain and temporal signatures, and grader is trained using training set, then that test set is defeated
Enter grader, tissue characterization result is assessed with the nicety of grading of grader.The present invention is based on ultrasonic RF time serieses, carry
The feature of three a point shape, frequency domain, time domain angles is taken, tissue is identified using grader, ultrasound medicine detection is can be widely applied to, surpasses
The fields such as sound tissue characterization.
Compared with prior art, the present invention having the following advantages that and advantageous effect:
1st, the present invention is based on ultrasonic RF time serieses, since the signal of ultrasonic RF time serieses is derived from same position, reduce
The error that the difference of ultrasonic wave propagation path and depth caused by subcutaneous fat thickness difference is brought, and do not need to depth attenuation
Compensation, determines sign precision so as to improve;
2nd, the present invention makes useful information more abundant from three shape, frequency domain, time domain angles is divided to carry out feature extraction;
3rd, the ultrasonic RF time serieses that the present invention is based on can obtain in conventional ultrasound diagnosis, will not generate extra expenses;
4th, whether the present invention uniformly has no particular/special requirement to tissue, and application range is more extensive.
Description of the drawings
Fig. 1 is the flow chart of the method for the present invention.
Fig. 2 is embodiment liver tissues of rats Type B figure.
Fig. 3 is liver tissues of rats ultrasound RF time series charts.
Fig. 4 is 5 classification random forest nicety of grading figure of test set rat, and " o " is true classification, and " * " is prediction classification.
Fig. 5 is 5 classification svm classifier precision figure of test set rat, and " o " is true classification, and " * " is prediction classification.
Specific embodiment
With reference to embodiment and attached drawing, the present invention is described in further detail, but embodiments of the present invention are unlimited
In this.
Embodiment
As shown in Figure 1, a kind of tissue characterization method based on ultrasonic radio frequency time series of the present embodiment, including walking as follows
Suddenly:
S1, structure ultrasound RF time serieses.
S1.1 uses the wideband line of the Sonix TOUCH and centre frequency 6.6MHz of Ultrasonix companies of Canada production
Hepatic tissue region under battle array ultrasonic probe scanning Wistar rat liver coatings, records multiple frames of ultrasonic echo RF signals.
S1.2 is demodulated the 100th frame data and shows its ultrasonic B figure, as shown in Figure 2.
S1.3 chooses the ROI that size is 70 × 20 on ultrasonic B figure, and the ultrasound that interception obtains 1400 points in ROI is returned
Wave RF signals take every in ROI its preceding 256 frame data to obtain 1400 points as 256 ultrasonic RF time serieses, surpass
Sound RF time series charts are as shown in Figure 3.
S2, feature extraction.
S2.1, fractal characteristic, the present embodiment employ 102 samples, wherein including 0 phase of degree of hepatic fibrosis sample 10
It is a, 1 26, phase sample, 2 45, phase samples, 3 15, phase samples, 46, phase samples.For 1400 ultrasound RF in each ROI
Time series calculates the Higuchi fractal dimensions of each RF time serieses first, is then averaging and obtains the sample
Higuchi fractal dimensions.The Higuchi fractal dimension solution procedures of single time series are as follows, and wherein L=256, k take 2~
16:
A) it is { x (l) to set length as the time series of L:1≤l≤L};
B) new time series is built by given time series { x (l) }
C) it calculatesLength Lm(k);
E) fitting a straight lineSlope is the Higuchi FRACTAL DIMENSIONs of ultrasound RF time serieses
Number;
S2.2, frequency domain character to 1400 ultrasound RF time serieses in sample ROI, make FFT transform, Ran Houqiu respectively
Spectrum amplitude mean value divided by maximum spectrum under identical frequency is taken to be worth to normalized spatial spectrum of the size in [0,1], to normalizing
Change frequency spectrum and carry out fitting a straight line, obtain slope S lope, the intercept Intercept of the samples normalization frequency spectrum.Meanwhile it calculates
The sum of 4 sections of spectrum energies to sample S1, S2, S3, S4;It is as follows to calculate step:
A) frequency spectrum X (w) is obtained as FFT transform to the ultrasonic RF time serieses { x (l) } that length is L;
B) average to frequency spectrum X (w) amplitudes under identical frequency Xave(w);
C) to Zave(w) normalized spatial spectrum is obtained as normalized
D) it is rightFitting a straight line is carried out, spectrum slope S lope, intercept Intercept is obtained, is asked according to the following formula
S1, S2, S3, S4:
Wherein L=256.
S2.3, temporal signatures to 1400 ultrasound RF time serieses in sample ROI, calculate each time series first
Kurtosis Kurtosis, peak value Peak, fuzzy entropy, zero crossing mZCI and zero crossing standard deviation nsZCI, be then averaging respectively
To the kurtosis Kurtosis of the sample, peak value Peak, fuzzy entropy, zero crossing mZCI and zero crossing standard deviation nsZCI.The single time
The kurtosis Kurtosis of sequence, peak value Peak, fuzzy entropy, zero crossing mZCI and zero crossing standard deviation nsZCI solution procedures are as follows:
2) peak value Peak is first 10 larger signal amplitude absolute value mean values in ultrasound RF time serieses;
3) fuzzy entropy calculating step is:
A) it is { x (l) to set length as the ultrasonic RF time serieses of L:1≤l≤L};
B) vector of one group of m dimension is regenerated according to sequence order
{x(i),x(i+1),…,x(i+m-1)}-x0(i),
C) vector is soughtWithBetween distance
Wherein, i, j=1,2 ..., L-m, j ≠ i
E) function
F) the new vector that b~e generates one group of m+1 dimension according to sequence order reconstruct is repeated;
G) it calculates
H) because L is limited, therefore valuation fuzzy entropy E (m, n, r, L) is calculated:
Wherein, L=256, m=n=2, r take 0.3 times of ultrasonic RF time serieses amplitude standard deviation;
4) zero crossing mZCI and zero crossing standard deviation nsZCI calculating step is:
A) it is { x (l) to set length as the ultrasonic RF time serieses of L:1≤l≥L};
C) the zero passage points in { y (l) } are searched, the points between two adjacent zero crossings is calculated, is denoted as d1,d2,…,dZ;
D) sequence of calculation d1,d2,…,dZMean value mZCI and Normalized standard deviation nsZCI:
Wherein, L=256.
S3, tissue identification and result.
In the present embodiment, organize a total of liver fibrosis 0,1,2,3,4 amount to 5 classifications, grader used be with
Machine forest and SVM, according to 7:Sample set is divided into training set and test set by 3 ratio, using training set feature vector to dividing
Class device is trained, and the parameter of random forest is set as:The number of tree is 300, and the depth of each tree is 6, iterations 150
Secondary, each selectable characteristic of node is 4;The parameter of SVM is set as:Kernel function is gaussian radial basis function, parameter g and
C is determined by ten folding cross validations;The feature vector of test set is finally inputted into trained grader, the results show that such as Fig. 4
Show, random forest nicety of grading has reached 96.6667%, and if Fig. 5 shows, SVM is up to that 93.3333%, Fig. 4 and Fig. 5 are shown
As a result it is feasible, effective to illustrate that the present invention is used for tissue identification.
Above-described embodiment is the preferable embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any Spirit Essences without departing from the present invention with made under principle change, modification, replacement, combine, simplification,
Equivalent substitute mode is should be, is included within protection scope of the present invention.
Claims (2)
- A kind of 1. tissue characterization method based on ultrasonic radio frequency time series, which is characterized in that include the following steps:Step 1, feature extraction:Each tissue samples are proceeded as follows, obtain corresponding feature vector;Step 2, tissue identification;The step 1 includes the following steps:(1-1) scans tissue, obtains its ultrasonic echo RF signal multiframes;(1-2) demodulates certain frame ultrasonic echo RF signal and shows Type B figure;(1-3) chooses the ROI that size is M × N on Type B figure;(1-4) takes every in ROI its preceding L frames RF signal to construct the ultrasonic RF time serieses that M × N number of length is L;Fractal characteristic, frequency domain and the temporal signatures of (1-5) extraction ultrasound RF time serieses are to get to its feature vector;The step 2 includes the following steps:Tissue samples are divided into test set and training set by (2-1) by a certain percentage;(2-2) trains grader using the feature vector of training set;(2-3) is with the assessment recognition result of grader under test set;In the step (1-5), the method for extracting ultrasound RF time series feature vectors is as follows:(1-5-1) fractal characteristic:The ultrasonic RF time serieses that the M × N number of length constructed in step (1-4) is L are calculated successively It is as follows to calculate step for fractal dimension:(1-5-1-1) note time series is { x (l):1≤l≤L }, x (l) represents the range value of l frame ultrasound RF time serieses;(1-5-1-2) is built new time series by given time series { x (l) }In formula, new time series starting pointK is signal Interval;(1-5-1-3) is calculatedLength Lm(k);In formula, Lm(k) it is newly-built time seriesLength of curve;(1-5-1-4) seeks Lm(k) mean valueIn formula, L (k) is length of curve mean value;(1-5-1-5) fitting a straight lineFrequency spectrum fitting a straight line slope is ultrasound RF time serieses Fractal dimension;(1-5-1-6) asks for the fractal dimension mean value of M × N number of ultrasound RF time serieses up to the fractal characteristic of the sample;(1-5-2) frequency domain character:It is extracted the sum of frequency spectrum fitting a straight line slope S lope, intercept Intercept and 4 sections of frequency spectrums S1, S2,S3,S4Totally 6 frequency domain characters, calculating step is:(1-5-2-1) obtains frequency spectrum X (w) to the ultrasonic RF time serieses { x (l) } that length is L as FFT transform;(1-5-2-2) averages to frequency spectrum X (w) amplitudes under identical frequency Xave(w);(1-5-2-3) is to Xave(w) normalized spatial spectrum is obtained as normalizedIn formula, | Xave(w) | it is Xave(w) absolute value, w are signal frequency;(1-5-2-4) is rightFitting a straight line is carried out, obtains frequency spectrum fitting a straight line slope S lope, intercept Intercept;RootS is sought according to formula1,S2,S3,S4:In formula, L is signal length,For normalized spatial spectrum;(1-5-3) temporal signatures:It is extracted kurtosis Kurtosis, peak value Peak, fuzzy entropy, zero crossing mZCI and zero crossing standard Poor nsZCI calculates this 5 features for the ultrasonic RF time serieses that the M × N number of length constructed in step (1-4) is L, step successively It is rapid as follows:(1-5-3-1) kurtosisIn formula, L is signal length,For ultrasonic RF time serieses amplitude mean value, x (l) represents l frame ultrasound RF time serieses Range value;(1-5-3-2) peak value Peak is the mean value of a absolute values by a relatively large margin of p before ultrasound RF time serieses;(1-5-3-3) calculates fuzzy entropy, and the calculating step of the fuzzy entropy is as follows:It is { x (l) that (1-5-3-3-1), which sets length as the ultrasonic RF time serieses of L,:1≤l≤L }, wherein, x (l) represents l frames and surpasses The range value of sound RF time serieses;(1-5-3-3-2) regenerates the vector of one group of m dimension according to sequence orderIn formula, It represents by starting point of i using m to be long The vector of degree, x0(i) it is its amplitude mean value;(1-5-3-3-3) seeks vectorWithBetween distanceIn formula, i, j=1,2 ..., L-m, j ≠ i,For vectorWithBetween distance, k is the sigtnal interval;(1-5-3-3-4) is definedWithSimilarityIn formula, n is exponential function boundary gradient, and r is border width;(1-5-3-3-5) function(1-5-3-3-6) repeats the new vector that 1-5-3-3-2 to 1-5-3-3-5 generates one group of m+1 dimension according to sequence order reconstruct:(1-5-3-3-7) is calculated(1-5-3-3-8) calculates valuation fuzzy entropy because L is limited:In formula, E (m, n, r, L) is valuation fuzzy entropy, and m is generation vector dimension, and n is exponential function boundary gradient, and r is wide for boundary Degree;(1-5-3-4) calculates zero crossing mZCI and zero crossing standard deviation nsZCI, the zero crossing mZCI and zero crossing standard deviation The calculating step of nsZCI is as follows:It is { x (l) that (1-5-3-4-1), which sets length as the ultrasonic RF time serieses of L,:1≤l≤L};(1-5-3-4-2) goes equalization to handle(1-5-3-4-3) searches the zero passage points in { y (l) }, calculates the points between two adjacent zero crossings, is denoted as d1, d2,…,dZ;(1-5-3-4-4) sequence of calculation d1,d2,…,dZMean value mZCI and Normalized standard deviation nsZCI, calculation formula it is as follows:In formula,For ultrasonic RF time serieses amplitude mean value, Z is zero crossing interval maximum number;(1-5-3-5) asks for kurtosis Kurtosis, peak value Peak, fuzzy entropy, the zero crossing of M × N number of ultrasound RF time serieses The mean value of this 5 features of mZCI and zero crossing standard deviation nsZCI obtains the temporal signatures of the sample.
- A kind of 2. tissue characterization method based on ultrasonic radio frequency time series according to claim 1, which is characterized in that institute It states in step (2-3), using nicety of grading as the evaluation index of assessment recognition result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510359391.0A CN105030279B (en) | 2015-06-24 | 2015-06-24 | A kind of tissue characterization method based on ultrasonic radio frequency time series |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510359391.0A CN105030279B (en) | 2015-06-24 | 2015-06-24 | A kind of tissue characterization method based on ultrasonic radio frequency time series |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105030279A CN105030279A (en) | 2015-11-11 |
CN105030279B true CN105030279B (en) | 2018-06-29 |
Family
ID=54437673
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510359391.0A Active CN105030279B (en) | 2015-06-24 | 2015-06-24 | A kind of tissue characterization method based on ultrasonic radio frequency time series |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105030279B (en) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10154826B2 (en) | 2013-07-17 | 2018-12-18 | Tissue Differentiation Intelligence, Llc | Device and method for identifying anatomical structures |
US10716536B2 (en) | 2013-07-17 | 2020-07-21 | Tissue Differentiation Intelligence, Llc | Identifying anatomical structures |
CN105447872A (en) * | 2015-12-03 | 2016-03-30 | 中山大学 | Method for automatically identifying liver tumor type in ultrasonic image |
US11701086B1 (en) | 2016-06-21 | 2023-07-18 | Tissue Differentiation Intelligence, Llc | Methods and systems for improved nerve detection |
CN106037799B (en) * | 2016-06-22 | 2019-04-09 | 华南理工大学 | Elastic parameter imaging method based on ultrasonic RF backscatter signal time frequency analysis |
CN106725592A (en) * | 2016-11-09 | 2017-05-31 | 飞依诺科技(苏州)有限公司 | The automatic identifying method and system of ultrasonic radio frequency metadata |
CN106491161B (en) * | 2016-11-15 | 2019-07-05 | 乐普(北京)医疗器械股份有限公司 | A kind of method and device of intelligent organization's identification |
CN107346541B (en) * | 2017-06-02 | 2020-02-18 | 华南理工大学 | Tissue characterization method based on ultrasonic radio frequency time series wavelet analysis |
CN107358156B (en) * | 2017-06-06 | 2020-05-19 | 华南理工大学 | Feature extraction method for ultrasonic tissue characterization based on Hilbert-Huang transform |
IT201900025303A1 (en) * | 2019-12-23 | 2021-06-23 | Sergio Casciaro | DEVICE AND METHOD FOR TISSUE CLASSIFICATION |
CN111855810B (en) * | 2020-07-20 | 2023-02-03 | 济南大学 | Rail foot damage identification method and system based on recurrent neural network |
CN114492519B (en) * | 2022-01-14 | 2023-08-29 | 西安交通大学 | Lung ultrasound special sign B line identification and classification method based on ultrasound echo radio frequency signals |
CN116147548B (en) * | 2023-04-19 | 2023-07-25 | 西南林业大学 | Nondestructive testing method and system for thickness of steel fiber RPC cover plate |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040034304A1 (en) * | 2001-12-21 | 2004-02-19 | Chikayoshi Sumi | Displacement measurement method and apparatus, strain measurement method and apparatus elasticity and visco-elasticity constants measurement apparatus, and the elasticity and visco-elasticity constants measurement apparatus-based treatment apparatus |
CN103479398A (en) * | 2013-09-16 | 2014-01-01 | 华南理工大学 | Method of detecting hepatic tissue microstructure based on ultrasonic radio frequency flow analysis |
CN103637821A (en) * | 2013-11-27 | 2014-03-19 | 华南理工大学 | Hepatic fibrosis degree Fisher identification method based on ultrasonic radio frequency (RF) time sequence |
-
2015
- 2015-06-24 CN CN201510359391.0A patent/CN105030279B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN105030279A (en) | 2015-11-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105030279B (en) | A kind of tissue characterization method based on ultrasonic radio frequency time series | |
CN103720489B (en) | Pathological tissues growth monitoring method and system | |
US6264609B1 (en) | Ultrasound apparatus and method for tissue characterization | |
Tsui et al. | Small-window parametric imaging based on information entropy for ultrasound tissue characterization | |
Lizzi et al. | Statistical framework for ultrasonic spectral parameter imaging | |
CN101667289B (en) | Retinal image segmentation method based on NSCT feature extraction and supervised classification | |
CN1954235B (en) | Improved method and device for local spectral analysis of an ultrasonic signal | |
Coster et al. | What starch grain is that?–A geometric morphometric approach to determining plant species origin | |
CN103479398A (en) | Method of detecting hepatic tissue microstructure based on ultrasonic radio frequency flow analysis | |
Guerriero et al. | Diagnosis of the most frequent benign ovarian cysts: is ultrasonography accurate and reproducible? | |
CN107358156A (en) | The feature extracting method of Ultrasonic tissue characterization based on Hilbert-Huang transform | |
CN107346541B (en) | Tissue characterization method based on ultrasonic radio frequency time series wavelet analysis | |
CN103637821A (en) | Hepatic fibrosis degree Fisher identification method based on ultrasonic radio frequency (RF) time sequence | |
Machado et al. | Characterization of in vitro healthy and pathological human liver tissue periodicity using backscattered ultrasound signals | |
CN106983524A (en) | A kind of parameter and its measuring method for reflecting that biological tissue is abnormal | |
CN110161119A (en) | Wind electricity blade defect identification method | |
Varghese et al. | Characterization of tissue microstructure scatterer distribution with spectral correlation | |
Rubert et al. | Scatterer number density considerations in reference phantom-based attenuation estimation | |
Narayanan et al. | Studies on ultrasonic scattering from quasi-periodic structures | |
Poonguzhali et al. | Automatic classification of focal lesions in ultrasound liver images using combined texture features | |
CN103637820A (en) | Hepatic fibrosis degree relative deviation identification method based on ultrasonic radio frequency (RF) time sequence | |
Granchi et al. | Multidimensional spectral analysis of the ultrasonic radiofrequency signal for characterization of media | |
CN114492519A (en) | Lung ultrasonic special sign B-line identification and classification method based on ultrasonic echo radio frequency signals | |
CN103854285A (en) | SAR image ground object cutting method based on random projection and improved spectral cluster | |
CN113907792B (en) | Nonlinear decorrelation fusion ultrasonic multi-parameter thermal coagulation detection and imaging method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |