WO2011155168A1 - 組織悪性腫瘍検出方法、組織悪性腫瘍検出装置 - Google Patents
組織悪性腫瘍検出方法、組織悪性腫瘍検出装置 Download PDFInfo
- Publication number
- WO2011155168A1 WO2011155168A1 PCT/JP2011/003156 JP2011003156W WO2011155168A1 WO 2011155168 A1 WO2011155168 A1 WO 2011155168A1 JP 2011003156 W JP2011003156 W JP 2011003156W WO 2011155168 A1 WO2011155168 A1 WO 2011155168A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- tissue
- beat
- pulsation
- waveform
- blocks
- Prior art date
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/08—Detecting organic movements or changes, e.g. tumours, cysts, swellings
- A61B8/0833—Detecting organic movements or changes, e.g. tumours, cysts, swellings involving detecting or locating foreign bodies or organic structures
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/48—Diagnostic techniques
- A61B8/485—Diagnostic techniques involving measuring strain or elastic properties
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/02—Measuring pulse or heart rate
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/5269—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves involving detection or reduction of artifacts
Definitions
- the present invention relates to a method for detecting tissue malignancy and the like, and more particularly to a method for detecting tissue malignancy and the like by ultrasound.
- CT and MRI typically detect the cancerous tissue using the rapid accumulation of contrast agent in extravascular space due to the increased permeability of neoplastic blood vessels.
- PET detects neoplastic tissue using the increased absorption of glucose molecules due to the increased metabolic demand of neoplastic cells.
- Ultrasound contrast agents utilize larger, typically gas-filled, bubbles to allow imaging of the neoplastic microcirculation.
- Patent Document 1 discloses a system for detecting early stage prostate cancer by combining tissue density and blood flow comparison.
- Patent Document 2 discloses a method of calculating the degree of newly proliferated microvessels in a tumor using power Doppler imaging and detecting the malignancy and metastatic ability of a cancerous tumor whose position is completely clear. It is done.
- Patent Document 3 discloses a method for detecting a tissue abnormality by comparing blood perfusion in a tissue before and after applying a heat source on the premise that heat perfusion to the tissue is increased by heat. ing.
- Tumor angiogenesis creates new blood vessels with a structure that is different from that of normal tissue defects. These new blood vessels associated with cancerous tumors are characterized by perivascular separation, vasodilation, and irregular shapes. Tumorous blood vessels do not have smooth muscle cells present in blood vessels of normal tissue and are not well formed to oxygenate all tissues. Furthermore, neoplastic blood vessels are also more porous and leaky than blood vessels of normal tissue.
- an object of the present invention is to provide a tissue malignant tumor detection method for more accurately determining a malignant tumor by more appropriately detecting the characteristics of the beating of a cancerous tumor.
- the tissue malignancy detection method is a tissue malignancy detection method for detecting a malignancy contained in the tissue by a scan signal obtained by scanning the tissue with ultrasound, and the tissue is scanned
- Block division step of dividing the divided area into a plurality of blocks, and a tissue beat, which is a time change of the displacement of the tissue caused by the pulsation of the tissue for each of the plurality of blocks
- a tissue beat estimation step of estimating based on the beats
- a beat related feature extraction step of extracting, from the tissue beats, a plurality of beat related features that are parameters related to the beat of the tissue for each of the plurality of blocks
- a distribution characteristic calculating step of calculating distribution characteristics of the plurality of pulsation related features for each of the plurality of blocks; Based on each of the plurality of blocks, and a malignant classification step of classifying whether malignant block or not a block including malignant tumors.
- a plurality of feature amounts corresponding to each of a plurality of effects of angiogenesis by the malignant tumor on the pulsation of the tissue can be calculated. From the plurality of feature quantities calculated in this manner, blocks included in the scan area can be classified into malignant blocks having a high probability of including malignant tumors and the other blocks. As a result, malignant tumors can be more accurately determined by more appropriately detecting the characteristics of the beating of cancerous tumors.
- the present invention can not only be realized as such a tissue malignant tumor detection method, but also be realized as a tissue malignant tumor detection device or a program that causes a computer to execute characteristic steps in the tissue malignant tumor detection method.
- a program can be distributed via a recording medium such as a compact disc read only memory (CD-ROM) and a transmission medium such as the Internet.
- CD-ROM compact disc read only memory
- the present invention is realized as a semiconductor integrated circuit (LSI) that realizes part or all of the function of such a tissue malignancy detection device, or a tissue malignancy detection system including such a tissue malignancy detection device Can be realized as
- LSI semiconductor integrated circuit
- FIG. 1 is a diagram showing functional blocks of a tissue malignant tumor detection apparatus according to an embodiment of the present invention.
- FIG. 2 is a flowchart showing the process flow of the tissue malignant tumor detection apparatus according to the embodiment of the present invention.
- FIG. 3 is a diagram showing an example of the configuration of a tissue beat estimation unit according to the embodiment of the present invention.
- FIG. 4 is a diagram showing an example of the specification of ultrasound raw RF data in the embodiment of the present invention.
- FIG. 5 is a diagram showing an example of the configuration of the pre-processing unit according to the embodiment of the present invention.
- FIG. 6 is a diagram showing an example of the configuration of the main component extraction unit according to the embodiment of the present invention.
- FIG. 1 is a diagram showing functional blocks of a tissue malignant tumor detection apparatus according to an embodiment of the present invention.
- FIG. 2 is a flowchart showing the process flow of the tissue malignant tumor detection apparatus according to the embodiment of the present invention.
- FIG. 3 is a diagram showing an
- FIG. 7 is a diagram showing an example of processing results by the extreme value identification unit and the pulsation adjustment unit according to the embodiment of the present invention.
- FIG. 8 is a diagram showing an example of the configuration of the amplitude feature calculation unit according to the embodiment of the present invention.
- FIG. 9 is a diagram showing an example of the configuration of the shape feature calculation unit according to the embodiment of the present invention.
- FIG. 10 is a diagram showing an example of the configuration of a cardiac cycle identification unit according to the embodiment of the present invention.
- FIG. 11 is a diagram showing an example of processing performed by the critical point identification unit according to the embodiment of the present invention.
- FIG. 12 is a diagram showing an example of a beating shape feature of a heartbeat according to the embodiment of the present invention.
- FIG. 13 shows a comparison of beating shape features of pulsatile and normal tissue beats.
- FIG. 14 is a diagram showing an example of the configuration of the main component extraction unit according to the embodiment of the present invention.
- FIG. 15 is a diagram showing an example of the result of each process of segmentation, tapering and representative extraction in the main component extraction unit according to the embodiment of the present invention.
- FIG. 16 is a diagram showing an example of the configuration of the spatial distribution calculation unit according to the embodiment of the present invention.
- FIG. 17 is a diagram showing an example of calculation processing of the gray level simultaneous occurrence probability matrix in the embodiment of the present invention.
- FIG. 18 is a diagram showing an example of the configuration of a tumor localization unit according to an embodiment of the present invention.
- FIG. 19 is a diagram for explaining an example of processing performed by the tumor localization unit in the embodiment of the present invention.
- FIG. 20 is a diagram showing an example of the configuration of a cardiac cycle delay calculation unit according to the embodiment of the present invention.
- FIG. 21 is a diagram showing an example of the configuration of the delay calculation unit according to the embodiment of the present invention.
- FIG. 22 is a diagram showing an example of a configuration of an autoregression coefficient calculation unit according to an embodiment of the present invention.
- FIG. 23 is a diagram showing an example of an output result of the tissue malignant tumor detection device according to the embodiment of the present invention when there is no tumor in the tissue.
- FIG. 20 is a diagram showing an example of the configuration of a cardiac cycle delay calculation unit according to the embodiment of the present invention.
- FIG. 21 is a diagram showing an example of the configuration of the delay calculation unit according to the embodiment of the present invention.
- FIG. 22 is a diagram showing an example of a configuration of an autoregression coefficient calculation unit according to an embodiment of the
- FIG. 24 is a diagram showing an example of an output result of the tissue malignant tumor detection apparatus according to the embodiment of the present invention when there is a tumor in a tissue.
- FIG. 25 is a block diagram showing a hardware configuration of a computer system for realizing the tissue malignant tumor detection apparatus according to the embodiment of the present invention.
- the tissue malignancy detection method is a tissue malignancy detection method for detecting a malignancy contained in the tissue by a scan signal obtained by scanning the tissue with ultrasound, and the tissue is scanned
- Block division step of dividing the divided area into a plurality of blocks, and a tissue beat, which is a time change of the displacement of the tissue caused by the pulsation of the tissue for each of the plurality of blocks
- a tissue beat estimation step of estimating based on the beats
- a beat related feature extraction step of extracting, from the tissue beats, a plurality of beat related features that are parameters related to the beat of the tissue for each of the plurality of blocks
- a distribution characteristic calculating step of calculating distribution characteristics of the plurality of pulsation related features for each of the plurality of blocks; Based on each of the plurality of blocks, and a malignant classification step of classifying whether malignant block or not a block including malignant tumors.
- a plurality of feature amounts corresponding to each of a plurality of effects of angiogenesis by the malignant tumor on the pulsation of the tissue can be calculated. From the plurality of feature quantities calculated in this manner, blocks included in the scan area can be classified into malignant blocks having a high probability of including malignant tumors and the other blocks. As a result, malignant tumors can be more accurately determined by more appropriately detecting the characteristics of the beating of cancerous tumors.
- a tumor localization step of identifying the position of a cancerous tumor may be included based on the block classified as the malignant block in the malignant classification step.
- the position of the malignant tumor can be accurately determined by associating the position of the malignant block with the tissue.
- the tissue pulsation estimation step calculates a tissue displacement which is a displacement at a spatial position of the tissue from the scan signal, and a spatial displacement of the tissue displacement from the calculated tissue displacement. Even if it includes a tissue strain calculating step of calculating a tissue strain which is a gradient, and a pulsating waveform generating step of generating a pulsating waveform which is a waveform of the tissue beat as the tissue displacement versus time or the tissue strain versus time. Good.
- a pre-processing step may be included to extract a component of the estimated tissue beat due to the heartbeat caused by the heartbeat.
- the pre-processing step further includes, among the estimated tissue beats, a cardiac power calculating step of calculating cardiac power which is power related to heart beat, and a magnitude of the cardiac power.
- Heart rate clustering step of clustering the region in the tissue, an extreme value identification step of identifying an extreme value of the amplitude of the heartbeat in the region of the tissue clustered, and the tissue beat based on the extreme value
- a heartbeat adjustment step of adjusting the amplitude of the pulsation waveform which is the waveform of
- the extreme value identification step further includes an extreme value point identification step of identifying peaks and valleys in the pulsation waveform, and an interference whose magnitude of deviation from other peaks is larger than a predetermined threshold. And an outlier rejection step for rejecting outliers of the amplitude of the pulsation waveform in order to eliminate interference valleys whose magnitudes of peaks and deviations from other valleys are larger than a predetermined threshold. It is also good.
- the pre-processing step further includes, from the pulsation waveform, a portion corresponding to an interference peak whose magnitude of deviation from the other peak is larger than a predetermined threshold, and a deviation from the other valley.
- the method may further include a beat adjustment step of adjusting the amplitude of the beat waveform so as to align in a direction.
- the tissue beat may be estimated for all scan points of each block included in the plurality of blocks.
- the tissue beat is estimated as one or several representative beats of each block included in the plurality of blocks, or a combination of the representative beats. It is also good.
- data obtained from a plurality of scan points can be comprehensively judged to determine the presence or absence of a malignant tumor.
- the beat related feature includes a beat amplitude feature that is a feature amount for the amplitude of the tissue beat, a beat shape feature that is a feature amount for the shape of the waveform of the tissue beat, and the tissue beat. It may be at least one of the pulse time features that are feature quantities for the time change of the movement waveform.
- the beating shape feature is L1 / L2 which is a ratio of a contraction period (L1) which is a period of a contraction portion of a cardiac cycle and an expansion period (L2) which is a period of an expansion portion of a cardiac cycle.
- a contraction midpoint which is a point on the contraction curve at which the amplitude is equal to a predetermined ratio with respect to the maximum amplitude
- the amplitude is previously defined with respect to the maximum amplitude Period L3 which is the ratio of the period (L3) between the expansion midpoint which is a point on the expansion curve which becomes equal to the ratio and the cycle of heartbeat (L4), and the peak of the amplitude in the contraction period
- Deviation of the expansion curve from a predefined curve connecting a contraction peak and an expansion end point which is the end point of the expansion period, a skewness representing asymmetry of heart beat, and a sharpness of the contraction peak Kurtosis and the expansion curve Among the deviation extreme that may be at least one.
- the beat shape feature includes a heart rate period calculating step of calculating a heart cycle from the tissue beat, a critical point identification step of identifying a critical point using the heart cycle, the heart cycle and the critical point,
- the shape feature calculating step may be calculated in the shape feature calculating step including the shape feature extracting step of extracting the pulsation shape feature on the basis of.
- the critical point of the cardiac cycle is a contraction start point which is a start point of a contracted part of the cardiac cycle, an expansion end point which is an end point of an extended part of the cardiac cycle, and an amplitude peak in the contraction period.
- Amplitude at the systolic midpoint which is the point on the systolic curve at which the amplitude is equal to a predefined ratio to the maximum amplitude during the systolic peak and the systolic portion of the cardiac cycle, and during the dilated portion of the cardiac cycle May include an extended midpoint, which is a point on the extended curve equaling a predefined ratio to the maximum amplitude.
- the critical point identification step whether the tissue beat has an upward contraction curve based on the search step for searching for the minimum point and the maximum point in the tissue beat, and the minimum point and the maximum point Pulsating direction identification step of identifying a pulsating direction indicating whether the squeezing curve has a downward curve, the critical point in the pulsating waveform using the maximum point, the minimum point, and the pulsing direction And a critical point determining step of determining
- the predefined curve may be a straight line.
- the beat related feature extraction step when the beat has an upward contraction curve, a positive differential sum between the expanded curve and the previously defined curve is calculated as the deviation, and the beat is calculated. If the movement has a downward contraction curve, a negative difference sum between the expansion curve and the predefined curve may be calculated as the deviation.
- the beat time characteristic is at least one of the delay of the heart cycle, the difference of the heart waveform which is the waveform of the heart beat, and the autoregression coefficient of the waveform of the tissue beat. Good.
- the delay of the cardiac cycle includes a reference cardiac cycle determining step of determining a reference cardiac cycle which is a reference cardiac cycle, and a delay calculating step of calculating a delay of a target cardiac cycle with respect to the reference cardiac cycle. It may be calculated in the cardiac cycle delay calculation step including.
- a cardiac cycle having the largest amplitude among the scan data may be selected as the reference cardiac cycle.
- the reference cardiac cycle may be determined from an electrocardiogram (ECG) signal.
- ECG electrocardiogram
- the difference between the cardiac waveforms can be calculated by calculating a reference cardiac waveform which is a reference cardiac waveform, and calculating a difference between each of a plurality of cardiac waveforms due to pulsation and the reference cardiac waveform.
- Delay calculation including a difference calculating step, and an entire difference calculating step of calculating an entire cardiac waveform difference value which is a value representing a difference between the plurality of cardiac waveforms and the reference cardiac waveform from the plurality of calculated differences. It may be calculated in the step.
- the overall cardiac waveform difference value may be a standard deviation of the calculated plurality of differences.
- the autoregression coefficient may be a beat resampling step of tapering a plurality of the beat waveforms so that the cardiac cycle is in the same period, and the beat tapered with a predetermined autoregressive model.
- the calculation may be performed in an autoregression coefficient calculation step including an autoregression operation step of obtaining an autoregression coefficient which is a coefficient of the autoregression model based on the dynamic waveform.
- the distribution characteristic may be at least one of a spatial distribution parameter and a statistical distribution parameter.
- the spatial distribution parameter includes energy of the beat related feature, entropy of the beat related feature, contrast of the beat related feature, homogeneity of the beat related feature, and It may be at least one of the correlations of the beat related features.
- the statistical distribution parameter includes an average value of the beat related features, a median value of the beat related features, a maximum value of the beat related features, a minimum value of the beat related features, and It may be at least one of a standard deviation of a beat related feature, a kurtosis of the beat related feature, and a skewness of the beat related feature.
- the pulsation related feature and the distribution characteristic thereof may be the median value, the entropy, the standard deviation, the mean value, and the maximum value of the pulsation amplitudes of the plurality of scan points included in each of the plurality of blocks.
- the tumor localization step further includes a target area identification step of defining a target area for each scan point of the scanned tissue, and a block belonging to the target area among the plurality of blocks.
- the method may include a tumor block division step, and a cancer probability calculation step of calculating the probability that the tissue is cancer based on the classification result of the block belonging to the target region according to the malignant classification step.
- the tumor localization step may further include an imaging step of displaying the probability of being a cancer at a scan point of the scanned tissue in a two-dimensional or three-dimensional image.
- a tissue malignancy detection apparatus for detecting a malignancy contained in the tissue according to a scan signal obtained by scanning the tissue with ultrasound, which comprises A block division unit configured to divide a scanned area into a plurality of blocks, and a tissue beat, which is a temporal change of displacement of the tissue caused by pulsation of the tissue, for each of the plurality of blocks; A tissue beat estimation unit for estimating based on the beats, and a beat related feature extraction unit for extracting, from the tissue beat, a plurality of beat related features that are parameters related to the beat of the tissue for each of the plurality of blocks; A distribution characteristic calculator configured to calculate distribution characteristics of the plurality of pulsation related features for each of the plurality of blocks, and the plurality of the plurality of pulsation related features based on the distribution characteristic.
- Each lock includes a malignant classification unit for classifying whether malignant block or not a block including malignant tumors.
- a tumor localization unit may be provided that identifies the position of a cancerous tumor based on the block classified as the malignant block in the malignant classification unit.
- angiogenesis by a cancerous tumor and the beating of the cancerous tumor produced thereby will be described in detail.
- angiogenesis Cancerous tumors need to supply themselves with nutrients and oxygen, and create new blood vessels to remove the waste products generated as a result of metabolism.
- the process of creating such new blood vessels is called angiogenesis. It is well known that in the early stages of cancerous tumors, angiogenesis is required as it grows beyond a diameter of about 2-3 millimeters. These new blood vessels grow around and inside the tumor, and with the increase of blood vessels, cancer cells reach the main blood vessels in the body and, as a result, the possibility of metastasis by these newly created blood vessels Also increases. Detecting angiogenesis may lead to detection of early stage cancerous tumors and may even lead to prediction of metastasis probability after treatment.
- cancerous tumors also called malignant tumors
- cancerous tumors differ from normal tissues or benign tumors in many respects.
- angiogenesis occurs at the level of the capillary bed.
- a beat in the tissue is created, resulting in the beating of the tissue by blood perfusion.
- the beating amplitude of cancerous tissue will be increased compared to surrounding normal tissue.
- blood inflow to and from cancerous tumors will increase.
- these new blood vessels associated with cancerous tumors are characterized by perivascular separation, vasodilation and irregular shape.
- Tumorous blood vessels do not have smooth muscle cells present in blood vessels of normal tissue and are not well formed to oxygenate all tissues.
- neoplastic blood vessels are also more porous and leaky than blood vessels of normal tissue.
- the beat waveform of the cancerous tumor is expected to have a more expanded shape than the beat of normal tissue, for example, if the spatial position is different and the beat pattern is different.
- the above difference in the microvasculature is also a difference in the appearance timing of the pulsation waveform, and exhibits the characteristic of time fluctuation. This also results in differences in the distribution characteristics of the statistically and spatially represented beats.
- cancerous tumor tissue and normal tissue can be classified by identifying differences in the beats detected from the tissue with blood perfusion.
- the main parameters from the beat are (1) beat amplitude, (2) beat shape, (3) so that the grade of the grade of the scanned tissue can be determined. 2.) Use the beat time characteristics and (4) their distribution characteristics as feature quantities.
- the vasculature of the tumor is more porous and leaked than normal vasculature and cancerous microvessels have no smooth muscle, they are in a state of vasodilation.
- the features extracted from the waveform ie, L1 / L2, L3 / L4, skewness, kurtosis, dilation of the dilation curve, and differences with respect to the pulsation shape measured by the diversion of the extrema of the dilation curve It may occur.
- L1 / L2 is the ratio of the contraction period (L1) of the pulsation to the expansion period (L2), and L1 / L2 and skewness indicate how much the contraction peak is delayed. The greater the delay, the more likely the beat is from a cancerous tumor.
- L3 / L4 is, for example, the ratio of the pulsation width (L3) to the total pulsation period (L4) at half height, and L3 / L4 and kurtosis are how wide and extended the pulsation waveform is Indicates what The more you expand, the more likely it is a cancerous beat.
- the deviation of the expansion curve and the deviation of the extreme value of the expansion curve represent the presence or absence of the overlapping ridge in the pulsation waveform and the curvature of the expansion curve. If there are no double bumps, it indicates the possibility of cancerous beats.
- the pulsation time feature is a feature value for the temporal change of the pulsation waveform, which represents how the pulsation changes with time. Specifically, it refers to the delay of cardiac cycle (also referred to as cardiac cycle), cardiac waveform difference, and autoregression coefficient.
- the spatial and statistical characteristics of the above features in specific tissue regions are also effective in expressing the basic pulsatile distribution, with differences in microvasculature, how the pulsatility is in the region Indicates whether it is distributed.
- classification results are combined to calculate a malignant score of the internal region of the scanned tissue, and these scores can be used to show the doctor or the patient the probability that the cancerous tumor exists in the scanned tissue.
- the medical imaging device may comprise some or all of the following components. Signal transceivers, data acquisition, data processing, and displays.
- the present invention teaches a method implemented in a data processing component of a medical imaging device. The input of such data processing component is a scan signal acquired from the data acquisition component, and the output is shown to the user through the display component. It is understood that other components of such medical imaging equipment are suitable for applications using the present invention.
- FIG. 1 is a diagram showing functional blocks of a tissue malignant tumor detection apparatus 90 according to an embodiment of the present invention.
- scannedSig indicates scan data received from a diagnostic imaging device that is used to detect beats in the scanned tissue.
- a diagnostic imaging device that is used to detect beats in the scanned tissue.
- An example of such a device is a medical ultrasound device in which scannedSig is IQ data indicating changes in radio frequency (RF) data or the amplitude and phase of a sine wave obtained by demodulating the RF data.
- the operation of the ultrasound instrument is based on transmitting high frequency ultrasound pulses towards the medium to be scanned, after which the pulses interact with the underlying structures along the path while reflecting and scattering, And receive the scattered signal. That is, scannedSig is a signal indicating the reflected wave of the ultrasonic pulse transmitted toward the tissue to be scanned. These signals are used to estimate morphological and functional features of underlying structures.
- One of the advantages of using ultrasound as a medical imaging technique is the ability to generate high frame rate images and detect small movements in tissue.
- the tissue malignancy detection apparatus 90 is a tissue malignancy detection apparatus for detecting a malignancy contained in a tissue by a scan signal obtained by scanning the tissue with ultrasonic waves, More specifically, a block division unit 100, a tissue beat estimation unit 101, a preprocessing unit 102, a beat related feature extraction unit 103, a distribution characteristic calculation unit 104, and a malignant classification unit 105 are provided.
- the preprocessing unit 102 is optional and is included for the purpose of illustration. Embodiments without the pretreatment unit 102 are also within the scope of the present invention.
- the block division unit 100 represents processing of dividing a scan area, which is an area obtained by scanning a tissue by ultrasonic waves, into tissue blocks (hereinafter also referred to as blocks) having a specific shape and size according to an application example.
- the shape of the block may be, but is not limited to, a square, a rectangle, a polygon or a circle.
- the size can be selected to be 1 millimeter, it is not limited to this.
- the tissue beat estimation unit 101 estimates a beat of the basic scanned tissue (hereinafter also referred to as a tissue beat) from the received scannedSig. That is, the tissue beat estimation unit 101 estimates, for each of the plurality of blocks, the tissue beat, which is a temporal change of the displacement of the tissue caused by the pulsation of the tissue, based on the scan signal.
- each block includes one or more scan signals.
- the output of the tissue beat estimation unit 101 is rawPulse, which is a signal representing a raw tissue beat.
- Parameters that can be used to represent beats in tissue include, but are not limited to, tissue displacement and tissue distortion.
- Tissue displacement is estimated as the absolute movement of tissue according to the original position at zero time.
- Tissue distortion is estimated as relative displacement between parts in tissue. That is, tissue strain is a spatial gradient of tissue displacement. Either tissue displacement versus time or tissue strain versus time at a particular tissue point (i.e., a scan signal obtained from a point in the tissue to be scanned) can be selected as representing tissue beat at that point.
- tissue pulsation is a temporal change of displacement at a spatial position of tissue. The tissue pulsation may include both tissue distortion and tissue displacement.
- the preprocessing unit 102 eliminates interference (so-called noise) in the estimated original beat.
- Raw tissue beats may include various types of tissue motion.
- One such movement is the heartbeat as a result of blood flow through the blood vessels according to the cardiac cycle of the heart.
- Another movement is the patient's respiratory cycle, which is several times slower than the cardiac cycle, usually much larger in amplitude compared to the heart beat. This movement is not important in the present invention.
- Other environmental and electronic interference may also occur.
- the pre-processing unit 102 outputs a pure beat with suppressed interference, that is, procPulse. When the interference included in the rawPulse output from the tissue beat estimation unit 101 is small, the tissue malignant tumor detection device 90 achieves the same effect even without the preprocessing unit 102.
- the beat related feature extraction unit 103 can extract features even without the pre-processing unit 102.
- the provision of the pre-processing unit 102 enables more accurate detection of a malignant tumor.
- the pulsation related feature extraction unit 103 performs, for each of a plurality of blocks, calculation processing for extracting a plurality of types of pulsation related features, which are parameters related to a pulsation waveform in the time domain, from tissue pulsations.
- the output of this block is pulseFeature, which is an input signal to the distribution characteristic calculation unit 104 and the malignant classification unit 105.
- the distribution characteristic calculation unit 104 calculates a parameter related to the distribution characteristic which is a representative characteristic of each of the blocks of the extracted characteristic.
- the output of the distribution characteristic calculation unit 104 is distributionFeature, which is also an input to the malignant classification unit 105.
- the malignant classification unit 105 calculates information necessary to classify an area in a tissue into benign or malignant, and classifies each of a plurality of blocks whether it is a malignant block which is a block including a malignant tumor or not.
- the output of this malignant classification unit 105 is maligScores.
- FIG. 2 is a flowchart showing the process flow of the tissue malignant tumor detection apparatus 90.
- the block division unit 100 divides the scan area indicated by the scan signal into blocks (S100).
- the tissue beat estimation unit 101 estimates a tissue beat for each of the plurality of blocks (S101).
- the pulsation related feature extraction unit 103 extracts, from the estimated tissue pulsation, a pulsation related feature representing a feature of the tissue beat in the time domain (S103).
- the distribution characteristic calculation unit 104 calculates a distribution characteristic indicating how the values of the pulsation related feature are distributed in each of the plurality of blocks in the scan area (S104).
- the malignant classification unit 105 determines the grade of the tumor located in the region in the tissue corresponding to each block, and classifies whether the tumor is benign or malignant (S105).
- the block division unit 100 in FIG. 1 divides a region obtained by scanning a tissue by ultrasound into tissue blocks which are smaller regions. All further processing takes place on this tissue block.
- the shape and size of the tissue block is determined by the practical application. The shape can be selected from a square, a rectangle, or a circle, but is not limited thereto. Size can choose 1 millimeter. According to the inventor's experiments, rectangular tissue blocks about 1 to 2 millimeters in size are effective.
- the output of block division is blocks which is information indicating the configuration of a tissue block.
- an imaging device provided with an ultrasound transducer scans tissue with ultrasound and uses raw ultrasound data, which is a scan signal output, to process unprocessed from tissue (ie, Estimate the beat before the interference removal process is applied.
- FIG. 3 illustrates a configuration for estimating tissue beats represented by tissue strain using ultrasound.
- the tissue pulsation estimation unit 101 includes a tissue displacement estimation unit 200 and a tissue distortion estimation unit 201.
- the ultrasonic raw RF data which is the output of the ultrasonic transducer is , ScannedSig (d, l, f).
- d indicates a depth direction parallel to the scan direction, also referred to as fast time
- l indicates a line direction orthogonal to the scan direction
- f indicates time or a frame
- slow time be called.
- the tissue displacement estimation unit 200 acquires scannedSig (d, l, f) as an input, and calculates disp (d, l, f) representing tissue displacement.
- Various methods of calculating displacement from raw ultrasound raw RF data include, but are not limited to, autocorrelation and cross correlation.
- the tissue distortion estimation unit 201 acquires disp (d, l, f) representing tissue displacement as an input, and calculates tissue distortion.
- tissue distortion As an example of the method of calculating the tissue strain, it is possible to calculate the spatial gradient of displacement along the depth direction, that is, calculating the derivative of disp (d, l, f) with respect to d.
- the specific depth d plotted against time and the tissue strain calculated at the specific line l indicate the tissue pulsing rawPulse (d, l, f) at a certain tissue point. Each tissue point is described by a corresponding depth and line. Thus, this is described by the particular beat waveform in rawPulse (d, l, f). This beat waveform includes any type of movement or interference that may occur in tissue.
- the preprocessing unit 102 of FIG. 1 can perform processing on the unprocessed beat rawPulse to remove interference and facilitate extraction of important beats.
- the pre-processing unit 102 is optional, and embodiments not including this block are also included in the scope of the present invention.
- the output of preprocessing is procPulse.
- FIG. 1 The configuration of such a pre-processing unit 102 is shown in FIG.
- the preprocessing unit 102 includes a noise filtering unit 400 and a main component extraction unit 401.
- the configuration of the preprocessing unit 102 is not limited to these. This is for illustrative purposes only and the steps performed by these components and their substeps can be performed in any order or combination, and any of these steps are omitted without affecting other steps. And any of these steps may be included in any of the other blocks of FIG. 1, and such an embodiment is also within the scope of the present invention.
- the noise filtering unit 400 is one of the components of the preprocessing unit 102 that may be implemented.
- the frequency of the heartbeat is 1-2 Hz and the frequency of the breathing cycle is less than 0.5 Hz. Therefore, by configuring the noise filtering unit 400 as a band pass filter that allows only the heartbeat frequency of 1 to 2 Hz to pass, both the environmental high frequency vibration noise and the low frequency breathing movement are characterized as heartbeat characteristics. Can be removed efficiently without affecting the
- the main component extraction unit 401 is an example of a component of another possible preprocessing unit 102, and the main beat component in the selected important area (ie, the beat included in the tissue beat). Extract the main components of the This is effective when the noise filtering unit 400 can not completely suppress interference, particularly irregular interference such as body movement.
- the input of this block may be a raw beat or a beat that has already been subjected to other pre-processing steps.
- the output of this block may be input to another preprocessing step, or may be used as it is as input to the pulsation related feature extraction unit 103 and the distribution characteristic calculation unit 104.
- the main component extraction unit 401 can perform main component extraction by using principal component analysis (PCA;) or independent component analysis (ICA;), but is not limited thereto.
- PCA principal component analysis
- ICA independent component analysis
- PCA or ICA takes beats in a predefined area and is most likely to beat uncorrelated (in the case of PCA) or independent (in the case of ICA) to other components (ie interference). Extract various components. This is done at the cost of reducing spatial resolution. In this way important key components can be selected.
- FIG. 6 is a block diagram showing another example of the main component extraction unit 401.
- the important beats include only the heart beat. That is, the main component extraction unit 401 extracts the heartbeat from the tissue beat.
- the main component extraction unit 401 has a cardiac power calculation unit 500, a heart rate clustering unit 501, an extreme value identification unit 502, and a pulsation adjustment unit 503.
- the cardiac power calculation unit 500 calculates signal power related to heart beat, that is, cardiacPow (d, l) from rawPulse (d, l, f) which is an unprocessed beat.
- the frequency of the heartbeat is known (indicated by fc, which is usually 1 to 2 Hz for humans).
- cardiacPow (d, l) is calculated as the value of the power spectrum component of the raw beat associated with fc.
- it is calculated as cardiacPow (d, l) as the value of the power spectrum component of the raw beat related to fc normalized by the total sum or partial sum of the power of all components .
- the heart rate clustering unit 501 identifies an area in tissue where the presence of amplitude or heart beat is high, that is, cardiac cluster (d, l).
- cardiac Cluster (d, l) is identified as an area in tissue having cardiac power of a magnitude within a predefined range.
- the cardiac Cluster (d, l) is identified to include a region of higher cardiac power within the tissue to occupy a predefined proportion to the entire tissue.
- the extremum identification unit 502 determines the peaks and valleys (ie, extrema points) of the beating waveform in the region within the tissue identified by the cardiac Cluster (d, l), and such peaks and beats belonging to the heartbeat. Output a valley, ie, cardiacExtrema (d, l). That is, the extreme value identifying unit 502 identifies the extreme value of the amplitude of the heartbeat in the region within the clustered tissue.
- One example of an embodiment for determining peaks and valleys is to estimate the first and second derivatives of the beating waveform with respect to time.
- the peaks represented as pulsePeaks (d, l) are identified at time instances where the first derivative is zero and the second derivative is negative
- the valleys represented as pulseTroughs (d, l) are:
- the first derivative is identified at time instances where the first derivative is zero and the second derivative is positive.
- Examples of peaks are the points 603, 604, 605 and 606 with reference to FIG. 7A
- the example of the valley is the point indicated by the points 600, 607, 601 and 602.
- Peaks and valleys that belong to the heart beat are assumed to be more regular than peaks and valleys of interference and other movements, so outlier rejection can be used as a way to perform this task, but It is not limited to
- the example of the peak and the valley of the heartbeat when rejecting the outliers are point 603, point 607, point 605, point 606, and the example of the interference peak and the valley of interference are point 600, point 604, point 601, point 602 It is. That is, the extreme value identifying unit 502 is an extreme point identifying unit (not shown) for identifying peaks and valleys in the pulsation waveform, and the magnitude of deviation from other peaks is larger than a predetermined threshold value.
- An outlier rejection unit (not shown) that rejects outliers in the amplitude of the pulsation waveform in order to eliminate interference peaks and interference valleys whose displacements from other valleys are greater than a predetermined threshold May be included.
- the interference peak (also referred to as interference peak) refers to a peak whose peak size is larger than a predetermined threshold value than the other peak sizes.
- the valley of interference (also referred to as an interference valley) refers to a valley having a valley size larger than a predetermined threshold value than the sizes of the other valleys.
- the heartbeat adjusting unit 503 acquires cardiac cluster (d, l) and cardiac extrema (d, l) as information for adjusting rawPulse (d, l, f) so as to have only a heartbeat.
- the output of this block is procPulse (d, l).
- the effect of the processing by the pulse adjustment unit 503 is shown in FIG.
- the length of time corresponding to only the heartbeat is identified from the peaks and valleys of the heart.
- the period corresponding to the interference is excluded from the beat.
- the amplitudes of the peak and the valley of the heartbeat are aligned in the time direction (graph 611), and the pulsation waveform is further added.
- the amplitude of is adjusted. That is, the pulsation adjusting unit 503 adjusts the magnitude of the amplitude of the pulsation waveform so that the magnitudes of the minimum value and the maximum value of the pulsation waveform become uniform.
- the peak amplitude and the valley amplitude used when making the amplitudes uniform are, for example, the average amplitudes of the peak and the valley shown by the broken line 609 and the broken line 610 in FIG. Absent. All other points that are not peaks and valleys are adjusted by the method of linear mapping, but are not limited to this.
- the fragmented heartbeat times are combined.
- the resulting beat is represented as a graph 612 with reference to FIG. 7 (D).
- Such a process ensures that the end of the fragmented heartbeat period must be of the same type (i.e., peaks to valleys or valleys) as the beginning of the next fragmented heartbeat period.
- segmentation, tapering and averaging are used to obtain representative beats of the tissue block. This is illustrated in FIG.
- the main component extraction unit 401 includes a heartbeat selection unit 1300, a beat tapering unit 1301, and a representative beat extraction unit 1302. Have.
- the heart beat selection unit 1300 receives the beat rawPulse (d, l, f) and selects a heart beat based on a certain reference. As an example of the selection criteria, one may consider that the signal energy in the cardiac frequency region is higher than in the other regions. As another example of the selection criteria, one may consider that the signal spectral flatness of the heart frequency region is lower than a threshold. If a heart beat is not selected, that is, the input beat does not meet the selection criteria, the heart beat may be zero.
- the output of the heartbeat selection unit 1300 is cardiac pulse (d, l, f).
- the pulsation tapering unit 1301 receives the cardiac pulse (d, l, f), gradually reduces each cardiac cycle, and makes them the same period so that these cardiac cycles can be correctly compared.
- the output of this block is taperedCardiacPulse (d, l, f).
- the representative beat extraction unit 1302 has taperedCardiacPulse (d, l, f) as an input, and calculates a representative beat at the tissue point from the tapered cardiac pulse of the beat waveform.
- taperedCardiacPulse d, l, f
- One example of such a calculation is a calculation that averages tapered heart pulses over several periods.
- Other examples of such calculations include those using PCA or ICA.
- FIG. 1 An example of signal segmentation, tapering and representative extraction is shown in FIG.
- the cardiac pulse is adjusted so that the waveform amplitude is either all positive or all negative.
- cardiac pulses of different tissue points are divided into continuous cardiac cycles according to the start and end points of the reference beat (reference beat).
- the result is shown as a graph 1400 of FIG.
- An example of a reference beat is a strong cardiac pulse in a scan plane, and another example of a reference beat can be an ECG signal.
- each cardiac cycle is gradually reduced (tapered) to have the same period by the window function.
- the result is shown as a graph 1402 in FIG.
- This window function is a customized Hann window 1401 shown in FIG. 15 (B).
- the window function may be any other function such as a Hamming window.
- the processed pulsation may be calculated at all scan points (that is, measurement points) included in each tissue block, There may be one or several representative beats of each tissue block. Also, such a combination of representative beats may be used.
- beat related feature extraction unit 103 of FIG. Extract motion related features.
- beat related features may be all or a subset of beat amplitude features, beat shape features, and beat time features. These are collectively called pulseFeature.
- the beat related feature extraction unit 103 may calculate a beat amplitude feature that is a feature amount for the amplitude of the beat waveform.
- Cancerous tumors need to create new blood vessels to supply nutrition and oxygen. With this increase in microvasculature, blood perfusion can cause the beat amplitude around the tissue and its distribution (statistical and spatial) to be higher than that of normal tissue.
- FIG. 7 An example of the configuration of the amplitude feature calculation unit 710 that calculates the amplitude feature of the pulsation is shown in FIG.
- FIG. 8 shows functional blocks of an amplitude feature calculation unit 710 provided in the beat related feature extraction unit 103 that calculates a beat amplitude feature.
- the amplitude feature calculation unit 710 includes a pulsation amplitude calculation unit 700 and an amplitude histogram calculation unit 701.
- the heartbeat amplitude calculator 700 calculates the cardiac beat amplitude at each tissue point designated by d and l, that is, cardiacAmplitude (d, l).
- cardiacAmplitudes (d, l) calculated in all tissue points or tissue blocks are grouped into various predefined bins.
- the amplitude histogram calculation unit 701 outputs “1” in the case of the tissue point or tissue block in which the pulsation amplitude belongs to bin, and outputs “0” in the other case.
- the amplitude range of each bin can be set individually, and the range of one bin does not affect the range of other bins. Both the value of cardiacAmplitude (d, l) and the value of the clustering result can be used as an output of the amplitude feature calculation unit 710.
- bins are [0.5, 1], which hypothetically corresponds to the neoplastic beat amplitude for tissue strain.
- Another example is [0.1, 0.5], which hypothetically corresponds to the beat amplitude of normal tissue for tissue strain.
- Other bins may be defined based on certain assumptions about the beat amplitude, and such an embodiment is also within the scope of the present invention.
- the amplitude histogram calculator 701 divides the entire scanned tissue into small regions, calculates an amplitude histogram for each small region, and the histogram follows a predefined pattern.
- a small area may be output.
- An example of tissue division is a rectangular grid.
- One example of an amplitude histogram is the probability distribution of beat amplitudes across a predefined set of bins. The range of each histogram bin can be set individually, and the range of one bin does not affect the range of other bins.
- An example of the predefined pattern is that the probability distribution is biased to the range of [0.5, 1].
- the pulsation related feature extraction unit 103 may calculate a pulsation shape feature that is a feature amount of the shape of the waveform of the tissue pulsation.
- New blood vessels associated with cancerous tissue include perivascular separation, vasodilation, and irregular shapes. Tumorous blood vessels do not have smooth muscle cells present in blood vessels of normal tissues, and are not sufficiently formed to provide oxygen and nutrients to all cancerous tissues. Furthermore, neoplastic blood vessels are also more porous and leaky than blood vessels of normal tissue. These factors cause neoplastic beats to have a more extended shape than normal tissue beats.
- FIG. 9 shows functional blocks of a shape feature calculation unit 711 included in the beat related feature extraction unit 103 that calculates a beat shape feature.
- the shape feature calculation unit 711 has a cardiac cycle identification unit 800 and a shape feature extraction unit 801.
- the cardiac cycle identification unit 800 identifies each cardiac cycle from procPulse (d, l, f).
- the critical point of each cardiac cycle, criticalPoints (d, l), can be extracted from this block.
- One embodiment of a cardiac cycle identification method is shown in FIG.
- the cardiac cycle identification unit 800 in this embodiment has a heartbeat period calculation unit 900 and a critical point identification unit 901.
- the heartbeat period calculation unit 900 calculates an average heartbeat period of the input beat from procPulse (d, l, f).
- the heartbeat period calculation unit 900 may calculate the cardiac cycle of each input beat.
- One example of this calculation is to convert the input beat into the frequency domain, search the basic frequency component, and select it as the average heartbeat period.
- the average distance between two adjacent peaks or two adjacent valleys is calculated in the pulsation adjustment unit 503, and the average distance is calculated as an average cardiac cycle (that is, an average cardiac cycle). It may be selected as
- the output of the heartbeat period calculation unit 900 is cardiacPeriod (d, l), where d indicates depth and l indicates line.
- the critical point identification unit 901 identifies the critical point of the cardiac cycle at each measurement point.
- the critical point may be, for example, a contraction start point, an expansion end point, a contraction peak, a contraction middle point, an expansion middle point, and the like, but is not limited thereto.
- the inputs of this block include procPulse (d, l, f) and cardiacPeriod (d, l), and the output of this block is criticalPoints (d, l).
- d indicates the depth and l indicates the line.
- the first step of the process performed by the critical point identification unit 901 is to search for local maxima and minima in procPulse (d, l).
- An example of this step is the following procedure. (1) Search the first maximum point 1000 within the cardiac cycle from time 0 seconds, (2) search the first minimum point 1001 from the time corresponding to the first maximum point 1000 within the cardiac cycle, 3) The second maximum point 1002 is searched from the first minimum point 1001 within the cardiac cycle, and (4) the desired number of maximum points and minimum points are searched, the above (1) to (3) Process it repeatedly.
- the second step of the process performed by the heartbeat period calculation unit 900 is to determine the direction of the heartbeat.
- the direction of the beat is either upward (with an upward contraction curve followed by a downward expansion curve) or downward (with an upward expansion curve after a downward contraction curve) throughout one cardiac cycle.
- the duration of the contraction portion of the heart beat is shorter than the duration of the dilation portion.
- the contraction curve is a waveform of the beat during the contraction period of the heart
- the dilation curve is the waveform of the beat during the expansion period of the heart.
- Point a and point c are two consecutive local maxima and point b is a local minimum between point a and c.
- the slopes of line ab and line bc may take an average of a predefined number of cardiac cycles. If the slope of line ab is greater than the slope of line bc, then the direction of the beat is downward and point a is the beginning of this cardiac cycle. If it is smaller, the direction of pulsation is upward, and point b is the start of one cardiac cycle.
- D1 is a period from the first maximum point 1000 to the first minimum point 1001 adjacent to the first maximum point 1000.
- D2 is a period from the first minimum point 1001 described above to the second maximum point 1002.
- D1 and D2 may take an average of a predefined number of cardiac cycles. If D2 is greater than D1, then the direction of pulsation is downward, and if smaller, the direction of pulsation is upward.
- the final step of the process performed by the critical point identification unit 901 is to identify critical points in each cardiac cycle.
- the critical points include, but are not limited to, contraction start point, expansion end point, contraction peak, contraction midpoint, and expansion midpoint.
- the contraction midpoint and the expansion midpoint are two points in the contraction curve and the expansion curve when the amplitude of the beat is equal to the pre-defined ratio to the maximum pulsation amplitude. An example of this ratio is one-half of the highest beat amplitude.
- the contraction start point is the first maximum point 1000.
- the extension end point is the next second maximum point 1002.
- the contraction peak is a first minimum point 1001.
- the contraction midpoint 1004 and the expansion midpoint 1005 are two points d and e on the contraction curve and the expansion curve corresponding to the amplitude f at which bf / bg is a predetermined value.
- the shape feature extraction unit 801 extracts useful features related to the waveform shape of the pulsation. This feature is described with reference to FIG.
- Characteristics related to the waveform shape of pulsation include L1 / L2, L3 / L4, pulsation distortion, pulsation kurtosis, deviation of the extended curve bc with respect to the straight line bc, and deviation of the extreme value in the expanded curve bc, etc. Included but not limited to them.
- the straight line bc may be a predetermined curve.
- L2 (d, l) is the duration of the dilation of the cardiac cycle. That is, in the case of the upward cardiac cycle, it is a period from the maximum point 1100 to the expansion end point 1101 with reference to FIG. 12 (A), and in the case of the downward cardiac cycle, FIG. 12 (B). , A period from the minimum point 1102 to the expansion end point 1103.
- L1 (d, l) is the duration of the contraction portion of the cardiac cycle. That is, in the case of the upward cardiac cycle, it is a period from the contraction start point 1104 to the maximum point 1100 with reference to FIG. 12 (A), and in the case of the downward cardiac cycle, referring to FIG. 12 (B). It is a period from the contraction start point 1105 to the minimum point 1102.
- neoplastic beats are likely to have delayed contraction peaks or prolonged contraction periods.
- L3 (d, l) is defined as the amplitude of the period of time from the contraction midpoint to the expansion midpoint when the amplitude of the beat is a predetermined ratio of the highest amplitude.
- the contraction midpoint is a point on the contraction curve at which the amplitude of the contraction curve is equal to a predetermined ratio with respect to the maximum amplitude during the contraction period.
- the expansion midpoint is a point on the expansion curve at which the amplitude of the expansion curve becomes equal to the ratio defined in advance with respect to the maximum amplitude during the expansion period.
- L4 (d, l) is a period of one pulse.
- the ratio L3 / L4 is used to measure the sharpness of the pulse. Tumorous beats are likely to have a wider, or more extended shape than normal beats. If L3 / L4 is higher, it means that the pulsation shape is more expanded. This means that the corresponding beat is more likely to be a neoplastic beat.
- Skewness is a measure of cardiac asymmetry and is expressed as pulse Skewness (d, l).
- a method of calculating pulse Skewness (d, l) there is a method using Equation 1 below.
- N is the number of frames in the cardiac cycle
- ⁇ and ⁇ are the mean and standard deviation of the pulsation amplitude within a cardiac cycle.
- L1 / L2 the shorter the contraction period and the longer the expansion period, the beat is likely to be a normal beat. Greater positive skewness indicates that the diastolic period of the cardiac cycle is longer than the systolic period of the cardiac cycle, indicating a lower probability of being neoplastic beats.
- Pulstosis is a measure of the sharpness of the contraction peak relative to the normal distribution and is expressed as pulse Kurtosis (d, l).
- pulse Kurtosis (d, l)
- the sum of the difference between the amplitudes of the extended curve bc and the difference between the amplitudes of the straight line bc is calculated as the deviation of the extended curve bc with respect to the straight line bc. This is calculated by the following Equation 3 in consideration of the heartbeat direction called pulseCurveDeviation (d, l).
- pulseCurveDeviation (d, l) is positive, it means that the width of the pulsation curve is wider, and if it is negative, it means the opposite. From this calculation, it can be understood that if the pulsation curve 1106 has a positive pulse curve deviation (d, l), the width is wider, and if the pulsation 1 107 has a negative pulse curve deviation (d, l) It can be seen that the width is not very wide. A wider beat is more likely to be a neoplastic beat.
- the difference in pulsation amplitude between the maximum value and the minimum value in the expanded curve bc is calculated as the deviation of the extremum in the expanded curve bc, and is referred to as pulseExtremaDeviation (d, l). If the expansion curve bc has no extrema, the deviation of the extremum is zero.
- the vasodilating state of the microvasculature in cancerous tissue eliminates the presence of overlapping bumps in the beat.
- a higher pulse Extrema Deviation (d, l) indicates that there is a high probability of the presence of overlapping ridges and a lower probability of neoplastic beats.
- FIG. 1 An example of the comparison of the neoplastic beat feature with the normal tissue beat feature is shown in FIG.
- any feature amount for example, L1 / L2, L3 / L4, pulse Skewness (d, l), pulse Kurtosis (d, l), pulse Curve Deviation (d, l), pulse Extrema Deviation
- the value of at least one of (d, l) may be normalized by a predefined value (eg, a value obtained from a region known to be a normal tissue), and may be predefined It may be normalized to a range (eg, [0, 1]).
- the value of any feature quantity or the value obtained by normalizing the value of the feature quantity is also set such that the higher the value, the higher the probability of being a neoplastic beat.
- the values of the original beat-related features ie, the values of the beat-related features before being normalized
- the values of the normalized beat-related features ie, the values of the normalized beat-related features
- the inverse of the beat-related features are also collectively referred to as “feature values”.
- the feature values are used to segment the scanned area of tissue in terms of high and low neoplastic beat rates.
- An embodiment of this segmentation is to set a threshold value for each feature and to classify regions according to high and low probability of neoplastic beat based on these threshold values, but it is not limited thereto.
- One example of threshold selection is to use an average value of feature values.
- the shape feature extraction unit 801 As the output pulseShapeFeatures of the shape feature extraction unit 801, it is possible to select a feature value or a clustering result based on the feature value.
- the beat related feature extraction unit 103 may calculate a beat time feature which is a feature of the time axis of the waveform of the tissue beat. Due to the different amplitude and shape of the beating, the entire beating waveform of the cancerous tumor is different from that of normal tissue. Furthermore, when the resistance to blood flow differs depending on the microvasculature structure, the arrival time of the beat will be different in cancerous tissue and normal tissue.
- the heartbeat time feature is, for example, at least one of cardiac cycle delay, cardiac waveform difference, and cardiac waveform autoregression coefficient. These are collectively called pulse Temporal Features.
- FIG. 20 illustrates an example of functional blocks of a cardiac cycle delay calculation unit 731 included in the pulsation related feature extraction unit 103 that calculates the delay amount of the cardiac cycle.
- the cardiac cycle delay calculating unit 731 has a reference cardiac cycle determining unit 1900 and a delay calculating unit 1901.
- the input to the cardiac cycle delay calculator 731 is a pulse, that is, procPulse (d, l, f).
- the output is then the delay of each beat (consisting of the cardiac cycle) relative to the reference beat (or reference cardiac cycle), ie, carDelay (d, l).
- Delay of the cardiac cycle means vascular resistance to blood flow.
- the reference cardiac cycle determination unit 1900 detects a reference beat of a reference beat or a cardiac cycle that is compared with all the beats.
- An example of such criteria is an electrocardiogram (ECG) signal. This represents the exact beginning of the cardiac cycle and the contraction peak. If no ECG signal is available, another example of such a criterion is the strongest beat in the scanned tissue.
- ECG electrocardiogram
- the output of the reference cardiac cycle determination unit 1900 is refCarCycle (f).
- the delay calculation unit 1901 calculates the delay of the cardiac cycle start point or the contraction peak in procPulse (d, l, f) relative to refCarCycle (f) as a reference. Before the calculation, resampling may be performed to standardize the cardiac cycle length so that it can be compared with refCarCycle (f).
- FIG. 19 An example of a functional block of the delay calculation unit 1901 that calculates a difference in cardiac waveform is shown in FIG.
- the inputs to the delay calculation unit 1901 are pulsation procPulse (d, l, f) and carDelay (d, l), and the output is the difference value of the cardiac cycle in the pulsation, that is, carDiff (d, l) is there.
- carDiff carDiff
- the delay calculation unit 1901 includes an individual difference calculation unit 2001 and an entire difference calculation unit 2002.
- the reference cardiac waveform calculation unit 2000 calculates a reference cardiac waveform to be compared with all the acquired cardiac waveforms.
- a reference cardiac waveform is calculated as the average of all acquired cardiac waveforms.
- such a reference cardiac waveform is calculated as the strongest cardiac waveform of all acquired cardiac waveforms.
- such reference cardiac waveforms may be defined without considering acquired cardiac waveforms, such as, for example, known cardiac waveforms cited from the literature.
- the output of this block is refCarWaveform (f).
- the individual difference calculation unit 2001 calculates the difference between the cardiac waveform and the reference cardiac waveform refCarWaveform (f) for each cardiac waveform of each pulsation procPulse (d, l, f). Before performing this calculation, these cardiac waveforms may be gradually reduced to the same length.
- the difference calculated by the individual difference calculation unit 2001 the sum (or integral value) of the absolute values of the differences between the beats procPulse (d, l, f) and the refCarWaveform (f) can be mentioned. Another example of such a difference is the root mean square difference.
- the overall difference calculation unit 2002 determines one value of the overall difference for each beat from the values of all the individual differences calculated by the individual difference calculation unit 2001.
- One example of such an overall difference value is to use the standard deviation of the individual difference values.
- FIG. 22 illustrates an example of functional blocks of an autoregression coefficient calculation unit 732 included in the pulsation related feature extraction unit 103 that calculates an autoregression coefficient of a waveform representing a heartbeat.
- the autoregression coefficient calculation unit 732 has a pulsation resampling unit 2100 and an autoregression operation unit 2101.
- any waveform can be described as an autoregressive process using autoregressive coefficients. That is, the current sample is a linear combination of past samples. Thus, these autoregressive coefficients may be used to describe the beat.
- the input to the autoregression coefficient calculation unit 732 is a beat procPulse (d, l, f), and the output is a model coefficient, that is, arCoeffs (d, l).
- the beat resampling unit 2100 resamples beats so that each cardiac cycle includes a standard number of samples. For example, at a frame rate of 40 frames per second (ie, the sampling rate in the time domain is 40 Hz), for a heart rate of 1 beat per second, 40 samples are included in one cardiac cycle. However, this number may vary, for example, if the configuration of the ultrasound transducer is different and the physiological conditions are different. If the number is different from 40 samples, the beat is resampled so that 40 samples are included in one cardiac cycle. The output of this block is resampPulse (d, l, f).
- the autoregressive equation calculation unit 2101 obtains a solution of the autoregressive equation (that is, an autoregressive coefficient). This applies to the beat.
- Equation 4 the autoregressive equation
- Xt is a current sample
- Xt-i is a past sample
- p is a model order
- ⁇ i is a model coefficient
- ⁇ t white noise.
- the solution of the model coefficients is the output feature, i.e. arCoeffs (d, l).
- the pulse-related feature extraction unit 103 may output, as its output, all or a subset of pulseAmpFeatures, pulseShapeFeaturres, and pulseTemporalFeatures.
- the distribution characteristic calculation unit 104 of FIG. 1 calculates distribution characteristics of these features in each tissue block or across a plurality of tissue blocks.
- the irregular shape of the microvasculature in cancerous tumors can be quantified by such distribution characteristics calculated from B-mode gray scale or beat related features.
- the calculated pulsation-related derived from the beating of the microvessels It is more preferable to calculate the distribution characteristic from the characteristic. This output is a distribution feature, or distributionFeature.
- the distribution feature is considered to use at least one of a statistical distribution parameter and a spatial distribution parameter.
- the statistical distribution parameter is, for example, an average value of feature values (ie, beat-related features), a median value of feature values, a maximum value of feature values, a minimum value of feature values, a standard deviation of feature values , The kurtosis of the value of the feature, and the skewness of the value of the feature.
- the spatial distribution parameter is, for example, at least one of energy of feature value, entropy of feature value, contrast of feature value, homogeneity of feature value, and correlation of feature value. The calculations for all these parameters are shown in FIG.
- FIG. 16 shows functional blocks of the spatial distribution calculation unit 740 included in the distribution characteristic calculation unit 104 that calculates the spatial distribution parameter.
- the spatial distribution calculation unit 740 has a gray level simultaneous occurrence probability matrix calculation unit 1500 and a spatial feature calculation unit 1501.
- the gray level simultaneous occurrence probability matrix calculating unit 1500 calculates how often the feature of the value i occurs in the specific spatial relationship with the feature of the value j in the block.
- Such specific spatial relationships are the various directions and distances, and are called offsets.
- offset specific four directions (horizontal, vertical, two diagonal directions) and distances in each direction as shown in FIG. 17 described later can be mentioned.
- a spatial feature calculator 1501 derives an average statistical feature from the co-occurrence probability matrix. This feature includes, but is not limited to, at least one of the following.
- Contrast (to measure local variations in the co-occurrence probability matrix). This is a measure of local variation. If the value of the feature is spatially larger and violently changed, the contrast will be higher. In such cases, the homogeneity is lower.
- Homogeneity (diagonal elements of co-occurrence probability matrix). The smaller the range of feature values, the higher the homogeneity. In such cases, the homogeneity will be higher.
- Correlation measures the joint probability that a specific feature set occurs. If the feature values are spatially closer to the linear structure, the correlation is higher.
- Entropy Statistical measure of feature randomness. This is a measure of the homogeneity of the feature's co-occurrence probability matrix. If the matrix elements are equal, then the entropy is maximized, meaning that the degree of change is the larger feature value.
- Equations 5 to 9 The equations for calculating spatial features described above are expressed as Equations 5 to 9 below.
- P d is a gray level co-occurrence probability matrix
- an entry (i, j) of P d is the number of occurrences of gray level pairs of i and j separated by a distance d.
- sigma x and sigma y respectively, the standard deviation of P d (x) and P d (y), the mu x and mu y, respectively, the average value of P d (x) and P d (y) is there.
- P d (x) and P d (y) are the following Equation 10 and Equation 11, respectively.
- FIG. 1500 An example of calculation processing of the gray level simultaneous occurrence probability matrix (also referred to as GLMC) performed by the gray level simultaneous occurrence probability matrix calculating unit 1500 is shown in FIG.
- the element (1, 1) has a value for the output of the GLCM shown in FIG. 17C. 1 is included. This is because there is only one instance in the input image shown in FIG. 17B in which the values of two horizontally adjacent pixels are 1 and 1, respectively. Further, in the output of the GLCM shown in FIG. 17C, the value (2) is included in the element (1, 2). This is because there are two instances in which the values of two horizontally adjacent pixels are 1 and 2 in the input image shown in FIG. 17B. Furthermore, in the output of the GLCM shown in FIG.
- the element (1, 3) contains the value 0. This is because instances in which the values of two pixels adjacent in the horizontal direction are 1 and 3 do not exist in the input image shown in FIG. 17 (B).
- the gray level co-occurrence probability matrix calculating unit 1500 continues processing the input image, scans the input image for another set of pixels (i, j), and records the sum in the corresponding element of GLCM. .
- the number of features calculated using the methods described herein may vary from less than one hundred to several hundred, depending on the application and the particular implementation. In many applications where such real time operation is required, such multiple features may be undesirable.
- Feature ranking algorithms can be used to rank features and to select top features for further classification tasks.
- the criteria for ranking features may be class separation criteria, which describe how the features are divided into cancerous data groups and normal data groups. The better the features are separated, the higher their rank.
- classification algorithms to evaluate performance may be used to intuitively group features and select the best configuration.
- a median value of L1 / L2 of a plurality of measurement points included in each block is a median value of L1 / L2 of a plurality of measurement points included in each block.
- the malignant classification unit 105 in FIG. 1 combines the output of the pulsation related feature extraction unit 103 and the output of the distribution characteristic calculation unit 104 to calculate malignant information of the scanned tissue, that is, maligScores. This is done using a classification algorithm.
- classification algorithms include, but are not limited to, AdaBoost and Support Vector Machine. Features may be ranked and selected prior to malignant classification. Such embodiments are also within the scope of the present invention.
- the classification algorithm takes the calculated feature as an input for each block (pulseFeature, distributionFeature), and outputs a malignant score indicating whether the block is normal or malignant under a predetermined setting.
- the example of the preset setting is to select a malignant score, regardless of whether the feature used in classification is selected or the parameter used by the algorithm to calculate the intermediate output value from the feature value. It may be a threshold used by the algorithm to determine from intermediate output values.
- Such predetermined settings are obtained through a training process that involves experimentation to collect real data with known characteristics (ie, information about normal and malignant tissue is known), and the classification algorithm is Train with data and characteristics.
- the predetermined setting can be determined by, for example, a learning algorithm in which a feature amount labeled in advance with at least one of a malignant tumor and a benign tumor or a normal tissue is a teacher data.
- a learning algorithm in which a feature amount labeled in advance with at least one of a malignant tumor and a benign tumor or a normal tissue is a teacher data.
- the malignant classification unit 105 does not necessarily use a learning algorithm. For example, a block having an outlier may be classified as a malignant block using a quartile value or the like, and an average value and a median value may be used. And so on may be classified as malignant blocks.
- the classification results may be further processed to obtain the cancer probability of the scanned area and to locate the cancerous tumor.
- An example of a functional block of the tumor localization unit included in the malignant classification unit 105 for identifying the position of the tumor is shown in FIG.
- the tumor localization unit 750 includes a tumor block division unit 1700, a cancer probability calculation unit 1701, and a thresholding unit 1702. The processing that these perform is further illustrated in FIG.
- the tumor block division unit 1700 defines a target area for calculating the cancer probability for each tissue point of the scanned tissue, and finds a tissue block belonging to the target area.
- the shape of the target area may be square, and its size may be 5 millimeters by 5 millimeters (see target areas 1800 and 1801 in FIG. 19A).
- the shape of the target area may be a cube, and its size may be 5 mm ⁇ 5 mm ⁇ 5 mm. Tissue blocks included in the defined region of interest are selected for further calculation.
- the cancer probability calculation unit 1701 counts the number of malignant blocks (as a classification result) in the selected tissue blocks to determine the cancer probability.
- the cancer probability calculation unit 1701 considers the distribution of malignant blocks. For example, clusters in which malignant blocks are continuous may be calculated as having a higher probability of cancer than clusters in which the malignant blocks are randomly distributed.
- the cancer probability can be regarded as the output of the tumor localization unit 750, and the probability value may be displayed as a color image to facilitate examination.
- the thresholding unit 1702 selects a threshold and compares the probability calculated by the cancer probability calculating unit 1701 with the threshold to determine which tissue point is cancerous and which tissue point is normal. Do. Tumor position is defined as a collection of cancerous tissue points based on this result.
- 23 and 24 show the results of animal experiments conducted to evaluate the performance of the tissue malignant tumor detection apparatus 90 according to the present invention.
- six consecutive ultrasound scan planes at 1 mm intervals are shown.
- the calculated cancer probability, the threshold result, and the corresponding B-mode image are displayed.
- FIG. 23 shows the results without tumor.
- An image 2200 shown in FIG. 23A shows the cancer probability calculated using the method according to the embodiment of the present invention, where white indicates higher cancer probability and black indicates lower cancer probability.
- the image 2201 shown in FIG. 23 (B) shows the position of the tumor after thresholding, and the white shows the position of the cancerous tumor.
- An image 2202 shown in FIG. 23C shows an actual B-mode image. From the image 2200 and the image 2201, the tissue malignant tumor detection device 90 according to the embodiment of the present invention can accurately show that there is no tumor in the tissue.
- FIG. 24 shows the result in the presence of a tumor, the actual position of which is shown in the image 2302 shown in FIG. 24 (C) as a white circle.
- the image 2300 shown in FIG. 24C shows the calculated cancer probability
- the image 2301 shown in FIG. 24C shows the position of the tumor after thresholding. From these two images, it can be seen that the tissue malignant tumor detection apparatus 90 according to the embodiment of the present invention accurately indicated the position of the tumor present in the tissue.
- the inventive steps described herein are for determining the grade of the scanned tissue.
- it may be considered to determine the position where the tumor is likely to exist in the tissue or to determine the grade of the selected tissue region without prior knowledge of the scanned tissue.
- Such information can also be used in parallel with or derived from other methods that are also used in such usage, or can be used to derive it.
- Such other methods include ultrasound B-mode imaging, ultrasound Doppler imaging, elasticity measurement, computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), and There is photoacoustic tomography (PAT).
- CT computed tomography
- MRI magnetic resonance imaging
- PET positron emission tomography
- PAT photoacoustic tomography
- One example of such an embodiment includes identifying a mass of tissue scanned by another method and using the present invention to determine the grade of the identified mass.
- the beat related feature is at least one of a beat amplitude feature, a beat shape feature, and a beat time feature.
- the beat related features include at least a beat shape feature or a beat time feature.
- FIG. 25 is a block diagram showing a hardware configuration of a computer system for realizing the tissue malignant tumor detection apparatus 90. As shown in FIG.
- the tissue malignant tumor detection apparatus 90 is implemented by a computer 34, a keyboard 36 and a mouse 38 for giving an instruction to the computer 34, a display 32 for presenting information such as calculation results of the computer 34, and the computer 34. It includes a CD-ROM (Compact Disc-Read Only Memory) device 40 for reading a program and a communication modem (not shown).
- a computer 34 a keyboard 36 and a mouse 38 for giving an instruction to the computer 34, a display 32 for presenting information such as calculation results of the computer 34, and the computer 34.
- It includes a CD-ROM (Compact Disc-Read Only Memory) device 40 for reading a program and a communication modem (not shown).
- CD-ROM Compact Disc-Read Only Memory
- a program which is a process performed by the tissue malignant tumor detection apparatus 90 is stored in a computer readable medium, CD-ROM 42, and read by the CD-ROM apparatus 40. Alternatively, it is read by the communication modem 52 through a computer network.
- the computer 34 includes a central processing unit (CPU) 44, a read only memory (ROM) 46, a random access memory (RAM) 48, a hard disk 50, a communication modem 52, and a bus 54.
- CPU central processing unit
- ROM read only memory
- RAM random access memory
- the CPU 44 executes the program read via the CD-ROM device 40 or the communication modem 52.
- the ROM 46 stores programs and data necessary for the operation of the computer 34.
- the RAM 48 stores data such as parameters at the time of program execution.
- the hard disk 50 stores programs, data, and the like.
- the communication modem 52 communicates with other computers via a computer network.
- the bus 54 mutually connects the CPU 44, the ROM 46, the RAM 48, the hard disk 50, the communication modem 52, the display 32, the keyboard 36, the mouse 38 and the CD-ROM device 40 to one another.
- the system LSI is a super-multifunctional LSI manufactured by integrating a plurality of components on one chip, and more specifically, a computer system including a microprocessor, a ROM, a RAM, and the like. .
- a computer program is stored in the RAM.
- the system LSI achieves its functions by the microprocessor operating according to the computer program.
- IC card or module is a computer system including a microprocessor, a ROM, a RAM, and the like.
- the IC card or module may include the above-described ultra-multifunctional LSI.
- the IC card or module achieves its functions by the microprocessor operating according to the computer program. This IC card or this module may be tamper resistant.
- the present invention may be the method described above. Further, the present invention may be a computer program that realizes these methods by a computer, or may be a digital signal composed of the computer program.
- the present invention is a computer-readable recording medium that can read the computer program or the digital signal, such as a flexible disk, a hard disk, a CD-ROM, an MO, a DVD, a DVD-ROM, a DVD-RAM, a BD (Blu-ray Disc (Registered trademark), a memory card such as a USB memory or an SD card, or a semiconductor memory may be used. Further, the present invention may be the digital signal recorded on these recording media.
- the computer program or the digital signal may be transmitted via a telecommunication line, a wireless or wired communication line, a network represented by the Internet, data broadcasting, and the like.
- the present invention may be a computer system comprising a microprocessor and a memory, wherein the memory stores the computer program, and the microprocessor operates according to the computer program.
- the computer program may be implemented by another independent computer system by recording and transporting the program or the digital signal on the recording medium, or by transporting the program via the network or the like.
- the present invention is applicable to tissue malignant tumor detection methods and the like, and in particular, it can be applied to tissue malignant tumor detection methods and the like by ultrasound.
Abstract
Description
34 コンピュータ
36 キーボード
38 マウス
40 CD-ROM装置
42 CD-ROM
44 CPU
46 ROM
48 RAM
50 ハードディスク
52 通信モデム
54 バス
90 組織悪性腫瘍検出装置
100 ブロック分割部
101 組織拍動推定部
102 前処理部
103 拍動関連特徴抽出部
104 分布特性算出部
105 悪性分類部
200 組織変位推定部
201 組織歪み推定部
400 ノイズフィルタリング部
401 主要コンポーネント抽出部
500 心電力算出部
501 心拍クラスタリング部
502 極値識別部
503 拍動調整部
700 拍動振幅算出部
701 振幅ヒストグラム算出部
710 振幅特徴算出部
711 形状特徴算出部
731 心周期遅延算出部
732 自己回帰係数算出部
740 空間分布算出部
750 腫瘍位置特定部
800 心周期識別部
801 形状特徴抽出部
900 心拍期間算出部
901 臨界点識別部
1300 心拍動選択部
1301 拍動テーパリング部
1302 代表的拍動抽出部
1500 グレーレベル同時生起確率行列算出部
1501 時間的特徴算出部
1700 腫瘍ブロック分割部
1701 癌確率算出部
1702 閾値化部
1900 基準心周期決定部
1901 遅延算出部
2000 基準心波形算出部
2001 個別差異算出部
2002 全体差異算出部
2100 拍動リサンプリング部
2101 自己回帰式演算部
Claims (31)
- 超音波で組織をスキャンして得られるスキャン信号により前記組織に含まれる悪性腫瘍を検出する組織悪性腫瘍検出方法であって、
前記組織をスキャンした領域を複数のブロックに分割するブロック分割ステップと、
前記複数のブロックのそれぞれごとに、前記組織が拍動することにより生じる前記組織の変位の時間変化である組織拍動を、前記スキャン信号に基づき推定する組織拍動推定ステップと、
前記複数のブロックのそれぞれごとに、前記組織の拍動に関するパラメータである複数の拍動関連特徴を前記組織拍動から抽出する拍動関連特徴抽出ステップと、
前記複数の拍動関連特徴の分布特性を前記複数のブロックのそれぞれごとに算出する分布特性算出ステップと、
前記分布特性に基づいて、前記複数のブロックのそれぞれが、悪性腫瘍を含むブロックである悪性ブロックか否かを分類する悪性分類ステップとを含む
組織悪性腫瘍検出方法。 - さらに、前記悪性分類ステップにおいて前記悪性ブロックであると分類されたブロックに基づき、癌性腫瘍の位置を特定する腫瘍位置特定ステップを含む
請求項1記載の組織悪性腫瘍検出方法。 - 前記組織拍動推定ステップは、
前記スキャン信号から組織の空間位置における変位である組織変位を算出する組織変位算出ステップと、
算出した前記組織変位から、前記組織変位の空間的な勾配である組織歪みを算出する組織歪み算出ステップと、
前記組織変位対時間または前記組織歪み対時間として前記組織拍動の波形である拍動波形を生成する拍動波形生成ステップとを含む
請求項1記載の組織悪性腫瘍検出方法。 - さらに、推定された前記組織拍動のうち、心臓の拍動に起因する心拍動による成分を抽出する前処理ステップを含む
請求項1記載の組織悪性腫瘍検出方法。 - 前記前処理ステップは、さらに、
推定された前記組織拍動のうち、心拍動に関連する電力である心電力を算出する心電力算出ステップと、
前記心電力の大きさに基づいて前記組織内の領域をクラスタリングする心拍クラスタリングステップと、
クラスタリングされた前記組織内の領域における前記心拍動の振幅の極値を識別する極値識別ステップと、
前記極値に基づいて前記組織拍動の波形である拍動波形の振幅を調整する心拍調整ステップとを含む
請求項4記載の組織悪性腫瘍検出方法。 - 前記極値識別ステップは、さらに、
前記拍動波形におけるピークと谷とを識別する極値点識別ステップと、
他のピークからのずれの大きさが事前に定められた閾値よりも大きい干渉ピーク、及び、他の谷からのずれの大きさが事前に定められた閾値よりも大きい干渉谷を削除するため、前記拍動波形の振幅の異常値を棄却する異常値棄却ステップとを含む
請求項5記載の組織悪性腫瘍検出方法。 - 前記前処理ステップは、さらに、
前記拍動波形から、他のピークからのずれの大きさが事前に定められた閾値よりも大きい干渉ピークに対応する部分と、他の谷からのずれの大きさが事前に定められた閾値よりも大きい干渉谷に対応する部分とを除去する干渉削除ステップと、
前記拍動波形のピークが時間方向に一列に並び、かつ前記拍動波形の谷が時間方向に一列に並ぶように、前記拍動波形の振幅を調整する拍動調整ステップとを含む
請求項5記載の組織悪性腫瘍検出方法。 - 前記組織拍動推定ステップでは、前記複数のブロックに含まれる各ブロックの全スキャンポイントに対して前記組織拍動を推定する
請求項1記載の組織悪性腫瘍検出方法。 - 前記組織拍動推定ステップでは、前記複数のブロックに含まれる各ブロックの、1つまたは数個の代表的拍動、または、前記代表的拍動の組み合わせとして前記組織拍動を推定する
請求項1記載の組織悪性腫瘍検出方法。 - 前記拍動関連特徴は、前記組織拍動の振幅についての特徴量である拍動振幅特徴、前記組織拍動の波形の形状についての特徴量である拍動形状特徴、及び、前記組織拍動の波形の時間変化についての特徴量である拍動時間特徴のうち、少なくとも1つである
請求項1記載の組織悪性腫瘍検出方法。 - 前記拍動形状特徴は、心周期の収縮部分の期間である収縮期間(L1)と心周期の拡張部分の期間である拡張期間(L2)との比率であるL1/L2と、
前記収縮期間において、振幅が最大振幅に対して予め定義された比率と等しくなる収縮曲線上の点である収縮中間点、及び、前記拡張期間において、振幅が前記最大振幅に対して予め定義された比率と等しくなる拡張曲線上の点である拡張中間点の間の期間(L3)と、心拍動の周期(L4)との比率であるL3/L4と、
前記収縮期間における振幅のピークである収縮ピークと前記拡張期間の終了点である拡張終了点とを結ぶ予め定義された曲線からの前記拡張曲線の偏差と、
心拍動の非対称性を表す歪度と、
前記収縮ピークの鋭さを表す尖度と、
前記拡張曲線に含まれる極値の偏差とのうち、少なくとも1つである
請求項10記載の組織悪性腫瘍検出方法。 - 前記拍動形状特徴は、
前記組織拍動から心周期を算出する心拍期間算出ステップと、
前記心周期を用いて臨界点を識別する臨界点識別ステップと、
前記心周期と前記臨界点とに基づいて前記拍動形状特徴を抽出する形状特徴抽出ステップとを含む形状特徴算出ステップにおいて算出される
請求項10記載の組織悪性腫瘍検出方法。 - 前記心周期の前記臨界点は、心周期の収縮部分の開始点である収縮開始点と、心周期の拡張部分の終了点である拡張終了点と、前記収縮期間における振幅のピークである収縮ピークと、心周期の収縮部分の期間において、振幅が最大振幅に対して予め定義された比率と等しくなる収縮曲線上の点である収縮中間点と、心周期の拡張部分の期間において、振幅が最大振幅に対して予め定義された比率と等しくなる拡張曲線上の点である拡張中間点とを含む
請求項12記載の組織悪性腫瘍検出方法。 - 前記臨界点識別ステップは、
前記組織拍動における極小点と極大点とを検索する検索ステップと、
前記極小点と極大点とに基づいて、前記組織拍動が上向きの収縮曲線を有するか、下向きの収縮曲線を有するかを表す拍動方向を識別する拍動方向識別ステップと、
前記極大点と、前記極小点と、前記拍動方向とを用いて、前記拍動波形における前記臨界点を求める臨界点決定ステップとを含む
請求項12記載の組織悪性腫瘍検出方法。 - 前記予め定義された曲線は直線である
請求項11記載の組織悪性腫瘍検出方法。 - 拍動関連特徴抽出ステップでは、前記拍動が上向きの収縮曲線を有する場合は、前記拡張曲線と前記予め定義された曲線との間の正の差分和を前記偏差として算出し、前記拍動が下向きの収縮曲線を有する場合は、前記拡張曲線と前記予め定義された曲線との間の負の差分和を前記偏差として算出する
請求項11記載の組織悪性腫瘍検出方法。 - 前記拍動時間特徴は、前記心周期の遅延と、前記心拍動の波形である心波形の差異と、前記組織拍動の波形の自己回帰係数とのうち、少なくとも1つである
請求項10記載の組織悪性腫瘍検出方法。 - 前記心周期の遅延は、
基準となる心周期である基準心周期を決定する基準心周期決定ステップと、
前記基準心周期に対する対象心周期の遅延を算出する遅延算出ステップとを含む心周期遅延算出ステップにおいて算出される
請求項17記載の組織悪性腫瘍検出方法。 - 前記基準心周期決定ステップでは、前記基準心周期として、スキャンデータのうち振幅が最も大きい心周期を選択する
請求項18記載の組織悪性腫瘍検出方法。 - 前記基準心周期は、心電図検査(ECG)信号から決定される
請求項18記載の組織悪性腫瘍検出方法。 - 前記心波形の差異は、
基準となる心波形である基準心波形を算出する基準心波形算出ステップと、
拍動による複数の心波形のそれぞれと前記基準心波形との差異を算出する個別差異算出ステップと、
算出された複数の前記差異から、前記複数の心波形と前記基準心波形との差異を代表する値である全体心波形差異値を算出する全体差異算出ステップとを含む遅延算出ステップにおいて算出される
請求項19記載の組織悪性腫瘍検出方法。 - 前記全体心波形差異値は、算出された複数の前記差異の標準偏差である
請求項21記載の組織悪性腫瘍検出方法。 - 前記自己回帰係数は、
前記心周期が同じ期間となるように複数の前記拍動波形をテーパリングする拍動リサンプリングステップと、
事前に定められた自己回帰モデルとテーパリングされた前記拍動波形とに基づいて前記自己回帰モデルが有する係数である自己回帰係数を求める自己回帰式演算ステップとを含む自己回帰係数算出ステップにおいて算出される
請求項17記載の組織悪性腫瘍検出方法。 - 前記分布特性は、空間的分布パラメータと、統計的分布パラメータとのうちの少なくとも1つである
請求項1記載の組織悪性腫瘍検出方法。 - 前記空間的分布パラメータは、前記拍動関連特徴のエネルギーと、前記拍動関連特徴のエントロピーと、前記拍動関連特徴のコントラストと、前記拍動関連特徴の均質性と、前記拍動関連特徴の相関性とのうちの少なくとも1つである
請求項24記載の組織悪性腫瘍検出方法。 - 前記統計的分布パラメータは、前記拍動関連特徴の平均値と、前記拍動関連特徴の中央値と、前記拍動関連特徴の最大値と、前記拍動関連特徴の最小値と、前記拍動関連特徴の標準偏差と、前記拍動関連特徴の尖度と、前記拍動関連特徴の歪度とのうちの少なくとも1つである
請求項24記載の組織悪性腫瘍検出方法。 - 前記拍動関連特徴およびその分布特性は、
前記複数のブロックのうちの各ブロックに含まれる複数のスキャンポイントの拍動振幅の中央値、エントロピー、標準偏差、平均値、および、最大値と、
前記複数のブロックのうちの各ブロックに含まれる複数のスキャンポイントの心波形差分の中央値、エントロピー、標準偏差、平均値、および、最大値と、
前記複数のブロックのうちの各ブロックに含まれる複数のスキャンポイントの心周期の収縮部分の期間である収縮期間と心周期の収縮部分の期間である拡張期間との比率の中央値と、
前記複数のブロックのうちの各ブロックにおける、代表的拍動の前記収縮期間と前記拡張期間との比率と、
前記複数のブロックのうちの各ブロックに含まれる複数のスキャンポイントの拡張曲線の偏差の最大値および標準偏差とのうちの少なくとも1つである
請求項1記載の組織悪性腫瘍検出方法。 - 前記腫瘍位置特定ステップは、さらに、
スキャンした前記組織のスキャンポイントごとに対象領域を規定する対象領域特定ステップと、
前記複数のブロックのうち、前記対象領域に属するブロックを特定する腫瘍ブロック分割ステップと、
前記対象領域に属するブロックの前記悪性分類ステップによる分類結果に基づいて、前記組織が癌である確率を算出する癌確率算出ステップとを含む
請求項2記載の組織悪性腫瘍検出方法。 - 前記腫瘍位置特定ステップは、さらに、
スキャンされた前記組織のスキャンポイントにおける前記癌である確率を、2次元または3次元画像で表示する画像化ステップを含む
請求項28記載の組織悪性腫瘍検出方法。 - 超音波で組織をスキャンして得られるスキャン信号により前記組織に含まれる悪性腫瘍を検出する組織悪性腫瘍検出装置であって、
前記組織をスキャンした領域を複数のブロックに分割するブロック分割部と、
前記複数のブロックのそれぞれごとに、前記組織が拍動することにより生じる前記組織の変位の時間変化である組織拍動を、前記スキャン信号に基づき推定する組織拍動推定部と、
前記複数のブロックのそれぞれごとに、前記組織の拍動に関するパラメータである複数の拍動関連特徴を前記組織拍動から抽出する拍動関連特徴抽出部と、
前記複数の拍動関連特徴の分布特性を前記複数のブロックのそれぞれごとに算出する分布特性算出部と、
前記分布特性に基づいて、前記複数のブロックのそれぞれが、悪性腫瘍を含むブロックである悪性ブロックか否かを分類する悪性分類部とを備える
組織悪性腫瘍検出装置。 - さらに、前記悪性分類部において前記悪性ブロックであると分類されたブロックに基づき、癌性腫瘍の位置を特定する腫瘍位置特定部を備える
請求項30記載の組織悪性腫瘍検出装置。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP11792127.0A EP2578163A1 (en) | 2010-06-07 | 2011-06-03 | Malignant tissue tumor detection method and malignant tissue tumor detection device |
CN201180003129.2A CN102469991B (zh) | 2010-06-07 | 2011-06-03 | 组织恶性肿瘤检测装置 |
US13/388,561 US20120136255A1 (en) | 2010-06-07 | 2011-06-03 | Tissue malignant tumor detection method and tissue malignant tumor detection apparatus |
JP2011544741A JPWO2011155168A1 (ja) | 2010-06-07 | 2011-06-03 | 組織悪性腫瘍検出方法、組織悪性腫瘍検出装置 |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2010130439 | 2010-06-07 | ||
JP2010-130439 | 2010-06-07 | ||
JP2011-085718 | 2011-04-07 | ||
JP2011085718 | 2011-04-07 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011155168A1 true WO2011155168A1 (ja) | 2011-12-15 |
Family
ID=45097785
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2011/003156 WO2011155168A1 (ja) | 2010-06-07 | 2011-06-03 | 組織悪性腫瘍検出方法、組織悪性腫瘍検出装置 |
Country Status (5)
Country | Link |
---|---|
US (1) | US20120136255A1 (ja) |
EP (1) | EP2578163A1 (ja) |
JP (1) | JPWO2011155168A1 (ja) |
CN (1) | CN102469991B (ja) |
WO (1) | WO2011155168A1 (ja) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015190180A1 (ja) * | 2014-06-11 | 2015-12-17 | オリンパス株式会社 | 医用診断装置、医用診断装置の作動方法および医用診断装置の作動プログラム |
TWI547266B (zh) * | 2014-06-11 | 2016-09-01 | 許百靈 | 核醫單光子影像測量腫瘤標準攝取值的方法及系統 |
KR20180031702A (ko) * | 2015-07-09 | 2018-03-28 | 아토믹 온콜로지 피티와이 리미티드 | 원자 치료 지수 |
JP2018516135A (ja) * | 2015-06-04 | 2018-06-21 | コーニンクレッカ フィリップス エヌ ヴェKoninklijke Philips N.V. | がん悪性度マップにより拡張された精密診断及び治療に対するシステム及び方法 |
WO2022054288A1 (ja) * | 2020-09-14 | 2022-03-17 | オリンパス株式会社 | 超音波観測装置、超音波観測装置の作動方法および超音波観測装置の作動プログラム |
Families Citing this family (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10814577B2 (en) | 2006-08-11 | 2020-10-27 | Gregorio Lim Tan | Self opening wide mouth carryout bag pack, apparatus and method of making same |
FR2965951B1 (fr) * | 2010-10-11 | 2013-10-04 | Olea Medical | Systeme et procede pour estimer une quantite d'interet d'un systeme dynamique artere/tissu/veine |
US10226227B2 (en) * | 2013-05-24 | 2019-03-12 | Sunnybrook Research Institute | System and method for classifying and characterizing tissues using first-order and second-order statistics of quantitative ultrasound parametric maps |
US9092691B1 (en) | 2014-07-18 | 2015-07-28 | Median Technologies | System for computing quantitative biomarkers of texture features in tomographic images |
EP2989986B1 (en) * | 2014-09-01 | 2019-12-18 | Samsung Medison Co., Ltd. | Ultrasound diagnosis apparatus and method of operating the same |
JP6497911B2 (ja) * | 2014-11-28 | 2019-04-10 | キヤノン株式会社 | 信号処理方法、信号処理装置、音響波処理方法および音響波処理装置 |
EP3282932A4 (en) * | 2015-04-13 | 2018-12-05 | Harbor Vascular, Inc. | Methods and system for assessment of peripheral perfusion |
CN107920798B (zh) * | 2015-09-01 | 2019-05-28 | 皇家飞利浦有限公司 | 用于显示身体部位的医学图像数据的装置 |
US11051790B2 (en) | 2015-11-10 | 2021-07-06 | Exact Imaging, Inc. | System comprising indicator features in high-resolution micro-ultrasound images |
EP3413790A4 (en) * | 2016-02-10 | 2019-10-16 | Balter, Inc. | SYSTEMS AND METHOD FOR EVALUATING LESIONS OF PIGMENTED TISSUE |
US10282588B2 (en) * | 2016-06-09 | 2019-05-07 | Siemens Healthcare Gmbh | Image-based tumor phenotyping with machine learning from synthetic data |
US11298072B2 (en) | 2016-07-01 | 2022-04-12 | Bostel Technologies, Llc | Dermoscopy diagnosis of cancerous lesions utilizing dual deep learning algorithms via visual and audio (sonification) outputs |
US11484247B2 (en) | 2016-07-01 | 2022-11-01 | Bostel Technologies, Llc | Phonodermoscopy, a medical device system and method for skin diagnosis |
CN109640830B (zh) * | 2016-07-14 | 2021-10-19 | 医视特有限公司 | 基于先例的超声聚焦 |
EP3570753A4 (en) * | 2017-02-23 | 2020-10-21 | Google LLC | METHOD AND SYSTEM TO AID IN PATHOLOGIST IDENTIFICATION OF TUMOR CELLS IN ENLARGED TISSUE IMAGES |
CA3088965A1 (en) | 2018-01-18 | 2019-07-25 | Neural Analytics, Inc. | Waveform visualization tool for facilitating medical diagnosis |
US11129587B2 (en) | 2018-01-22 | 2021-09-28 | Novasignal Corp. | Systems and methods for detecting neurological conditions |
TWI682169B (zh) * | 2018-03-29 | 2020-01-11 | 佳世達科技股份有限公司 | 超音波成像方法 |
CN108663657B (zh) * | 2018-03-30 | 2019-04-23 | 清华大学 | 基于柔性超声换能器单元的移动振源追踪方法及装置 |
EP3730058A1 (en) * | 2019-04-24 | 2020-10-28 | Koninklijke Philips N.V. | Fetal ultrasound processing unit for separating heart rate signals |
CN111275093B (zh) * | 2020-01-17 | 2024-01-26 | 上海乐普云智科技股份有限公司 | 一种用于多标签标注心电信号的心搏分类方法和装置 |
CN112598648B (zh) * | 2020-12-24 | 2022-08-26 | 重庆邮电大学 | 一种基于图像梯度方向的图像接缝裁剪篡改检测方法 |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1980002365A1 (en) | 1979-05-04 | 1980-11-13 | Rasor Ass Inc | Ultrasonic image enhancement |
US5233994A (en) | 1991-05-13 | 1993-08-10 | Advanced Technology Laboratories, Inc. | Detection of tissue abnormality through blood perfusion differentiation |
WO2001028594A2 (en) | 1999-10-22 | 2001-04-26 | Fbit Ltd. | Tumor detection by imaging and therapy of tumors |
JP2003116855A (ja) * | 2001-10-19 | 2003-04-22 | Aloka Co Ltd | 超音波診断装置 |
US20030149358A1 (en) | 2002-02-06 | 2003-08-07 | Nissan Maskil | Ultrasonic system for non-invasive early prostate cancer detection |
US20040199077A1 (en) | 2003-01-03 | 2004-10-07 | Xiaohui Hao | Detection of tumor halos in ultrasound images |
US20050027188A1 (en) | 2002-12-13 | 2005-02-03 | Metaxas Dimitris N. | Method and apparatus for automatically detecting breast lesions and tumors in images |
US20060241463A1 (en) | 2004-12-08 | 2006-10-26 | Yio-Wha Shau | Quantitative non-invasive method for detecting degree of malignancy in tumors and application thereof |
JP2009183564A (ja) * | 2008-02-07 | 2009-08-20 | Hitachi Medical Corp | 超音波診断装置 |
JP2010051554A (ja) * | 2008-08-28 | 2010-03-11 | Konica Minolta Medical & Graphic Inc | 超音波診断装置 |
JP2010512900A (ja) * | 2006-12-21 | 2010-04-30 | アンスティテュ ギュスタヴ ルーシ −アイジーアール− | 腫瘍の血管新生の定量化方法およびシステム |
Family Cites Families (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5183046A (en) * | 1988-10-17 | 1993-02-02 | Board Of Regents Of The University Of Washington | Ultrasonic plethysmograph |
US5687291A (en) * | 1996-06-27 | 1997-11-11 | The United States Of America As Represented By The Secretary Of The Army | Method and apparatus for estimating a cognitive decision made in response to a known stimulus from the corresponding single-event evoked cerebral potential |
US7914442B1 (en) * | 1999-03-01 | 2011-03-29 | Gazdzinski Robert F | Endoscopic smart probe and method |
US6510337B1 (en) * | 1999-11-26 | 2003-01-21 | Koninklijke Philips Electronics, N.V. | Multi-phase cardiac imager |
US8790272B2 (en) * | 2002-03-26 | 2014-07-29 | Adidas Ag | Method and system for extracting cardiac parameters from plethysmographic signals |
US6733461B2 (en) * | 2002-08-01 | 2004-05-11 | Hypertension Diagnostics, Inc. | Methods and apparatus for measuring arterial compliance, improving pressure calibration, and computing flow from pressure data |
JP3932482B2 (ja) * | 2002-10-18 | 2007-06-20 | 株式会社日立メディコ | 超音波診断装置 |
JP4733938B2 (ja) * | 2004-07-16 | 2011-07-27 | 株式会社東芝 | 超音波診断装置および超音波画像処理装置 |
US9066679B2 (en) * | 2004-08-31 | 2015-06-30 | University Of Washington | Ultrasonic technique for assessing wall vibrations in stenosed blood vessels |
JP5529378B2 (ja) * | 2004-08-31 | 2014-06-25 | ユニヴァーシティ オブ ワシントン | 狭窄血管における壁振動を評価するための超音波技法 |
JP4827745B2 (ja) * | 2005-01-19 | 2011-11-30 | 株式会社日立メディコ | 画像診断支援装置及び画像診断支援プログラム |
US7392075B2 (en) * | 2005-03-03 | 2008-06-24 | Nellcor Puritan Bennett Incorporated | Method for enhancing pulse oximetry calculations in the presence of correlated artifacts |
JP5160227B2 (ja) * | 2005-05-09 | 2013-03-13 | 株式会社日立メディコ | 超音波診断装置及び超音波画像表示方法 |
CN101431942B (zh) * | 2006-03-20 | 2013-08-28 | 松下电器产业株式会社 | 超声波诊断装置 |
US9005126B2 (en) * | 2007-05-03 | 2015-04-14 | University Of Washington | Ultrasonic tissue displacement/strain imaging of brain function |
US8366619B2 (en) * | 2009-05-13 | 2013-02-05 | University Of Washington | Nodule screening using ultrasound elastography |
WO2011094487A2 (en) * | 2010-01-29 | 2011-08-04 | Edwards Lifesciences Corporation | Elimination of the effects of irregular cardiac cycles in the determination of cardiovascular parameters |
JP6010050B2 (ja) * | 2011-02-03 | 2016-10-19 | パルティ、ヨーラム | 経胸腔的な心肺モニター |
-
2011
- 2011-06-03 CN CN201180003129.2A patent/CN102469991B/zh not_active Expired - Fee Related
- 2011-06-03 WO PCT/JP2011/003156 patent/WO2011155168A1/ja active Application Filing
- 2011-06-03 US US13/388,561 patent/US20120136255A1/en not_active Abandoned
- 2011-06-03 JP JP2011544741A patent/JPWO2011155168A1/ja not_active Withdrawn
- 2011-06-03 EP EP11792127.0A patent/EP2578163A1/en not_active Withdrawn
Patent Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4276885A (en) | 1979-05-04 | 1981-07-07 | Rasor Associates, Inc | Ultrasonic image enhancement |
WO1980002365A1 (en) | 1979-05-04 | 1980-11-13 | Rasor Ass Inc | Ultrasonic image enhancement |
US5233994A (en) | 1991-05-13 | 1993-08-10 | Advanced Technology Laboratories, Inc. | Detection of tissue abnormality through blood perfusion differentiation |
WO2001028594A2 (en) | 1999-10-22 | 2001-04-26 | Fbit Ltd. | Tumor detection by imaging and therapy of tumors |
JP2003116855A (ja) * | 2001-10-19 | 2003-04-22 | Aloka Co Ltd | 超音波診断装置 |
US20030149358A1 (en) | 2002-02-06 | 2003-08-07 | Nissan Maskil | Ultrasonic system for non-invasive early prostate cancer detection |
US7466848B2 (en) | 2002-12-13 | 2008-12-16 | Rutgers, The State University Of New Jersey | Method and apparatus for automatically detecting breast lesions and tumors in images |
US20050027188A1 (en) | 2002-12-13 | 2005-02-03 | Metaxas Dimitris N. | Method and apparatus for automatically detecting breast lesions and tumors in images |
US20040199077A1 (en) | 2003-01-03 | 2004-10-07 | Xiaohui Hao | Detection of tumor halos in ultrasound images |
US6984211B2 (en) | 2003-01-03 | 2006-01-10 | Mayo Foundation For Medical Education And Research | Detection of tumor halos in ultrasound images |
US20060241463A1 (en) | 2004-12-08 | 2006-10-26 | Yio-Wha Shau | Quantitative non-invasive method for detecting degree of malignancy in tumors and application thereof |
JP2010512900A (ja) * | 2006-12-21 | 2010-04-30 | アンスティテュ ギュスタヴ ルーシ −アイジーアール− | 腫瘍の血管新生の定量化方法およびシステム |
JP2009183564A (ja) * | 2008-02-07 | 2009-08-20 | Hitachi Medical Corp | 超音波診断装置 |
JP2010051554A (ja) * | 2008-08-28 | 2010-03-11 | Konica Minolta Medical & Graphic Inc | 超音波診断装置 |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015190180A1 (ja) * | 2014-06-11 | 2015-12-17 | オリンパス株式会社 | 医用診断装置、医用診断装置の作動方法および医用診断装置の作動プログラム |
JP5897227B1 (ja) * | 2014-06-11 | 2016-03-30 | オリンパス株式会社 | 医用診断装置、医用診断装置の作動方法および医用診断装置の作動プログラム |
TWI547266B (zh) * | 2014-06-11 | 2016-09-01 | 許百靈 | 核醫單光子影像測量腫瘤標準攝取值的方法及系統 |
US9655593B2 (en) | 2014-06-11 | 2017-05-23 | Olympus Corporation | Medical diagnostic apparatus, method for operating medical diagnostic apparatus, and computer-readable recording medium |
JP2018516135A (ja) * | 2015-06-04 | 2018-06-21 | コーニンクレッカ フィリップス エヌ ヴェKoninklijke Philips N.V. | がん悪性度マップにより拡張された精密診断及び治療に対するシステム及び方法 |
KR20180031702A (ko) * | 2015-07-09 | 2018-03-28 | 아토믹 온콜로지 피티와이 리미티드 | 원자 치료 지수 |
KR102046265B1 (ko) * | 2015-07-09 | 2019-11-18 | 아토믹 온콜로지 피티와이 리미티드 | 원자 치료 지수 |
WO2022054288A1 (ja) * | 2020-09-14 | 2022-03-17 | オリンパス株式会社 | 超音波観測装置、超音波観測装置の作動方法および超音波観測装置の作動プログラム |
Also Published As
Publication number | Publication date |
---|---|
US20120136255A1 (en) | 2012-05-31 |
JPWO2011155168A1 (ja) | 2013-08-01 |
CN102469991B (zh) | 2016-05-18 |
EP2578163A1 (en) | 2013-04-10 |
CN102469991A (zh) | 2012-05-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2011155168A1 (ja) | 組織悪性腫瘍検出方法、組織悪性腫瘍検出装置 | |
Golemati et al. | Using the Hough transform to segment ultrasound images of longitudinal and transverse sections of the carotid artery | |
Meiburger et al. | Automated localization and segmentation techniques for B-mode ultrasound images: A review | |
US11049246B2 (en) | Rapid calculation method and system for plaque stability index based on medical image sequence | |
Molinari et al. | Intima-media thickness: setting a standard for a completely automated method of ultrasound measurement | |
US20080051660A1 (en) | Methods and apparatuses for medical imaging | |
US9924927B2 (en) | Method and apparatus for video interpretation of carotid intima-media thickness | |
US20110257545A1 (en) | Imaging based symptomatic classification and cardiovascular stroke risk score estimation | |
KR101580254B1 (ko) | 주파수 영역에서의 의료 영상 진단 장치 및 방법 | |
US20110257505A1 (en) | Atheromatic?: imaging based symptomatic classification and cardiovascular stroke index estimation | |
CN103260526B (zh) | 具有峰值强度检测功能的超声成像系统和方法 | |
Loizou et al. | An integrated system for the segmentation of atherosclerotic carotid plaque ultrasound video | |
US20130253319A1 (en) | Method and system for acquiring and analyzing multiple image data loops | |
Rossi et al. | Automatic recognition of the common carotid artery in longitudinal ultrasound B-mode scans | |
WO2024067527A1 (zh) | 一种髋关节角度检测系统和方法 | |
Santhiyakumari et al. | Detection of the intima and media layer thickness of ultrasound common carotid artery image using efficient active contour segmentation technique | |
CN111275755A (zh) | 基于人工智能的二尖瓣瓣口面积检测方法、系统和设备 | |
Butakoff et al. | Left-ventricular epi-and endocardium extraction from 3D ultrasound images using an automatically constructed 3D ASM | |
Scaramuzzino et al. | Longitudinal motion assessment of the carotid artery using speckle tracking and scale-invariant feature transform | |
Zheng et al. | An off-line gating method for suppressing motion artifacts in ICUSsequence | |
Raj et al. | An insight into elasticity analysis of common carotid artery using ultrasonography | |
Panicker et al. | Employing acoustic features to aid neural networks towards platform agnostic learning in lung ultrasound imaging | |
Kachenoura et al. | An automated four-point scale scoring of segmental wall motion in echocardiography using quantified parametric images | |
Berthomier et al. | Scattering operator and spectral clustering for ultrasound images: Application on deep venous thrombi | |
Isguder et al. | Manifold learning for image-based gating of intravascular ultrasound (IVUS) pullback sequences |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 201180003129.2 Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2011544741 Country of ref document: JP |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 11792127 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13388561 Country of ref document: US Ref document number: 2011792127 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |