DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant’s arguments with respect to claim(s) 1-6, 8, 10-18, & 20-21 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-6, 8, 10-18, & 20-21 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1 recites the limitation “calculating a correlation value between each navigator signal of the plurality of navigator signals” and “first neural network by inputting thereto the calculated correlation value” which renders the claim unclear. It is unclear whether a single correlation value is inputted or all calculated correlation values are inputted.
Claim 1 also recites the limitation “a number of weighting elements each representing a certainty level of the collected data being corrupted by the motion of the object” which renders the claim unclear. It is unclear whether each weighting elements represents a single K-space point or the entire data set. For the purposes of this examination it is interpreted as being for a single k-space point.
Claim 15 recites the limitation " the deep learning framework " in Line 9. There is insufficient antecedent basis for this limitation in the claim.
Claim 20 recites the limitation “calculating a correlation value between each navigator signal of the plurality of navigator signals” and “first neural network by inputting thereto the calculated correlation value” which renders the claim unclear. It is unclear whether a single correlation value is inputted or all calculated correlation values are inputted.
Claim 20 also recites the limitation “a number of weighting elements each representing a certainty level of the collected data being corrupted by the motion of the object” which renders the claim unclear. It is unclear whether each weighting elements represents a single K-space point or the entire data set. For the purposes of this examination it is interpreted as being for a single k-space point.
The following is a quotation of 35 U.S.C. 112(d):
(d) REFERENCE IN DEPENDENT FORMS.—Subject to subsection (e), a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
The following is a quotation of pre-AIA 35 U.S.C. 112, fourth paragraph:
Subject to the following paragraph [i.e., the fifth paragraph of pre-AIA 35 U.S.C. 112], a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claim 3 is rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends. Claim 3 recites “assigning, based on a comparison of the correlation value calculated for each navigator signal, of the plurality of navigator signals, with an empirical correlation threshold, a particular value to the number of weighting elements of the weighting matrix” which in view of Claim 1 and the Applicant’s Specification fails to further limit the invention. As stated in Claim 1 the data consistency weighting matrix is obtained by inputting the correlation values into a first neural network, in Applicants Spec [0053] thresholding is identified as a distinct alternative embodiment to the neural network. It is thus viewed that claim 3 is a substitution of the methods described in Claim 1 rather than a further limitation. Applicant may cancel the claim(s), amend the claim(s) to place the claim(s) in proper dependent form, rewrite the claim(s) in independent form, or present a sufficient showing that the dependent claim(s) complies with the statutory requirements.
Claim 6 is rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends. Claim 6 recites “and the estimating step is omitted.” which in view of Claim 1 fails to further limit the invention. As stated in Claim 1, the estimating step is essential. Applicant may cancel the claim(s), amend the claim(s) to place the claim(s) in proper dependent form, rewrite the claim(s) in independent form, or present a sufficient showing that the dependent claim(s) complies with the statutory requirements.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1-2, 6, 8, 16-18, & 20 are rejected under 35 U.S.C. 103 as being unpatentable over Splithoff (US20220342018A1; hereinafter referred to as Splithoff) in view of Beck (US20150346307A1) and further in view of Cordero et al (L. Cordero-Grande et al., “Sensitivity encoding for aligned Multishot Magnetic Resonance Reconstruction,” IEEE Transactions on Computational Imaging, vol. 2, no. 3, pp. 266–280, Sep. 2016; hereinafter referred to as Cordero).
Regarding Claim 1, Splithoff discloses a method for motion correction in a magnetic resonance imaging system (“Systems and Methods that identify the effect of motion during a medical imaging procedure.” [Abstract]), the method comprising:
receiving data collected from imaging an object by the magnetic resonance imaging system, or image data reconstructed from the collected data (“In response to applied RF pulse signals, the RF coil 18 receives MR signals, e.g., signals from the excited protons within the body as the protons return to an equilibrium position established by the static and gradient magnetic fields. The MR signals are detected and processed by a detector within RF module and the control unit 20 to provide an MR dataset to a processor 22 for processing into an image. In some embodiments, the processor 22 is located in the control unit 20, in other embodiments, the processor 22 is located remotely. A two or three-dimensional k-space storage array of individual data elements in a memory 24 of the control unit 20 stores corresponding individual frequency components including an MR dataset.” [0021]);
estimating a motion parameter indicating a motion of the object (“the motion assessment includes one or more values for respective degree of freedoms describing the 3D motion state of the patient at the time of the acquisition of a chunk of the data relative to an initial position. The initial position may be defined by a rapid motion-free initial scan and the patient's movements are quantified into different positions/poses after the sampling of the current chunk of the acquisition (e.g., echo train/shot/line of the k-space in multi-shot acquisitions).” [0042]);
obtaining a data-consistency weighting matrix output from a first neural network by inputting thereto the calculated correlation value, the data-consistency weighting matrix including a number of weighting elements each representing a certainty level of the collected data being corrupted by the motion of the object (“In an embodiment, at Act A150, the control unit 20 determines or calculates a coil mixing matrix including either the CMM as described above or a coil mixing error matrix (also referred to as the difference coil mixing error matrix) for the respective chunk of MR data based on an error between the linearly combined motion data and the linearly combined reference data by determining the linear combination of the error. In certain embodiments, the control unit 20 only uses the CMM and thus the coil mixing error matrix may not be calculated (or may be calculated and not used as an input to the neural network). The coil mixing error matrix may be the difference between a subset of data from the motion free reference and from the definitive imaging procedure being performed and thus may be referred to as the difference coil mixing error matrix.” [0035], “At Act A160, the control unit 20 inputs the coil mixing error matrix and/or the second coil mixing matrix (CMM) into a neural network trained to output a motion assessment for the acquired MRI data. Additional information may be input into the neural network when available. The additional information may include a data consistency error of the current ET, the object size relative to the image matrix size and the relative energy of the current ET to the whole. The additional information may help account for the differences in the depicted object.” [0036], “At Act A170, the control unit 20 provides the motion assessment generated by the neural network to an operator… the motion assessment is or includes a motion score that quantifies the extent of the motion. The motion score may be compared to a threshold score… the control unit 20 ranks the motion scores for each echo train and replaces or reacquires the data for each of the n-th highest ranked echo trains or above a certain level. For example, the control unit 20 may replace the data for the worst 5, 10, or 20% of the acquired data among other levels.” [0037]).
Splithoff does not specifically disclose that the motion parameter is based on a plurality of navigator signals acquired by non-imaging while the collected data is being collected; determining, based on the plurality of navigator signals, a reference navigator signal; calculating a correlation value between each navigator signal of the plurality of navigator signals and the reference navigator signal; and generating, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data.
However, in a similar field of endeavor, Beck teaches a method and magnetic resonance (MR) apparatus for performing an MR examination with prospective motion correction [Abstract].
Beck also teaches the motion parameter is based on a plurality of navigator signals acquired by non-imaging while the collected data is being collected; determining, based on the plurality of navigator signals, a reference navigator signal (“This object is achieved in a method and magnetic resonance (MR) apparatus for performing an MR examination with prospective motion correction, wherein multiple MR signals are acquired. For each MR signal, signal, an examination volume is established and a navigator volume is established for recording navigator signals. The examination volume and the navigator volume are not identical. At least one navigator reference signal is acquired at a time t0, and at least one navigator signal is acquired at a time t1>t0. Motion information is determined from the navigator signal and the navigator reference signal, and the recording parameters are set as a function of the motion information. At least one further magnetic resonance signal is acquired with this setting.” [0016], “The measured signals and the navigator signals are both magnetic resonance signals; one or more images or a spectrum are obtained from the measured signals, and motion information is obtained from the navigator signals.” [0049]);
calculating a correlation value between each navigator signal of the plurality of navigator signals and the reference navigator signal (“the motion information can be determined using a cross-correlation analysis. In this, several cross-correlation coefficients are calculated and the cross-correlation coefficient with the highest numerical value indicates the position of the best match between the reference data and the data obtained from the current navigator echo.” [0025])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff as outlined above with the motion parameter is based on a plurality of navigator signals acquired by non-imaging while the collected data is being collected; determining, based on the plurality of navigator signals, a reference navigator signal; calculating a correlation value between each navigator signal of the plurality of navigator signals and the reference navigator signal as taught by Beck, because it is less prone to error for performing a magnetic resonance examination with a prospective motion correction [0015].
Splithoff in view of Beck does not specifically teach generating, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data.
However, in a similar field of endeavor, Cordero teaches a framework for the recon struction of magnetic resonance images in the presence of rigid motion [Abstract].
Cordero also teaches generating, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data (“The generalized reconstruction with rigid motion correction for parallel multi shot imaging can be formulated in matrix form as: (ˆx, ˆ T)=argminx,T∥AFSTx−y∥2 2, (1) where y denotes the measured k-space data, x the image to be reconstructed, T the rigid motion transformation matrix, S the coil sensitivity matrix, F the Discrete Fourier transform (DFT) encompassing applied k-space oversampling or down sampling, and A a sampling matrix. The forward model for this formulation, originally proposed in[13] for the estimation of x assuming T is known, is depicted in Fig.1 and the different terms are described in Section II-B” [A. Generalized Rigid Motion-Corrected Multishot Reconstruction], “5) T is a matrix of size NS×N given by T= ⎡ ⎢ ⎣ T1 . . . TS ⎤ ⎥ ⎦, where Ts is a matrix of size N×N corresponding to the rigid transformation the underlying structure has been subject to when acquiring the shots. This matrix is described just below.” [B. Model Terms])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff in view of Beck as outlined above with generating, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data as taught by Cordero, because it generally improve the quality of reconstructed images [Abstract].
Regarding Claim 2, Splithoff discloses that the collected data is k-space data acquired by a plurality of shots, each shot acquiring a plurality of k-space lines, each k-space line comprising a plurality of k-space points, each navigator signal of the plurality of navigator signals is acquired during one of the plurality of shots (“the motion assessment includes one or more values for respective degree of freedoms describing the 3D motion state of the patient at the time of the acquisition of a chunk of the data relative to an initial position. The initial position may be defined by a rapid motion-free initial scan and the patient's movements are quantified into different positions/poses after the sampling of the current chunk of the acquisition (e.g., echo train/shot/line of the k-space in multi-shot acquisitions).” [0042]),
Splithoff in view of Beck does not specifically teach each weighting element of the data-consistency weighting matrix corresponds to a k- space point and represents a certainty level of the corresponding k-space point being corrupted by the motion of the object.
However, in a similar field of endeavor, Cordero teaches each weighting element of the data-consistency weighting matrix corresponds to a k- space point and represents a certainty level of the corresponding k-space point being corrupted by the motion of the object (“We want to reconstruct a 3D image of size N=N1N2N3, with Nl the number of voxels along dimension l using a coil array of C elements from M=ESC samples of a discretized k-space grid of size K=K1K2K3, where E denotes the number of sample d points per shot and S is the number of shots. The different terms included in (1) can be represented by the following matrices: A is a matrix of size M×KSC given by A where Asc is a matrix of size E×K that takes the value 1 if the sample of the shot s corresponds to the k-space location indexed by k and 0 otherwise” [B. Model Terms])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Splithoff in view of Beck as outlined above with each weighting element of the data-consistency weighting matrix corresponds to a k- space point and represents a certainty level of the corresponding k-space point being corrupted by the motion of the object as taught by Cordero, because it generally improve the quality of reconstructed images [Abstract].
Regarding Claim 8, Splithoff discloses further comprising: obtaining motion-free navigator data; determining different motions to be simulated; generating motion-affected navigator data by simulating a corresponding influence of each motion on the motion-free navigator data; and using the motion-free navigator data and the motion-affected navigator data to train the neural network, so as to learn a mapping from the motion-affected navigator data to a corresponding data consistency weighting matrix (“the method comprising: acquiring a motion free reference; calculating, based on the motion free reference, a scout coil mixing matrix representing a linear combination of coils of the imaging system;” [0005], “The training data for the model includes ground truth data or gold standard data acquired or simulated prior to training the neural network. Ground truth data and gold standard data is data that includes correct or reasonably accurate labels that are verified manually or by some other accurate method. The training data may be acquired at any point prior to inputting the training data into the neural network. The neural network may input the training data (e.g., CMM data, CMEM data and other information) and output a prediction or classification, for example of motion. In an example the prediction may include a motion score that quantifies the amount of motion in the input data. In another example, the prediction or classification may include motion parameters that quantify and describe the detected motion. The prediction is compared to the annotations (e.g., motion scores or values for respective degree of freedoms describing the 3D motion state of the patient at the time of the acquisition of a chunk of the data relative to an initial position) from the training data. A loss function may be used to identify the errors from the comparison. The loss function serves as a measurement of how far the current set of predictions are from the corresponding true values. Some examples of loss functions that may be used include Mean-Squared-Error, Root-Mean-Squared-Error, L2 norm, and Cross-entropy loss. Mean Squared Error loss, or MSE for short, is calculated as the average of the squared differences between the predicted and actual values. Root-Mean Squared Error is similarly calculated as the average of the root squared differences between the predicted and actual values. The L2 norm is similar to the Root-Mean Squared error and is calculated as the square root of the sum of the squared vector values. The max norm that is calculated as the maximum vector values. For cross-entropy loss each predicted probability is compared to the actual class output value (0 or 1) and a score is calculated that penalizes the probability based on the distance from the expected value. The penalty may be logarithmic, offering a small score for small differences (0.1 or 0.2) and enormous score for a large difference (0.9 or 1.0). During training and over repeated iterations, the neural network attempts to minimize the loss function as the result of a lower error between the actual and the predicted values means the neural network has done a good job in learning. Different optimization algorithms may be used to minimize the loss function, such as, for example, gradient descent, Stochastic gradient descent, Batch gradient descent, Mini-Batch gradient descent, among others. The process of inputting, outputting, comparing, and adjusting is repeated for a predetermined number of iterations with the goal of minimizing the loss function. Once adjusted and trained, neural network is ready to be applied to unseen data. In an embodiment, the neural network may be replaced by a model that uses, for example, a pattern matching technique. The model may be configured or trained using iterative machine learning techniques.” [0039]).
Regarding Claim 16, Splithoff discloses all limitations noted above except that the step of receiving the plurality of navigator signals further comprises acquiring the plurality of navigator signals from a non- imaging k-space echo inserted into a pulse sequence of the magnetic resonance imaging system, a respiratory bellow, an electrocardiogram signal, a camera with an external marker, a camera without an external marker, or a pilot-tone-based motion detection signal.
However, in a similar field of endeavor, Beck teaches the step of receiving the plurality of navigator signals further comprises acquiring the plurality of navigator signals from a non- imaging k-space echo inserted into a pulse sequence of the magnetic resonance imaging system, a respiratory bellow, an electrocardiogram signal, a camera with an external marker, a camera without an external marker, or a pilot-tone-based motion detection signal (“This object is achieved in a method and magnetic resonance (MR) apparatus for performing an MR examination with prospective motion correction, wherein multiple MR signals are acquired. For each MR signal, signal, an examination volume is established and a navigator volume is established for recording navigator signals. The examination volume and the navigator volume are not identical. At least one navigator reference signal is acquired at a time t0, and at least one navigator signal is acquired at a time t1>t0. Motion information is determined from the navigator signal and the navigator reference signal, and the recording parameters are set as a function of the motion information. At least one further magnetic resonance signal is acquired with this setting.” [0016], “The measured signals and the navigator signals are both magnetic resonance signals; one or more images or a spectrum are obtained from the measured signals, and motion information is obtained from the navigator signals.” [0049])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff as outlined above with the step of receiving the plurality of navigator signals further comprises acquiring the plurality of navigator signals from a non- imaging k-space echo inserted into a pulse sequence of the magnetic resonance imaging system, a respiratory bellow, an electrocardiogram signal, a camera with an external marker, a camera without an external marker, or a pilot-tone-based motion detection signal as taught by Beck, because it is less prone to error for performing a magnetic resonance examination with a prospective motion correction [0015].
Regarding Claim 17, Splithoff discloses all limitations noted above except that the step of receiving the plurality of navigator signals further comprises acquiring the plurality of navigator signals in a form of a 3D volume, a 2D image, or a 1D signal.
However, in a similar field of endeavor, Beck teaches the step of receiving the plurality of navigator signals further comprises acquiring the plurality of navigator signals in a form of a 3D volume, a 2D image, or a 1D signal (“This object is achieved in a method and magnetic resonance (MR) apparatus for performing an MR examination with prospective motion correction, wherein multiple MR signals are acquired. For each MR signal, signal, an examination volume is established and a navigator volume is established for recording navigator signals. The examination volume and the navigator volume are not identical. At least one navigator reference signal is acquired at a time t0, and at least one navigator signal is acquired at a time t1>t0. Motion information is determined from the navigator signal and the navigator reference signal, and the recording parameters are set as a function of the motion information. At least one further magnetic resonance signal is acquired with this setting.” [0016], “The measured signals and the navigator signals are both magnetic resonance signals; one or more images or a spectrum are obtained from the measured signals, and motion information is obtained from the navigator signals.” [0049])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff as outlined above with the step of receiving the plurality of navigator signals further comprises acquiring the plurality of navigator signals in a form of a 3D volume, a 2D image, or a 1D signal as taught by Beck, because it is less prone to error for performing a magnetic resonance examination with a prospective motion correction [0015].
Regarding Claim 18, Splithoff discloses all limitations noted above except that the estimating step further comprises estimating, as the motion parameter, at least one of a distance of a translation and an angle degree of a rotation.
However, in a similar field of endeavor, Beck teaches the estimating step further comprises estimating, as the motion parameter, at least one of a distance of a translation and an angle degree of a rotation (“3D image data record can be used as a navigator signal and navigator reference signal in each case to determine motion information. With the coverage of the 3D volume every motion can be captured. Thus six degrees of freedom of motion can be covered, namely three translatory and three rotational degrees of freedom.” [0029])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff as outlined above with the estimating step further comprises estimating, as the motion parameter, at least one of a distance of a translation and an angle degree of a rotation as taught by Beck, because it is less prone to error for performing a magnetic resonance examination with a prospective motion correction [0015].
Regarding Claim 20, Splithoff discloses a apparatus for motion correction in a magnetic resonance imaging system (“Systems and Methods that identify the effect of motion during a medical imaging procedure.” [Abstract]), the apparatus comprising:
processing circuitry configured to: receive data collected from imaging an object by the magnetic resonance imaging system, or image data reconstructed from the collected data (“In response to applied RF pulse signals, the RF coil 18 receives MR signals, e.g., signals from the excited protons within the body as the protons return to an equilibrium position established by the static and gradient magnetic fields. The MR signals are detected and processed by a detector within RF module and the control unit 20 to provide an MR dataset to a processor 22 for processing into an image. In some embodiments, the processor 22 is located in the control unit 20, in other embodiments, the processor 22 is located remotely. A two or three-dimensional k-space storage array of individual data elements in a memory 24 of the control unit 20 stores corresponding individual frequency components including an MR dataset.” [0021]);
estimate a motion parameter indicating a motion of the object (“the motion assessment includes one or more values for respective degree of freedoms describing the 3D motion state of the patient at the time of the acquisition of a chunk of the data relative to an initial position. The initial position may be defined by a rapid motion-free initial scan and the patient's movements are quantified into different positions/poses after the sampling of the current chunk of the acquisition (e.g., echo train/shot/line of the k-space in multi-shot acquisitions).” [0042]);
obtain a data-consistency weighting matrix output from a first neural network by inputting thereto the calculated correlation value, the data-consistency weighting matrix including a number of weighting elements each representing a certainty level of the collected data being corrupted by the motion of the object (“In an embodiment, at Act A150, the control unit 20 determines or calculates a coil mixing matrix including either the CMM as described above or a coil mixing error matrix (also referred to as the difference coil mixing error matrix) for the respective chunk of MR data based on an error between the linearly combined motion data and the linearly combined reference data by determining the linear combination of the error. In certain embodiments, the control unit 20 only uses the CMM and thus the coil mixing error matrix may not be calculated (or may be calculated and not used as an input to the neural network). The coil mixing error matrix may be the difference between a subset of data from the motion free reference and from the definitive imaging procedure being performed and thus may be referred to as the difference coil mixing error matrix.” [0035], “At Act A160, the control unit 20 inputs the coil mixing error matrix and/or the second coil mixing matrix (CMM) into a neural network trained to output a motion assessment for the acquired MRI data. Additional information may be input into the neural network when available. The additional information may include a data consistency error of the current ET, the object size relative to the image matrix size and the relative energy of the current ET to the whole. The additional information may help account for the differences in the depicted object.” [0036], “At Act A170, the control unit 20 provides the motion assessment generated by the neural network to an operator… the motion assessment is or includes a motion score that quantifies the extent of the motion. The motion score may be compared to a threshold score… the control unit 20 ranks the motion scores for each echo train and replaces or reacquires the data for each of the n-th highest ranked echo trains or above a certain level. For example, the control unit 20 may replace the data for the worst 5, 10, or 20% of the acquired data among other levels.” [0037]).
Splithoff does not specifically disclose that the motion parameter is based on a plurality of navigator signals acquired by non-imaging while the collected data is being collected; determine, based on the plurality of navigator signals, a reference navigator signal; calculate a correlation value between each navigator signal of the plurality of navigator signals and the reference navigator signal; and generate, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data.
However, in a similar field of endeavor, Beck teaches a method and magnetic resonance (MR) apparatus for performing an MR examination with prospective motion correction [Abstract].
Beck also teaches the motion parameter is based on a plurality of navigator signals acquired by non-imaging while the collected data is being collected; determining, based on the plurality of navigator signals, a reference navigator signal (“This object is achieved in a method and magnetic resonance (MR) apparatus for performing an MR examination with prospective motion correction, wherein multiple MR signals are acquired. For each MR signal, signal, an examination volume is established and a navigator volume is established for recording navigator signals. The examination volume and the navigator volume are not identical. At least one navigator reference signal is acquired at a time t0, and at least one navigator signal is acquired at a time t1>t0. Motion information is determined from the navigator signal and the navigator reference signal, and the recording parameters are set as a function of the motion information. At least one further magnetic resonance signal is acquired with this setting.” [0016], “The measured signals and the navigator signals are both magnetic resonance signals; one or more images or a spectrum are obtained from the measured signals, and motion information is obtained from the navigator signals.” [0049]);
calculate a correlation value between each navigator signal of the plurality of navigator signals and the reference navigator signal (“the motion information can be determined using a cross-correlation analysis. In this, several cross-correlation coefficients are calculated and the cross-correlation coefficient with the highest numerical value indicates the position of the best match between the reference data and the data obtained from the current navigator echo.” [0025])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff as outlined above with the motion parameter is based on a plurality of navigator signals acquired by non-imaging while the collected data is being collected; determining, based on the plurality of navigator signals, a reference navigator signal; calculating a correlation value between each navigator signal of the plurality of navigator signals and the reference navigator signal as taught by Beck, because it is less prone to error for performing a magnetic resonance examination with a prospective motion correction [0015].
Splithoff in view of Beck does not specifically teach generate, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data.
However, in a similar field of endeavor, Cordero teaches a framework for the recon struction of magnetic resonance images in the presence of rigid motion [Abstract].
Cordero also teaches generate, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data (“The generalized reconstruction with rigid motion correction for parallel multi shot imaging can be formulated in matrix form as: (ˆx, ˆ T)=argminx,T∥AFSTx−y∥2 2, (1) where y denotes the measured k-space data, x the image to be reconstructed, T the rigid motion transformation matrix, S the coil sensitivity matrix, F the Discrete Fourier transform (DFT) encompassing applied k-space oversampling or down sampling, and A a sampling matrix. The forward model for this formulation, originally proposed in[13] for the estimation of x assuming T is known, is depicted in Fig.1 and the different terms are described in Section II-B” [A. Generalized Rigid Motion-Corrected Multishot Reconstruction], “5) T is a matrix of size NS×N given by T= ⎡ ⎢ ⎣ T1 . . . TS ⎤ ⎥ ⎦, where Ts is a matrix of size N×N corresponding to the rigid transformation the underlying structure has been subject to when acquiring the shots. This matrix is described just below.” [B. Model Terms])
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff in view of Beck as outlined above with generate, based on the estimated motion parameter, the obtained data-consistency weighting matrix and the received data or reconstructed image data, motion-corrected image data as taught by Cordero, because it generally improve the quality of reconstructed images [Abstract].
Claim 3 is rejected under 35 U.S.C. 103 as being unpatentable over Splithoff in view of Beck and further in view of Cordero as applied to Claim 2 above, and further in view of Nielsen et al (US20150212182A1; hereinafter referred to as Nielsen).
Regarding Claim 3, Splithoff in view of Beck and further in view of Cordero discloses all limitations noted above except that the step of obtaining the data-consistency weighting matrix further comprises: assigning, based on a comparison of the correlation value calculated for each navigator signal, of the plurality of navigator signals, with an empirical correlation threshold, a particular value to the number of weighting elements of the weighting matrix, the number of weighting elements corresponding to a number of k-space points acquired by one of the plurality of shots that corresponds to the navigator signal.
However, in a similar field of endeavor, Nielsen teaches the modification of the acquisition of magnetic resonance imaging using a dissimilarity matrix classification [0001].
Nielsen also teaches that the step of obtaining the data-consistency weighting matrix further comprises: assigning, based on a comparison of the correlation value calculated for each navigator signal, of the plurality of navigator signals, with an empirical correlation threshold, a particular value to the number of weighting elements of the weighting matrix, the number of weighting elements corresponding to a number of k-space points acquired by one of the plurality of shots that corresponds to the navigator signal (“To use the method you need to define a metric which calculates the distance between two clusters. Again different options exist, but all are based on the dissimilarity matrix values for the elements which belong to the clusters: you can take the minimum, maximum or average of the dissimilarity values of all elements involved.” [0028], “This gives you a decreasing sequence of clusters with an increasing inter-cluster dissimilarity at which the fusion of two clusters takes place. If you plot the fusion dissimilarity against the number of clusters you will notice strong jumps in this graph if the navigator shots can be separated into distinct groups. By putting a threshold to the maximum acceptable inter-cluster distance you select one particular clustering from the hierarchical chain of clusters.” [0029], “the statistical analysis algorithm is operable to determine matrix classification by performing any one of the following: performing a Bayesian analysis, thresholding the dissimilarity matrix, calculating a standard deviation of the dissimilarity matrix, identifying the elements of the dissimilarity matrix outside of a predetermined range and performing a probability based selection.” [0031], “the metric is any one of the following: computing the sum of the squared complex difference between navigator vectors, calculating the difference in the magnitude of the navigator vectors, calculating the absolute value of the difference between navigator vectors, and calculating the correlation between navigator signals. The metric may also include normalizing the navigator vectors. For example the navigator vectors may be normalized using the minimum value of the D-matrix as part of calculating the metric.” [0040]).
It would have been obvious to an ordinary skilled person in the art before the effective filing
date of the claimed invention to modify the system of Splithoff in view of Beck and further in view of Cordero as outlined above with the step of obtaining the data-consistency weighting matrix further comprises: assigning, based on a comparison of the correlation value calculated for each navigator signal, of the plurality of navigator signals, with an empirical correlation threshold, a particular value to the number of weighting elements of the weighting matrix, the number of weighting elements corresponding to a number of k-space points acquired by one of the plurality of shots that corresponds to the navigator signal as taught by Beck, because it can improve the quality of an MR data set by reacquiring data which are disturbed by motion [0118].
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Splithoff in view of Beck in view of Cordero and further in view of Nielsen as applied to Claim 3 above, and further in view of Jiang et al (W. Jiang et al., “Motion robust high resolution 3D free‐breathing pulmonary MRI using dynamic 3D image self‐navigator,” Magnetic Resonance in Medicine, vol. 79, no. 6, pp. 2954–2967, Oct. 2017;hereinafter referred to as Jiang).
Regarding Claim 4, Splithoff in view of Beck in view of Cordero and further in view of Nielsen discloses all limitations noted above except that the assigned particular value is a value within a predefined range, and the assigning step further comprises: assigning, in response to the comparison indicating a stronger correlation, the particular value, which approaches a first end of the predefined range to represent a higher certainty level of the number of k-space points not being corrupted by the motion, and assigning, in response to the comparison indicating a weaker correlation, the particular value, which approaches a second end of the predefined range to represent a higher certainty level of the number of k-space points being corrupted by the motion.
However, in a similar field of endeavor, Jiang teaches a motion robust high resolution 3D free- breathing pulmonary MRI utilizing a novel dynamic 3D image navigator derived directly from imaging data [Abstract].
Jiang also teaches that the assigned particular value is a value within a predefined range, and the assigning step further comprises: assigning, in response to the comparison indicating a stronger correlation, the particular value, which approaches a first end of the predefined range to represent a higher certainty level of the number of k-space points not being corrupted by the motion, and assigning, in response to the comparison indicating a weaker correlation, the particular value, which approaches a second end of the predefined range to represent a higher certainty level of the number of k-space points being corrupted by the motion (“The concept of soft- gating is illustrated in the top branch of Figure 1. The weights effectively take account for motion induced data inconsistency. We use the soft-gating approach by modifying the basic image reconstruction model (Eq. [2]) to incorporate appropriate weights W: Here, W is a diagonal matrix containing the soft-gating weights, which are applied to the data consistency term. Let w[n] be the vector representing the diagonal entries of W. A different weight w[n] is estimated for each radial spoke n, ranging between 0 and 1: where d[n] represents the estimated respiratory motion with respect to the end of expiration or the end of inspiration (we picked the end of expiration state in this manuscript since more time is typically spent in expiration during a respiration cycle), threshold is a threshold of the respiratory motion, and α is a scaling factor. For data experiencing more respiratory motion corruption, their weights are smaller and thus they contribute less to the data consistency term in Equation [3]. “ [Motion Compensated Reconstruction])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Polak in view of Splithoff in view of Beck in view of Cordero and further in view of Nielsen as outlined above with the assigned particular value is a value within a predefined range, and the assigning step further comprises: assigning, in response to the comparison indicating a stronger correlation, the particular value, which approaches a first end of the predefined range to represent a higher certainty level of the number of k-space points not being corrupted by the motion, and assigning, in response to the comparison indicating a weaker correlation, the particular value, which approaches a second end of the predefined range to represent a higher certainty level of the number of k-space points being corrupted by the motion as taught by Jiang, because soft gating is a computationally efficient iterative method in which the data consistency term in the optimization is preferentially weighted based on distances from the chosen respiratory motion state [Introduction].
Claims 5-6 are rejected under 35 U.S.C. 103 as being unpatentable over Splithoff in view of Beck in view of Cordero and further in view of Nielsen as applied to Claim 3 above, and further in view of Oksuz et al (I. Oksuz et al., “Detection and correction of cardiac MRI motion artefacts during Reconstruction from K-Space,” Lecture Notes in Computer Science, pp. 695–703, 2019; hereinafter referred to as Oksuz).
Regarding Claim 5, Splithoff in view of Beck in view of Cordero and further in view of Nielsen discloses all limitations noted above except that the assigned particular value is either a first predefined value or a second predefined value, and the assigning step further comprises: assigning, in response to the comparison indicating the correlation beyond the empirical correlation threshold, the first predefined value to represent a high certainty level of the number of k-space points not being corrupted by the motion, and assigning, in response to the comparison indicating the correlation short of the empirical correlation threshold, the second predefined value to represent a high certainty level of the number of k-space points being corrupted by the motion.
However, in a similar field of endeavor, Oksuz teaches a method to automat ically detect and correct motion-related artefacts in CMR acquisitions during reconstruction from k-space data [Abstract].
Oksuz also teaches the assigned particular value is either a first predefined value or a second predefined value, and the assigning step further comprises: assigning, in response to the comparison indicating the correlation beyond the empirical correlation threshold, the first predefined value to represent a high certainty level of the number of k-space points not being corrupted by the motion, and assigning, in response to the comparison indicating the correlation short of the empirical correlation threshold, the second predefined value to represent a high certainty level of the number of k-space points being corrupted by the motion (“The architecture of our network follows a similar architecture to [14], which was originally developed for video classification using a spatio-temporal 3D CNN. In our case we use the third dimension as the time component and use 2D+time mid-ventricular sequences as the input to the network. Each image sequence has 50 time frames. The network has 4 convolutional layers and 4 pooling layers, 1 fully-connected layer and a softmax loss layer to predict corrupted k-space lines.” [3.1 Network Architecture], “The detection loss is the cross entropy loss, defined as: Ldetection(pr, y) = 1 Nl −(ylog(pr) +(1−y)log(1−pr)) where y is a binary indicator (0 or 1) indicating if a k-space line is corrupted or not and pr is predicted probability of the line being uncorrupted. Np denotes the total number of k-space lines in an image.” [3.2 Loss Function and Training])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Polak in view of Splithoff in view of Beck in view of Cordero and further in view of Nielsen as outlined above with the assigned particular value is either a first predefined value or a second predefined value, and the assigning step further comprises: assigning, in response to the comparison indicating the correlation beyond the empirical correlation threshold, the first predefined value to represent a high certainty level of the number of k-space points not being corrupted by the motion, and assigning, in response to the comparison indicating the correlation short of the empirical correlation threshold, the second predefined value to represent a high certainty level of the number of k-space points being corrupted by the motion as taught by Oksuz, because it preserves the quality of uncorrupted images and therefore can be also utilized as a general image reconstruction algorithm [Abstract].
Regarding Claim 6, Splithoff in view of Beck in view of Cordero and further in view of Nielsen discloses all limitations noted above except that the second predefined value is set at 0 to reject the number of k-space points because of the motion, such that the number of k-space points are not to be used in reconstruction of the image data, and the estimating step is omitted.
However, in a similar field of endeavor, Oksuz teaches the second predefined value is set at 0 to reject the number of k-space points because of the motion, such that the number of k-space points are not to be used in reconstruction of the image data, and the estimating step is omitted (“By training both networks end-to-end we are able to ignore motion corrupted k-space lines during the reconstruction.” [1 Introduction], “The architecture of our network follows a similar architecture to [14], which was originally developed for video classification using a spatio-temporal 3D CNN. In our case we use the third dimension as the time component and use 2D+time mid-ventricular sequences as the input to the network. Each image sequence has 50 time frames. The network has 4 convolutional layers and 4 pooling layers, 1 fully-connected layer and a softmax loss layer to predict corrupted k-space lines.” [3.1 Network Architecture], “The detection loss is the cross entropy loss, defined as: Ldetection(pr, y) = 1 Nl −(ylog(pr) +(1−y)log(1−pr)) where y is a binary indicator (0 or 1) indicating if a k-space line is corrupted or not and pr is predicted probability of the line being uncorrupted. Np denotes the total number of k-space lines in an image.” [3.2 Loss Function and Training])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Polak in view of Splithoff in view of Beck in view of Cordero and further in view of Nielsen as outlined above with the second predefined value is set at 0 to reject the number of k-space points because of the motion, such that the number of k-space points are not to be used in reconstruction of the image data, and the estimating step is omitted as taught by Oksuz, because it preserves the quality of uncorrupted images and therefore can be also utilized as a general image reconstruction algorithm [Abstract].
Claim 10-12, 14-15, & 21 is rejected under 35 U.S.C. 103 as being unpatentable over Splithoff in view of Beck in view of Cordero as applied to Claim 2 above, and further in view of Oksuz.
Regarding Claim 10, Splithoff in view of Beck in view of Cordero discloses all limitations noted above except that the step of generating the motion-corrected image data further comprises: applying the received data or the reconstructed image data, the weighting matrix, and the motion parameter to a neural network; and obtaining, as the motion-corrected image data, an output of the neural network.
However, in a similar field of endeavor, Oksuz teaches the step of generating the motion-corrected image data further comprises: applying the received data or the reconstructed image data, the weighting matrix, and the motion parameter to a neural network; and obtaining, as the motion-corrected image data, an output of the neural network (“The first network is an artefact detection network which is used to identify potentially corrupted k-space lines and hence define a data consistency term. and the second network is a recurrent convolutional neural network (RCNN) used for reconstruction using this data-consistency term [12]” [3. Methods], “we use the output of this k-space artefact detection network to introduce a data consistency term to be used by an image reconstruction network.” [1 Introduction], “Fig.2: The CNN architecture for motion artefact correction. The proposed net work architecture consists of two building blocks 1) A corrupted k-space line detection network to define the data-consistency term; 2) A recurrent neural network (RCNN) architecture to correct image artefacts” [Fig. 2], Fig. 2 is included below)
PNG
media_image1.png
248
620
media_image1.png
Greyscale
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Polak in view of Splithoff in view of Beck in view of Cordero as outlined above with the second predefined value is set at 0 to reject the number of k-space points because of the motion, such that the number of k-space points are not to be used in reconstruction of the image data, and the estimating step is omitted as taught by Oksuz, because it preserves the quality of uncorrupted images and therefore can be also utilized as a general image reconstruction algorithm [Abstract].
Regarding Claim 11, Splithoff in view of Beck in view of Cordero discloses all limitations noted above except that the applying step further comprises applying the obtained data or the reconstructed image data, the generated weighting matrix, and the estimated motion parameter to a model-driven deep learning framework having a pre-determined number of iterations, wherein the model-driven deep learning framework includes a neural network and performs a data consistency process.
However, in a similar field of endeavor, Oksuz teaches the applying step further comprises applying the obtained data or the reconstructed image data, the generated weighting matrix, and the estimated motion parameter to a model-driven deep learning framework having a pre-determined number of iterations, wherein the model-driven deep learning framework includes a neural network and performs a data consistency process (“This network reconstructs high quality cardiac MR images from highly undersampled k-space data by jointly exploiting the dependencies of the temporal sequences as well as the iterative nature of traditional optimisation algorithms.” [3.1 Network Architecture])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Polak in view of Splithoff in view of Beck in view of Cordero as outlined above with the applying step further comprises applying the obtained data or the reconstructed image data, the generated weighting matrix, and the estimated motion parameter to a model-driven deep learning framework having a pre-determined number of iterations, wherein the model-driven deep learning framework includes a neural network and performs a data consistency process as taught by Oksuz, because it preserves the quality of uncorrupted images and therefore can be also utilized as a general image reconstruction algorithm [Abstract].
Regarding Claim 12, Splithoff in view of Beck in view of Cordero discloses all limitations noted above except that the neural network is a U-net, a residual U-net, a residual network, an inception-residual network, or a linear convolutional network.
However, in a similar field of endeavor, Oksuz teaches the neural network is a U-net, a residual U-net, a residual network, an inception-residual network, or a linear convolutional network (“This network reconstructs high quality cardiac MR images from highly undersampled k-space data by jointly exploiting the dependencies of the temporal sequences as well as the iterative nature of traditional optimisation algorithms.” [3.1 Network Architecture])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Polak in view of Splithoff in view of Beck in view of Cordero as outlined above with the neural network is a U-net, a residual U-net, a residual network, an inception-residual network, or a linear convolutional network as taught by Oksuz, because it preserves the quality of uncorrupted images and therefore can be also utilized as a general image reconstruction algorithm [Abstract].
Regarding Claim 14, Splithoff in view of Beck discloses all limitations noted above except that the data consistency process uses a conjugate gradient iteration algorithm, a proximal gradient algorithm, an orthogonal matching pursuit algorithm, an iterative hard thresholding algorithm, a split Bregman-based algorithm, or a gradient descent algorithm.
However, in a similar field of endeavor, Cordero teaches that the data consistency process uses a conjugate gradient iteration algorithm, a proximal gradient algorithm, an orthogonal matching pursuit algorithm, an iterative hard thresholding algorithm, a split Bregman-based algorithm, or a gradient descent algorithm (“The first of these subproblems, i.e., that of reconstructing the image x in the presence of rigid motion, is considered in [13], where the system THSHFHAHAFSTˆx=THSHFHAHy (6) is solved by means of the conjugate gradient (CG) algorithm.” [C. Problem Solving]).
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Splithoff in view of Beck as outlined above with the data consistency process uses a conjugate gradient iteration algorithm, a proximal gradient algorithm, an orthogonal matching pursuit algorithm, an iterative hard thresholding algorithm, a split Bregman-based algorithm, or a gradient descent algorithm as taught by Cordero, because it preserves the quality of uncorrupted images and therefore can be also utilized as a general image reconstruction algorithm [Abstract].
Regarding Claim 15, Splithoff discloses that further comprising: obtaining fully-sampled motion-free k-space data acquired by a plurality of shots; generating motion-corrupted k-space data by simulating corresponding influences caused by motions having different motion parameters on different shots; generating navigator data corresponding to the motions having different motion parameters; generating the data-consistency weighting matrixes based on the navigator data; and using the fully-sampled motion-free k-space data, the motion-corrupted k-space data, the data-consistency weighting matrixes, and the motion parameters to train the deep learning framework, so as to learn a mapping from the motion-corrupted k-space data to the fully- sampled motion-free image data (“the method comprising: acquiring a motion free reference; calculating, based on the motion free reference, a scout coil mixing matrix representing a linear combination of coils of the imaging system;” [0005], “The training data for the model includes ground truth data or gold standard data acquired or simulated prior to training the neural network. Ground truth data and gold standard data is data that includes correct or reasonably accurate labels that are verified manually or by some other accurate method. The training data may be acquired at any point prior to inputting the training data into the neural network. The neural network may input the training data (e.g., CMM data, CMEM data and other information) and output a prediction or classification, for example of motion. In an example the prediction may include a motion score that quantifies the amount of motion in the input data. In another example, the prediction or classification may include motion parameters that quantify and describe the detected motion. The prediction is compared to the annotations (e.g., motion scores or values for respective degree of freedoms describing the 3D motion state of the patient at the time of the acquisition of a chunk of the data relative to an initial position) from the training data. A loss function may be used to identify the errors from the comparison. The loss function serves as a measurement of how far the current set of predictions are from the corresponding true values. Some examples of loss functions that may be used include Mean-Squared-Error, Root-Mean-Squared-Error, L2 norm, and Cross-entropy loss. Mean Squared Error loss, or MSE for short, is calculated as the average of the squared differences between the predicted and actual values. Root-Mean Squared Error is similarly calculated as the average of the root squared differences between the predicted and actual values. The L2 norm is similar to the Root-Mean Squared error and is calculated as the square root of the sum of the squared vector values. The max norm that is calculated as the maximum vector values. For cross-entropy loss each predicted probability is compared to the actual class output value (0 or 1) and a score is calculated that penalizes the probability based on the distance from the expected value. The penalty may be logarithmic, offering a small score for small differences (0.1 or 0.2) and enormous score for a large difference (0.9 or 1.0). During training and over repeated iterations, the neural network attempts to minimize the loss function as the result of a lower error between the actual and the predicted values means the neural network has done a good job in learning. Different optimization algorithms may be used to minimize the loss function, such as, for example, gradient descent, Stochastic gradient descent, Batch gradient descent, Mini-Batch gradient descent, among others. The process of inputting, outputting, comparing, and adjusting is repeated for a predetermined number of iterations with the goal of minimizing the loss function. Once adjusted and trained, neural network is ready to be applied to unseen data. In an embodiment, the neural network may be replaced by a model that uses, for example, a pattern matching technique. The model may be configured or trained using iterative machine learning techniques.” [0039]).
Regarding Claim 21, Splithoff discloses that the step of obtaining the data-consistency weighting matrix further comprises obtaining the data-consistency weighting matrix output from a first neural network by inputting thereto the calculated correlation value (“In an embodiment, at Act A150, the control unit 20 determines or calculates a coil mixing matrix including either the CMM as described above or a coil mixing error matrix (also referred to as the difference coil mixing error matrix) for the respective chunk of MR data based on an error between the linearly combined motion data and the linearly combined reference data by determining the linear combination of the error. In certain embodiments, the control unit 20 only uses the CMM and thus the coil mixing error matrix may not be calculated (or may be calculated and not used as an input to the neural network). The coil mixing error matrix may be the difference between a subset of data from the motion free reference and from the definitive imaging procedure being performed and thus may be referred to as the difference coil mixing error matrix.” [0035], “At Act A160, the control unit 20 inputs the coil mixing error matrix and/or the second coil mixing matrix (CMM) into a neural network trained to output a motion assessment for the acquired MRI data. Additional information may be input into the neural network when available. The additional information may include a data consistency error of the current ET, the object size relative to the image matrix size and the relative energy of the current ET to the whole. The additional information may help account for the differences in the depicted object.” [0036], “At Act A170, the control unit 20 provides the motion assessment generated by the neural network to an operator… the motion assessment is or includes a motion score that quantifies the extent of the motion. The motion score may be compared to a threshold score… the control unit 20 ranks the motion scores for each echo train and replaces or reacquires the data for each of the n-th highest ranked echo trains or above a certain level. For example, the control unit 20 may replace the data for the worst 5, 10, or 20% of the acquired data among other levels.” [0037]),
Splithoff in view of Beck in view of Cordero does not specifically teach the step of generating the motion-corrected image data further comprises inputting the estimated motion parameter, the obtained data-consistency weighting matrix, and the received data or the reconstructed image data to a second neural network, and obtaining, as the motion-corrected image data, an output of the second neural network.
However, in a similar field of endeavor, Oksuz teaches that the step of generating the motion-corrected image data further comprises inputting the estimated motion parameter, the obtained data-consistency weighting matrix, and the received data or the reconstructed image data to a second neural network, and obtaining, as the motion-corrected image data, an output of the second neural network (“We propose a k-space artefact detection network that generates an individual data consistency term for any given acquisition and converts the image artefact correction task to an undersampled image reconstruction problem, which is sub sequently addressed by an algorithm developed for reconstruction of undersam pled CMR acquisitions (see the illustration in Fig. 1). Our proposed method is evaluated using 300 cine SSFP (2D+time) CMR datasets from the UK Biobank.” [1 Introduction], “Second, we use the output of this k-space artefact detection network to introduce a data consistency term to be used by an image reconstruction network.”).
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Splithoff in view of Beck in view of Cordero and further in view of Oksuz as outlined above with the neural network is a U-net, a residual U-net, a residual network, an inception-residual network, or a linear convolutional network as taught by Aggarwal, because it translates to improved performance, primarily when the available GPU memory restricts the number of iterations [Abstract].
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Splithoff in view of Beck in view of Cordero and further in view of Oksuz as applied to Claim 12 above, and further in view of Aggarwal et al (H. K. Aggarwal, M. P. Mani, and M. Jacob, “MoDL: Model-based deep learning architecture for inverse problems,” IEEE Transactions on Medical Imaging, vol. 38, no. 2, pp. 394–405, Feb. 2019; hereinafter referred to as Aggarwal).
Regarding Claim 13, Splithoff in view of Beck in view of Cordero and further in view of Oksuz discloses all limitations noted above except that the neural network is a complex U-net, and a plurality of parameters of the complex U-net are shared across the pre- determined number of iterations.
However, in a similar field of endeavor, Aggarwal teaches that the neural network is a complex U-net, and a plurality of parameters of the complex U-net are shared across the pre- determined number of iterations (“or many measurement operators (e.g Fourier sampling, blurring, projection imaging), AHA is a translation-invariant operator; the convolutional structure makes it possible for CNNs to solve such problems [34]. However, the receptive field of the CNN has to be comparable to the support of the point spread function corresponding to AHA. In applications involving Fourier sampling or projection imaging, the receptive field of the CNNs has to be the same as that of the image; large networks such as UNET with several layers are required to obtain such a large receptive field.” [B. Deep learned image reconstruction: the state-of-the-art])
It would have been obvious to an ordinary skilled person in the art before the effective filing date of the claimed invention to modify the system of Splithoff in view of Beck in view of Cordero and further in view of Oksuz as outlined above with the neural network is a U-net, a residual U-net, a residual network, an inception-residual network, or a linear convolutional network as taught by Aggarwal, because it translates to improved performance, primarily when the available GPU memory restricts the number of iterations [Abstract].
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to STEVEN MALDONADO whose telephone number is 703-756-1421. The examiner can normally be reached 8:00 am-4:00 pm PST M-Th Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at
http://www.uspto.gov/interviewpractice. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Christopher Koharski can be reached on (571) 272-7230. Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Steven Maldonado/
Patent Examiner, Art Unit 3797
/CHRISTOPHER KOHARSKI/Supervisory Patent Examiner, Art Unit 3797
Prosecution Insights