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 .
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
Claims 1-5, 7, 9-12, and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Zeng et al, (US-PGPUB 20200134887) in view of Applicant’s Admitted Prior Art, “AAPA”, and further in view of Fasil et al, (“Robust partial Fourier reconstruction for diffusion-weighted imaging using a recurrent convolutional neural network”, Magnetic Resonance in Medicine, Nov. 7, 2021)
In regards to claim 1, Zeng discloses a computer-implemented method of
processing magnetic resonance imaging (MRI) data, the method comprising:
obtaining MRI data that k-space (Fourier domain) data 102 to produce acquired k-space data 104, [i.e., obtaining MRI data, “k-space data 104”, that samples k-space, “sampling k-space (Fourier domain) data 102”]);
under-sampled k-space as input and the reconstructed image as output, based on learning processing pipeline. During MRI acquisition, sampling pattern mask 100 is used for sampling k-space (Fourier domain) data 102 to produce acquired k-space data 104, and an image 108 is reconstructed from the acquired k-space data 104 by the environment 106, [i.e., based on one or more settings of the partial Fourier acquisition scheme, “implicit by using the under-sampled k-space as input”, configuring a processing pipeline, “implicit by using learning processing pipeline”, to reconstruct an MRI image, “image 108”, based on the MRI data, “implicit by k-space data”]); and
providing the MRI data to the processing pipeline and obtaining, from the processing pipeline, the MRI image, (see at least: Fig. 1A, Par. 0026, where the k-space data 104, “i.e., MRI data”, is provides to the learning processing pipeline, “i.e., processing pipeline”, and obtaining the MRI image 108 from the learning processing pipeline”]).
Although disclosing reconstructing an MRI image based on the under-sampled k-space; Zeng does not expressly disclose that the MRI image being reconstructed based on one or more settings of the partial Fourier acquisition scheme; and that the obtained MRI data asymmetrically samples k-space based on a partial Fourier acquisition scheme.
However, the obtaining MRI data that asymmetrically samples k-space based on a partial Fourier acquisition scheme, is well-known in the art, (see at least: Applicant’s Admitted Prior Art, “AAPA”, par. 0005, lines 1-2). Moreover, the reconstructing MRI image based on one or more settings of the partial Fourier acquisition scheme, is not new. In fact, Fasil et al discloses the reconstructing MRI image based on one or more settings of the partial Fourier acquisition scheme, (see at least: Fig. 1, Page 2021, section 3.1, Network inputs were generated by retrospective PF sampling of the coil-combined data followed by an inverse Fourier transform; and from Page 2020, section 2.1, the forward model in PF-sampled MRI can be expressed as:
y
=
A
x
+
n
,
[i.e., the reconstruction information for the MRI image is included in the form of partial Fourier sampling, which implicit the reconstruction of MRI image based on Partial Fourier sampling]).
Zeng, AAPA, and Fasil are combinable because they are both concerned with MRI image reconstruction. Therefore, it would have been obvious to a person of ordinary skill in the art, to modify Zeng, to include the reconstruction information for the MRI image in the form of partial Fourier sampling, as though by Fasil, in order to reconstruct the MRI image (see Fig. 1, and section 2.1).
In regards to claim 2, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 1.
Fasil further discloses wherein the processing pipeline comprises: a trained function trained to solve, in a current estimate of the MRI image, a deblurring task to reduce blurring artifacts, (see at least: Fig. 1, where CNN corresponds to the trained function. Further, page 2026, section 4.3, both POCS and DRPF alleviate blurring introduced by zero-filling effectively, and page 2029, section 5, [i.e., a trained function, trained, “see at least: Fig. 1, training CNN”, to solve, in a current estimate of the MRI image, “MRI image shown in Fig. 7”, a deblurring task to reduce blurring artifacts, “both POCS and DRPF alleviate blurring”]); and
a filter block arranged downstream of the trained function and configured to enforce data consistency between an output of the trained function and the MRI data, (see at least: page 2022, section 3.2, using the average of the ground-truth repetitions computed in the same manner, end-to-end training of the network was performed by minimizing a loss function, …. All convolutions used kernels of size 3X3 with zero-padding, [i.e., a filter block arranged downstream of the trained function, “kernels filters”, and configured to enforce data consistency between an output of the trained function and the MRI data, “implicit by minimizing a loss function based on the ground-truth”]).
In regards to claim 3, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Fasil further discloses wherein: the processing pipeline comprises multiple iterations of the trained function and the filter block, (see at least: Fig. 1, k iterations at bottom block of Fig. 1, correspond to the multiple iterations of the trained function and right bottom block of Fig. 1, corresponds to the filter block); and an input of the trained function in a subsequent one of the multiple iterations is based on an output of the filter block in a preceding one of the multiple iterations, (see at least: Fig. 1, where the input of the second CNN of Fig. 1, is the output of preceding first CNN of the convolutional kernels, “the filter block” in a preceding one of the multiple iterations)
In regards to claim 4, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Fasil further discloses wherein the filter block is adapted to prioritize, by hard filtering or soft filtering, measured samples included in the MRI data over reconstructed samples included in the output of the trained function, (see at least: page 2022, section 3.2, right-hand-column, the filter block (kernels) of Fig. 1, is implicitly adapted to prioritize, by hard filtering or soft filtering, measured samples included in the MRI data over reconstructed samples included in the output of the trained function, based on performing hard projection for data consistency for all iterations).
In regards to claim 5, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Zeng further discloses wherein the trained function comprises a generative neural network comprising: a first input that is based on randomized data, a second input that is based on the one or more settings, and a third input that is based on the current estimate of the MRI image, (see at least: par. 0029, the network 122 is preferably a generative adversarial network … using randomly-weighted sampling patterns; and from par. 0012, the image quality metric of the reward (110 in Fig. 1), uses discriminators from trained generative adversarial networks, [i.e., the trained function, (fig. 1a), comprises a generative neural network, (122 in Fig. 1), comprising a first input that is based on randomized data, “implicit by using randomly-weighted sampling patterns”, a second input that is based on the one or more settings, (104 in Fig. 1A), and a third input that is based on the current estimate of the MRI image, “discriminators as shown in Fig. 1c”]).
In regards to claim 7, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Fasil further discloses wherein the configuring of the processing pipeline comprises: providing, to the filter block, a representation of at least one of a first section of the k-space or a second section of the k-space, (see at least: Fig. 1, where k iterations, “representation of k-space”, are provided to the filter block, “upper middle block, (CNN and DC), in Fig. 1”), the first section being sampled by the partial Fourier acquisition scheme and the second section not being sampled by the partial Fourier acquisition scheme, (see at least: section 3.2.3, implicit by acquiring scans: one without PF sampling and one with prospective PF sampling of 5/8).
In regards to claim 9, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Fasil further discloses wherein the trained function is further trained to solve, in the current estimate of the MRI image, a denoising task adapted to reduce noise in the MRI image, (see at least: page 2022, section 3.2, right-hand-column, second paragraph, implicit by “denoising”)
In regards to claim 10, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Zeng further discloses wherein the configuring of the processing pipeline comprises determining combination parameters for a combination of outputs of multiple parallel branches of the processing pipeline, (see at least: Fig. 1, par. 0031, as for output, the reconstruction technique could output any combination of k-space, image, and arbitrary data, “i.e., the parameters k-space, image, and arbitrary data, are implicitly determined, and output as combined parameters”).
In regards to claim 11, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
Fasil further discloses wherein the one or more settings comprise at least one of a partial Fourier factor of the partial Fourier acquisition scheme, or at least one direction associated with the partial Fourier acquisition scheme, (see at least: Abstract, and section 3.3, “deep recurrent partial Fourier”; and Fig. 1, “PF parameter”).
Regarding claim 12, claim 12 recites substantially similar limitations as set forth in claim 1. As such, claim 10 is rejected for at least similar rational.
The Examiner further acknowledged the following additional limitation(s): “a non-transitory computer-readable storage medium with an executable program stored thereon, that when executed, instructs a processor to perform the method of claim 1”.
However, Zeng discloses the “non-transitory computer-readable storage medium with an executable program stored thereon”, (see at least: par. 0025, “reconstruction techniques may include, for example, algorithms”, which implicit the use of the non-transitory computer-readable storage medium).
Regarding claim 18, claim 18 recites substantially similar limitations as set forth in claim 1. As such, claim 18 is rejected for at least similar rational.
The Examiner further acknowledged the following additional limitation(s): “an apparatus, comprising: one or more processors; and memory storing instructions that, when executed by the one or more processors”. However, Zeng et al discloses the “apparatus, comprising: one or more processors; and memory storing instructions that, when executed by the one or more processors”, (see at least: Fig. 2, Par. 0027, the MRI computer 212, implicitly comprises processor and memory)
Regarding claim 19, claim 19 recites substantially similar limitations as set forth in claim 2. As such, claim 19 is rejected for at least similar rational.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Zeng et al, “AAPA”, and Fasil et al, as applied to claim 2 above; and further in view of Wu et al, (US-PGPUB 20230055826)
The combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
The combine teaching Zeng, AAPA, and Fasil as whole does not expressly disclose wherein: the trained function is adapted to determine a latent representation of the current estimate of the MRI image in a coordinate space, the coordinate space comprising a respective dimension for each of at least one k-space direction associated with the partial Fourier acquisition scheme; and the processing pipeline comprises a coordinate decoder coupled in-between the trained function and the filter block, the coordinate decoder being adapted to determine an output based on the one or more settings, (see at least: Fig. 1, Par. 0016, providing deep learning based methods for magnetic resonance imaging (MRI) image reconstruction from partial Fourier-space (i.e., k-space) data; and from par. 0019, such deep learning model is also applicable to the MRI image reconstruction from 3D or 4D k-space data, in which partial Fourier sampling can exist in more than one direction during phase or/and frequency encoding, [i.e., determining a latent representation of the current estimate of the MRI image in a coordinate space, “implicit by MRI image reconstruction from 3D or 4D k-space data, in which partial Fourier sampling can exist in more than one direction]. Further, from Par. 0008, it is disclosed that the kernel sizes of the five convolutional layers are 9×9, 7×7, 5×5, 5×5 and 3×3, respectively, and the numbers of kernels are 128, 64, 32, 32 and 2, respectively, and such models can be expanded to 3D and multi-channel data, [i.e., the processing pipeline, “implicit by 3D and multi-channel data”, comprises a coordinate decoder, coupled in-between the trained function and the filter block, “kernel layers are implicitly coupled in-between the trained function and the filter block”, the coordinate decoder being adapted to determine an output based on the one or more settings, “the output is implicitly determined based on the partial Fourier 3D k-space data”]).
Zeng, AAPA, Fasil, and Wu et al are combinable because they all concerned with MRI image reconstruction. Therefore, it would have been obvious to a person of ordinary skill in the art, to modify Zeng, AAPA, and Fasil, to use the 3D k-space data, as though by Wu, in order to perform the MRI image reconstruction from 3D k-space data, in which partial Fourier sampling can exist in more than one direction, (Wu, par. 0019)
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Zeng et al, “AAPA”, and Fasil et al, as applied to claim 2 above; and further in view of Chen et al, (US-PGPUB 20240062438)
The combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
The combine teaching Zeng, AAPA, and Fasil as whole does not expressly disclose wherein the trained function is further trained to solve, in the current estimate of the MRI image, a super-resolution task adapted to increase a resolution of the MRI image.
However, Chen et al discloses wherein the trained function is further trained to solve, in the current estimate of the MRI image, a super-resolution task adapted to increase a resolution of the MRI image, (see at least: Abstract, and Par. 0014, implicit by solving an inverse problem, where the invertible neural network may be trained to learn a mapping from the ground truth to the input data, and may subsequently apply an inverse of the mapping (e.g., at an inference time) to complete, such as super-resolution, “i.e., adapting a super-resolution task to increase a resolution of the MRI image”).
Zeng, AAPA, Fasil, and Chen are combinable because they all concerned with MRI image reconstruction. Therefore, it would have been obvious to a person of ordinary skill in the art, to modify Zeng, AAPA, and Fasil, to apply the inverse mapping, as though by Chen, in order to perform the image super-resolution, (Chen, see Abstract, last two lines)
Claims 13-17 are rejected under 35 U.S.C. 103 as being unpatentable over Zeng et al, “AAPA”, and Fasil et al, as applied to claim 2 above; and further in view of Xiao et al, (” Partial Fourier reconstruction of complex MR images using complex-valued convolutional neural networks”, Magnetic Resonance in Medicine, 09/16/2029)
In regards to claim 13, the combine teaching Zeng, AAPA, and Fasil as whole discloses the limitations of claim 2.
The combine teaching Zeng, AAPA, and Fasil as whole does not expressly
disclose obtaining ground-truth MRI data that fully samples the k-space; augmenting the ground-truth MRI data by discarding samples in the k-space based on multiple settings of the partial Fourier acquisition scheme, to obtain multiple training MRI data for the multiple settings; and training the trained function by inputting each of the multiple training MRI data and minimizing a loss determined based on an output of the trained function for each of the multiple training MRI data and with respect to the ground-truth MRI data.
However, Xiao discloses obtaining ground-truth MRI data that fully samples the k-space, (see at least: section 2.4, implicit by obtaining fully sampled T1w GRE, T2w FSE, T2*w G RE and PDw FSE data);
augmenting the ground-truth MRI data by discarding samples in the k-space based on multiple settings of the partial Fourier acquisition scheme, to obtain multiple training MRI data for the multiple settings, (see at least: page 1003, left-hand-column, under section 2.4, all database used for training and testing were retrospectively PF sampled with unacquired k-space data zero filled for specific PF fractions, or specific PF fractions, which implicit that the data was retroactively under-sampled to result in data with different partial Fourier factors); and
training the trained function by inputting each of the multiple training MRI data and minimizing a loss determined based on an output of the trained function for each of the multiple training MRI data and with respect to the ground-truth MRI data, (see at least: page 1002, left-hand-column, under section 2.3, during training, the parameters are estimated using the training dataset by minimizing a loss function, and equation 9, “i.e., implicitly minimizing a loss … with respect to the ground-truth MRI data”).
Zeng, AAPA, Fasil, and Xiao are combinable because they all concerned with MRI image reconstruction. Therefore, it would have been obvious to a person of ordinary skill in the art, to modify Zeng, AAPA, and Fasil, to estimate parameters during training based on minimizing a loss function, as though by Xiao, in order to effectively reconstruct MRI data, (Xiao, Abstract, “conclusion”)
In regards to claims 14-15, the combine teaching Zeng, AAPA, Fasil, and Xiao as whole discloses the limitations of claim 13.
Xiao further discloses wherein the configuring of the processing pipeline comprises: selecting between multiple predefined sets of parameters for at least a section of the processing pipeline; and/or determining a set of parameters of at least a section of the processing pipeline by interpolating in-between predefined sets of parameters, wherein: each of the predefined sets of parameters is associated with a respective predefined partial Fourier factor, and the interpolating is based on a relation of the partial Fourier factor of the partial Fourier acquisition scheme with respect to the predefined partial Fourier factors, (see at least: section 2.4, second paragraph, where the acquisition predefined set of parameters are selected according the neural network used for the parallel MRI reconstruction; and from page 1003, left-hand-column, the datasets were then retrospectively PF sampled with unacquired k-space data zero filled
for specific PF fractions deep, which implicit that the data was retroactively under-sampled to result in data with different partial Fourier factors, which implicit that each of the predefined sets of parameters is associated with a respective predefined partial Fourier factor).
In regards to claim 16 the combine teaching Zeng, AAPA, Fasil, and Xiao as whole discloses the limitations of claim 13.
Zeng further discloses wherein the configuring of the processing pipeline comprises determining combination parameters for a combination of outputs of multiple parallel branches of the processing pipeline, (see at least: Fig. 1, par. 0031, as for output, the reconstruction technique could output any combination of k-space, image, and arbitrary data, “i.e., the parameters k-space, image, and arbitrary data, are implicitly determined, and combined as output parameters of multiple parallel branches of the processing pipeline”).
In regards to claim 17 the combine teaching Zeng, AAPA, Fasil, and Xiao as whole discloses the limitations of claim 13.
Fasil further discloses wherein the one or more settings comprise at least one of a partial Fourier factor of the partial Fourier acquisition scheme, or at least one direction associated with the partial Fourier acquisition scheme, (see at least: Abstract, and section 3.3, “deep recurrent partial Fourier”; and Fig. 1, “PF parameter”).
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to AMARA ABDI whose telephone number is (571)272-0273. The examiner can normally be reached 9:00am-5:30pm.
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, Vu Le can be reached at (571) 272-7332. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
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.
/AMARA ABDI/Primary Examiner, Art Unit 2668 03/13/2026