Detailed Action
1. Claims 1-20 are pending in this Application.
Notice of Pre-AIA or AIA Status
2. 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 Amendment
3. Applicant’s response to the last Office Action filed on 08/11/2025 has been entered and made of record.
4. Claims1 and 9 have been amended. New claims 17-20 have been added.
.
Response to Argument
5. The Applicant’s argument filed on 12/12/2025 is fully consider. For Examiner response see discussion below.
i. Applicant have amended claim 1 and 9 by adding “generating, from a set of frequency representations of the noisy patches, a set of atoms for the dictionary that provide”, and substantially argue that the applied prior art Li teaches "the initial dictionaries for DL-GSGR and K-SVD are chosen as the overcomplete DCT." Li, page 3452. Accordingly, these approaches depend on pre- existing DCT atoms rather than generating atoms from frequency representations of the noisy patches as recited by amended claims 1 and 9.”
As to above argument Examiner respectfully disagrees with the Applicant’s argument for the reason discuss below:
Li specifically disclosed “The initial dictionaries for DL-GSGR and K-SVD are chosen as the overcomplete DCT. The fixed (group) sparsity level is used as the stopping criterions of DL-GSGR and K-SVD in (group) sparse coding stage. In this experiment, the fixed (group) sparsity level is set to 3. The RMSE is used as the objective function. The objective function values of the K-SVD and DL-GSGR (with different group size) are reported as a function of iteration number.” See Section B 1st and 2nd pars.
The above statement describe noise estimation based on DCT function, which is a function of frequency, the dictionary which is learned (updated) from the noisy data using the K-SVD algorithm to best match the image structure, and the noisy image is reconstructed by representing patches as a sparse combination of atoms from the learned dictionary. Thus, based on the above discussing Li teaches a method of generating atoms from frequency representations of the noisy patches based on DCT and K-SVD dictionary learning, where the DCT is frequency domain function.
More specifically Noise Estimation, K-SVD Dictionary Learning described as follows
Noise Estimation: The DCT is used to transform image patches, where the high-frequency coefficients (corresponding to high DCT frequencies) are assumed to be dominated by noise rather than image structure.
K-SVD Dictionary Learning: The dictionary is then learned (updated) from the noisy data using the K-SVD algorithm to best match the image structure.
Sparse Representation: The noisy image is reconstructed by representing patches as a sparse combination of atoms from the learned dictionary
Claim Rejections - 35 USC § 102
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 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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless -
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale or otherwise available to the public before the effective filing date of the claimed invention.
6. Claims 1-3, 5-11 and 13-20 are rejected under 35 U.S.C. 102(a)(1)/102(a)(2) as being anticipated by LI et al., ( hereafter LI), “Group-Sparse Representation With Dictionary Learning for Medical Image Denoising and Fusion”, IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 59, NO. 12, DECEMBER 2012.
As to claim 1, LI teaches A method for denoising a medical image (Abstract, 3-D a medical image denoising and fusion) , comprising:
obtaining a medical image of a subject (Fig.3, page 3452 right col. section B , page 3455, right col., section V, the registered multimodality medical images Xk (k = 1, 2, . . .,K) are obtained and divided into overlapping patches {xki }i=1 P, with a square area of n pixels and ordered as column vectors, where K and Pare the number of images and the total patch number of each image, respectively.);
determining a set of noisy patches from the medical image (Fig. 1, page 3453 section IV A, 3D Medical Image Denoising Model, this section cover a method of 3D medical image denoising , where the image is corrupted by Gaussian noise. The denoising process is carried out based on a plurality of local patches with a square area of n pixels extracted from the kth slice of noisy medical image Y, and ordered as n-dimension column vector, where Rik is a binary matrix extracting the patch from 3-D image);
determining a dictionary based upon the set of noisy patches by generating, from a set of frequency representations of the noisy patches, a set of atoms for the dictionary that provide a sparse representation of the medical image (Fig.1, Table 1, Section B 1st and 2nd pars., The initial dictionaries for DL-GSGR and K-SVD are chosen as the overcomplete DCT. The fixed (group) sparsity level is used as the stopping criterions of DL-GSGR and K-SVD in (group) sparse coding stage. In this experiment, the fixed (group) sparsity level is set to 3. The RMSE is used as the objective function. The objective function values of the K-SVD and DL-GSGR (with different group size) are reported as a function of iteration number.” Thus, based on the above discussing Li teaches generating atoms from frequency representations of the noisy patches based on DCT and K-SVD Dictionary Learning);
denoising the set of noisy patches to create a set of denoised patches (section A right col., 2nd par., equation (14) , equation(14) that describes denoised patch );
denoise the medical image by reconstructing the denoised set of patches into a denoised version of the medical image (Fig.1 , Table 1, (page 3454 right col, last par., the 3-D DGSTR is compared with several 2-D denoising methods (each slice is performed separately including the K-SVD2 [11], BM3D3 [2], and NLM4 [30]) and 3-D Denoising methods (several slices are performed as a whole including 3-D K-SVD and VBM3D [31]). Based on Fig. 2, for a tradeoff, the parameters setting of 2-D K-SVD, 3-D K-SVD, and 3-D DGSTR are listed in Table I. The dictionaries used in 2-D K-SVD, 3-D K-SVD, and 3-D DGSTR are adaptively (Fig.4learned from the noisy image.); and
presenting the denoised version of the medical image for viewing by a user (Figs.3-5, the figures display the denoising result of the medical image ).
As to claim 2, LI teaches the dictionary comprises a noise vector incorporated into the dictionary per the noisy patches (page 3453 right col., last par., yki= Rki Y be the ith patch with a square area of n pixels extracted from the kth slice of noisy medical image Y, and ordered as n-dimension column vector, where Rki is a binary matrix extracting the patch from 3-D image.).
As to claim 3, LI teaches the noisy patches are denoised using an orthogonal matching pursuit (OMP) algorithm (page 3451, right col., II. GROUP SPARSE REPRESENTATION, II. GROUP SPARSE REPRESENTATION, Group Orthogonal Matching Pursuit (GOMP) [26] is a popular greedy algorithm for group sparse representation, which derives from OMP. Different from the OMP (selecting one atom each time), the GOMP seeks the group sparse coefficients by searching the most correlative group each time);
As to claim 5, LI teaches denoising the medical image by reconstructing the denoised set of patches includes using an orthogonal matching pursuit (OMP) algorithm to incorporate the dictionary into a sparse optimization of the medical image, separating noise in the medical image ( page 3454, left col., equation 15-18, the proposed 3-D medical image denoising model (14) can be rewritten as shown in equations 17 and 18. The denoised 3-D patch Xki is calculated as Dˆα. Then, each 3-D noisy patch extracted from Y is performed by the same way. At last, averaging the overlapping pixels yields the final denoised result. In (equation 18), the dictionary D plays a crucial role, which affects the denoised results. The DL-GSGR method is adopted to learn the dictionary D from the training set.).
As to claim 6, LI teaches when reconstructing the denoised set of patches into a denoised version of the medical image, overlapping patches are averaged (page 3457, left col., 3rd par., Then, the fused vectors are reshaped into patches with a square area of n pixels. At last, the final fused image is obtained by averaging the overlapping pixels)
As to claim 7, LI teaches the medical image is a magnetic resonance imaging (MRI) image (Figs.3-5).
As to claim 8, LI teaches the medical image is a computed tomography (CT) image ( see Fig.6).
As to claim 9, LI teaches A system for denoising a medical image, comprising: a memory; and a processor communicatively coupled to the memory; wherein the memory stores a set of instructions which, when executed by the processor, cause the processor to: obtain a medical image of a subject (inherent, page 3454-3455 ,Section B Experiment, LI specifically teaches a medical image fusion, it can be used for clinical diagnosis, computer-aided diagnosis (see section page 3455, right col., section V , first par.,). Li further teaches experimental result based on computer simulation that include of Gaussian noise simulation and Rician noise simulation, and image fusion see Fig.2, tables I-III); regarding the remaining limitation of claim 9, all the claim limitations are set forth and rejected as per discussion for claim 1.
Regarding claim 10, all the claim limitations are set forth and rejected as per discussion for claim 9 and 2.
Regarding claim 11, all the claim limitations are set forth and rejected as per discussion for claim 9 and 3.
Regarding claim 13, all the claim limitations are set forth and rejected as per discussion for claim 9 and 5.
Regarding claim 14, all the claim limitations are set forth and rejected as per discussion for claim 9 and 6.
Regarding claim 15, all the claim limitations are set forth and rejected as per discussion for claim 9 and 7.
Regarding claim 16, all the claim limitations are set forth and rejected as per discussion for claim 9 and 8.
As to claim 17, LI teaches using the orthogonal matching pursuit (OMP) algorithm comprises selecting a subset of the set of atoms that project the noisy patches into denoised representations (page 3451 right col., 1st and 2nd pars., Group Orthogonal Matching Pursuit (GOMP) [26] is a popular greedy algorithm for group sparse representation, which derives from OMP. Different from the OMP (selecting one atom each time), the GOMP seeks the group sparse coefficients by searching the most correlative group each time).
Regarding claim 18, all the claim limitations are set forth and rejected as per discussion for claim 17.
As to claim 19, LI teaches generating the set of atoms comprises extracting from the set of frequency representations of the noisy patches atoms based on a lasso minimization (page 3451, right col., 1st and 2nd pars., equation ).
Regarding claim 20, all the claim limitations are set forth and rejected as per discussion for claim 19.
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 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 of this title, 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.
6. Claim 4 and 12 are rejected under 35 U.S.C. 103(a) as being unpatentable over
LI, “Group-Sparse Representation With Dictionary Learning for Medical Image Denoising and Fusion”, in view of LIU et al., (hereafter LIU), “A least angle regression assessment algorithm based on joint dictionary for visible and near-infrared spectrum denoising” , Optik, pub. 4 May 2021.
As to claim 4, LI teaches “the noisy patches are denoised” as discussed in claim 1 above but fails to teach “wherein the noisy patches are denoised using a least-angle regression (LARS) algorithm”.
On the other hand, LIU teaches the noisy Abstract, page 2 2nd par., LIU specifically teaches a least angle regression assessment (LARA) algorithm for joint dictionary to denoise the mixed spectrum).
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate a well-known LARS algorithm taught by LIU into LI to denoise the medical images taught by Li.
The suggestion/motivation for doing so would have been allows user of LI to maximize a competitional efficient of denoising, especially for high-dimensional datasets where the number of features significantly exceeds the number of samples.
Regarding claim 12, all the claim limitations are set forth and rejected as per discussion for claim 9 and 4.
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.
Contact Information
Any inquiry concerning this communication or earlier communication from the examiner should be directed to Mekonen Bekele whose telephone number is (469) 295-9077.The examiner can normally be reached on Monday -Friday from 9:00AM to 6:50 PM Eastern Time.
If attempt to reach the examiner by telephone are unsuccessful, the examiner’s supervisor Eng, George can be reached on (571) 272-7495.The fax phone number for the organization where the application or proceeding is assigned is 571-237-8300. Information regarding the status of an application may be obtained from the patent Application Information Retrieval (PAIR) system. Status information for published application may be obtained from either Private PAIR or Public PAIR.
Status information for unpublished application is available through Privet PAIR only.
For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have question on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866.217-919 (tool-free)
/MEKONEN T BEKELE/Primary Examiner, Art Unit 2699