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 .
Claim Interpretation
In regards to claim 1, the terminology that has been added via amendment is left extremely unclear, as such, additional interpretation is required to make sense of some terms. The term, “pixel-level predicted annotation images” does not make sense. An image or a digital image by definition is a two-dimensional matrix/array of pixels. “Pixel-level” would mean on the level of an individual pixel. As such, a “pixel-level predicted annotation image” would be a two-dimensional array of pixels at the level of exactly one pixel. This doesn’t make logical sense as something that is made up of pixels cannot be at the same level as a singular pixel. As such, the term is being interpreted as, “a predicted annotation image comprised of pixels”. Another term in the amended language have this issue, such as “pixel-wise annotation images” which informed this interpretation as it further clarifies that it is merely an image comprised of pixels that correspond to other pixels. Another term is “generate fake images as pixel-level image data” as this similarly is making a matrix of pixels into a singular pixel, so this is being interpreted to have “fake images comprised of pixels”. If applicant takes issue with these interpretations, they may address them in a response.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-10 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claims contain subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventors, at the time the application was filed, had possession of the claimed invention. There are a variety of new matter elements added in the amended claims 1 and 7, “wherein the segmentation model is configured to generate pixel-level predicted annotation images corresponding to respective input images”; “inputting real images, each represented as a multi-dimensional pixel matrix”; “annotation images as pixel-wise annotation images comprising pixel values corresponding to respective pixels of the input images”; and for claim 1 specifically “computing by one or more processors, a loss function based on differences between pixel-level image data of the real images and pixel-level data of the fake images”; and for claim 7 specifically “computing by one or more processors, a loss function based on differences between pixel-level image data of the real images and pixel-level image data of the fake images”. Each of these additions to the claims are new matter. The specification recites the term pixel four times in paragraphs 14 and paragraph 28. Each recitation of pixels is related to the image generator being a GAN or Generative Adversarial model with pixel to pixel correspondence. For the loss specific aspects, it is recited five times in the specification in paragraphs 9, 15, 39, and 43, there is no recitation of the loss even using a function let alone using “pixel-level image data” or “pixel-level data”. As a matter of fact, the term pixel is recited more times in the newly amended claim language than anywhere else in the application. For the “multi-dimensional pixel matrix” has its own enablement issues outside of it being purely new matter, but the specification and drawings make no mention of anything beyond a standard two-dimensional pixel matrix or an image. Anything like, a third dimensional pixel matrix, fourth dimensional pixel matrix or so on is clearly not recited by the application at any point. The various disclosures of “pixel-level” or “pixel-wise” images given are similarly new matter as there is no disclosure in the specification that an image or a two-dimensional array of pixels can simply be just one pixel. As such, the dependent claims for claims 1 and 7 inherit these issues, and they are similarly rejected.
Claims 1-10 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, because the specification, while being enabling for “two-dimensional pixel matrix”, does not reasonably provide enablement for “multi-dimensional pixel matrix”. The specification does not enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the invention commensurate in scope with these claims. The breadth of the claims is fairly broad as the “multi-dimensional pixel matrix” would allow for any number of additional dimensions added to a matrix so long as that number is greater than one. The nature of the invention seems to be limited to exclusively two-dimensional pixel matrices or images rather than allowing for three-dimensional pixel matrices or even higher dimensional matrices. The state of the prior art does have at least voxels which would be roughly analogous to a three-dimensional pixel matrix, while higher order dimensions seem beyond the scope of the prior art. A person of ordinary skill in the art would likely be entirely unfamiliar with anything beyond a three-dimensional pixel matrix, and they would similarly be unable to apply the methods disclosed onto a voxel without serious experimentation. The level of predictability in the art is difficult to surmise as any higher order pixel matrix would be difficult to find. Voxels would be the only seemingly analogous form in the art, and they likely have a decent amount of art that would cover something similar to this. The inventor has essentially provided no direction at all in regards to anything beyond a two-dimensional pixel matrix. There are no working examples of anything greater than a three-dimensional pixel matrix, and a three-dimensional pixel matrix is analogous to a voxel. There would need to be a massive amount of experimentation in a frankly unknown field as what would constitute a fourth-dimensional pixel matrix or beyond is practically a complete unknown, and as such, to use or make the invention, would require a person of ordinary skill in the art to essentially have to invent something completely new just to make the invention, let alone to use it. As all dependent claims rely on claims with this terminology, they are similarly rejected.
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 1-10 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.
Where applicant acts as his or her own lexicographer to specifically define a term of a claim contrary to its ordinary meaning, the written description must clearly redefine the claim term and set forth the uncommon definition so as to put one reasonably skilled in the art on notice that the applicant intended to so redefine that claim term. Process Control Corp. v. HydReclaim Corp., 190 F.3d 1350, 1357, 52 USPQ2d 1029, 1033 (Fed. Cir. 1999). The terms “pixel-wise annotation images”; “fake images as pixel-level image data”; and “pixel-level predicted annotation images” in claims 1 and 7 is used by the claim to mean “an image comprised of pixels,” while the accepted meaning is “an image made at the level of a single pixel.” The term is indefinite because the specification does not clearly redefine the term.
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 (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 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.
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 1-10 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Rajapakse et al. (US 20250022137 A1), hereinafter referred to as Rajapakse.
In regards to claim 1, Rajapakse discloses a method for training a segmentation model, comprising: using first training images to train a segmentation model, wherein the segmentation model is configured to generate pixel-level predicted annotation images corresponding to respective input images (Paragraph 105, Describes that the two segmentation models, the U-Nets, use images input for training purposes where the images produced are made of pixels with specific pixel data produced); using second training images to train an image generator; inputting real images, each represented as a multi-dimensional pixel matrix (Abstract, Discloses that the image generators or GANs are trained using image pairs and all digital images are a 2-dimensional pixel matrix inherently), into the segmentation model to generate predicted annotation images as pixel-wise annotation images comprising pixel values corresponding to respective pixels of the input images (Paragraphs 90, 116-117, and paragraphs 97-99, Describes that the datasets contain real MR scans, pre-annotated segmentation masks, and synthetic CT scans. Since the MR scans are real, this would read upon the claim, and it further discloses that the annotations can come from machine learning. Paragraphs 116 and 117 discloses further labelling of medical images. With paragraphs 97-99 explicitly mentioning pixel-level loss being a factor); inputting the predicted annotation images into the image generator to generate fake images as pixel-level image data corresponding to the predicted annotation images (Abstract and paragraphs 97-99, Describes that the GAN generates synthetic images from the image pairs which would include the annotated images with the later paragraphs establishing that they are done using pixel-level loss functions); and computing, by one or more processors, a loss function based on differences between pixel-level image data of the real images and pixel-level data of the fake images (Paragraph 85 and Paragraphs 97-99, The disclosed adversarial loss in paragraph 85 would cover that for the image generator with paragraph 98 corroborating that assessment and paragraph 97 details adversarial training is used for the U-nets as well. Paragraph 99 details the use of a function that checks the pixel level loss), and updating parameters of the segmentation model and the image generator based on the computed loss (Paragraph 85 and Paragraphs 97-98, The disclosed adversarial loss in paragraph 85 would cover that for the image generator with paragraph 98 corroborating that assessment and paragraph 97 details adversarial training is used for the U-nets as well).
In regards to claim 2, Rajapakse discloses wherein the first training images are labeled images (Paragraphs 75 and 96, Describes that the datasets contain MR scans, pre-annotated tissue masks, and CT scans. The pre-annotated images would be labelled).
In regards to claim 3, Rajapakse discloses wherein the second training images are labeled images (Paragraphs 75 and 96, Describes that the datasets contain MR scans, pre-annotated tissue masks, and CT scans. The pre annotated images would be labeled.).
In regards to claim 4, Rajapakse discloses wherein the real images comprise labeled images and unlabeled images (Paragraphs 75 and 96, Describes that the datasets contain MR scans, pre-annotated tissue masks, and CT scans. The pre-annotated images are real as they are used to train the system to make fake MR images and the MR scans are real images which were used to train it to generate fake MR images. Paragraph 96 also states that real data is used which would include real versions of the labelled and unlabeled data).
In regards to claim 5, Rajapakse discloses wherein the segmentation model is based on a Visual Geometry Group (VGG) U-net model (Paragraph 105, Describes that the two segmentation models, the U-Nets, use images input for training purposes.).
In regards to claim 6, Rajapakse discloses wherein the image generator is based on a Generative Adversarial Network (GAN) model with pixel to pixel correspondence (Paragraph 85, The paragraph discloses the usage of a GAN where the training enforces pixel-level loss).
In regards to claim 7, it is similar to claim 1, and it is similarly rejected. The aspects of processors and computer storage media are adequately covered by paragraph 123 of Rajapakse.
In regards to claim 8, it is similar to claim 2, and it is similarly rejected.
In regards to claim 9, it is similar to claim 3, and it is similarly rejected.
In regards to claim 10, it is similar to claim 4, and it is similarly rejected.
Response to Amendment
The amendment, entered 2/11/2026, overcomes the specification objections and the 101 rejections. However, the amended claim language introduces a number of new matter issues along with issues around claim interpretation and 112(b).
Response to Arguments
Applicant's arguments filed 2/11/2026 have been fully considered but they are not persuasive.
In regards to the 102 arguments, applicant alleges that the term “predicted annotation images” is actually supposed to mean, “segmentation masks”. Now, the BRI of “predicted annotation images” is significantly broader than just one kind of mask with it being merely an image that is annotated. Further, it would not be reasonable in this case to limit the interpretation to this as the specification, claims, and drawings make no mention of “segmentation masks” or even “masks” in general. Further, the reference that allegedly does not teach the use of “segmentation masks” being used in an analogous manner explicitly mentions the usage of segmentation masks even in the portions cited by the applicant’s own argument, and it uses and references segmentation masks when applicant’s own specification, drawings, and claims make no mention of the term at all. So, even if applicant’s interpretation was applicable, arguing that the reference does not teach their usage is contrary to the reference itself as cited by the applicant.
Argument further alleges that applicant’s claim 1 discloses a closed loop structure rather than a linear structure. However, the language of claim 1 does not demonstrate any form of “closed loop” as it addresses a set of linear steps that are taken to update the model with no mention of any “closed loop”.
Lastly, argument alleges that Rajapakse does not compare a real image and a fake image as the fake image is constructed from a segmentation mask. As previously noted, the specification, claims, and drawings make no reference to a “segmentation mask” or any kind of “mask” at all. Secondly, reading the abstract of Rajapakse, “GAN to generate a synthetic mask for the synthetic MR image based on relationships between features of the synthetic MR image and segmentation mask of each data group.” So, the synthetic images of Rajapakse do use the segmentation masks explicitly. Even if applicant’s segmentation mask argument was applicable, Rajapakse would still cover the claimed aspects. As applicant’s argument about segmentation masks is not applicable, a “synthetic image” and a “fake image” are essentially synonyms of each other with the specification and claims requiring nothing more than the fake image be generated from a predicted annotation image. As such, Rajapakse’s fake image and their calculated loss are more than applicable in this case.
For these reasons, the arguments are not persuasive.
Conclusion
THIS ACTION IS MADE FINAL. 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 CONOR AIDAN O'MALLEY whose telephone number is (571)272-0226. The examiner can normally be reached Monday - Friday 9:00 am. - 5:00 pm. EST.
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, Andrew Moyer can be reached at 5722729523. 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.
CONOR AIDAN. O'MALLEY
Examiner
Art Unit 2675
/CONOR A O'MALLEY/Examiner, Art Unit 2675
/ANDREW M MOYER/Supervisory Patent Examiner, Art Unit 2675
Prosecution Insights