DETAILED ACTION
Accordingly, claim 46 withdrawn from consideration as being directed to a non-elected invention. See 37 CFR 1.142(b) and MPEP § 821.03.
Applicant’s arguments, see remarks, pages 7,8 filed 6/30/2026, with respect to double patenting have been fully considered and are persuasive. The double patenting rejection of claims 1-20 has been withdrawn.
Claim(s) 1,5,6,7,10,11,12,18,20 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1):
Claim(s) 2 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of ZHU et al. (CN 111275128 A) with SEARCH machine translation:
Claim(s) 3 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Xu et al. (US 2018/0374245 A1):
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Xu et al. (US 2018/0374245 A1) as applied in claim 3 further in view of ALLMENDINGER et al. (US 2019/0130571 A1):
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of THOMAS (DE 10211485 A1) with SEARCH machine translation:
Claim(s) 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of DEY et al. (US 2011/0275933 A1):
Claim(s) 13,14,15 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Bryant et al. (US 5,751,910):
Claim(s) 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Wei et al. (Real-time tumor localization with single x-ray projection at arbitrary gantry angles using a convolutional neural network (CNN)):
Claim(s) 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Holt (US 12,579,718 B1):
Claim(s) 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of CHU et al. (US 2021/0073678 A1):
Response to Amendment
The amendment was received 6/30/2026. Claims 1-20 and 46 are pending:
PNG
media_image1.png
768
177
media_image1.png
Greyscale
Election/Restrictions
Newly submitted claim 46 directed to an invention that is independent or distinct from the invention originally claimed for the following reasons:
Restriction to one of the following inventions is required under 35 U.S.C. 121:
I. Claims 1-20, drawn to training; learning, classified in G06T 2207/20081:
PNG
media_image2.png
161
1058
media_image2.png
Greyscale
II. Claim 46, drawn to Categorising the entire scene, e.g., birthday party or wedding scene, classified in G06V 20/35:
PNG
media_image3.png
129
1066
media_image3.png
Greyscale
The inventions are independent or distinct, each from the other because:
Inventions I and II are related as subcombinations disclosed as usable together in a single combination. The subcombinations are distinct if they do not overlap in scope and are not obvious variants, and if it is shown that at least one subcombination is separately usable. In the instant case, subcombination I has separate utility such as special programming. See MPEP § 806.05(d).
The examiner has required restriction between subcombinations usable together. Where applicant elects a subcombination and claims thereto are subsequently found allowable, any claim(s) depending from or otherwise requiring all the limitations of the allowable subcombination will be examined for patentability in accordance with 37 CFR 1.104. See MPEP § 821.04(a). Applicant is advised that if any claim presented in a divisional application is anticipated by, or includes all the limitations of, a claim that is allowable in the present application, such claim may be subject to provisional statutory and/or nonstatutory double patenting rejections over the claims of the instant application.
Since applicant has received an action on the merits for the originally presented invention, this invention has been constructively elected by original presentation for prosecution on the merits.
Accordingly, claim 46 withdrawn from consideration as being directed to a non-elected invention. See 37 CFR 1.142(b) and MPEP § 821.03.
To preserve a right to petition, the reply to this action must distinctly and specifically point out supposed errors in the restriction requirement. Otherwise, the election shall be treated as a final election without traverse. Traversal must be timely. Failure to timely traverse the requirement will result in the loss of right to petition under 37 CFR 1.144. If claims are subsequently added, applicant must indicate which of the subsequently added claims are readable upon the elected invention.
Should applicant traverse on the ground that the inventions are not patentably distinct, applicant should submit evidence or identify such evidence now of record showing the inventions to be obvious variants or clearly admit on the record that this is the case. In either instance, if the examiner finds one of the inventions unpatentable over the prior art, the evidence or admission may be used in a rejection under 35 U.S.C. 103 or pre-AIA 35 U.S.C. 103(a) of the other invention.
35 USC § 101 – Positive Statement
Claims 1-20 is directed to reflecting an improvement1 in a technical field (Medicine/Medical) of imaging2 in view of applicant’s disclosure [0011]. Thus claims 1-20 statutory under 35 USC 101.
Response to Arguments
Double Patenting
Applicant’s arguments, see remarks, pages 7,8 filed 6/30/2026, with respect to double patenting have been fully considered and are persuasive. The double patenting rejection of claims 1-20 has been withdrawn.
Claim Rejections - 35 U.S. C § 102
Applicant's arguments filed 6/30/2026 have been fully considered but they are not persuasive:
A. Bengtsson is Directed to a Fundamentally Different Technical Field and
Application.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 9, 2nd para: “microscopy” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 9, 3rd para: “image quality”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
B. Bengtsson Does Not Disclose the Claimed "Improvement Selection."
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 9, 5th para: “user-facing improvement selection mechanism”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 9, 7th para: “a user-received selection”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 9, last para: “a user-selectable improvement”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 10, 1st para: “a user selection that drives”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
C. Bengtsson's "Partitioning" is Fundamentally Different from the Claimed
Partitioning.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 10, 3rd para: “partitions projections of the same sample”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 10, 4th para: “same object with decorrelated artifacts”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 10, 4th para: “decorrelate noise and artifacts across subsets so that neural network training can learn to distinguish signal from noise”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicant’s arguments, see remarks, pages 10,11:
Thus, claim 1 is distinct from Bengtsson because Bengtsson's subsets are inherently different in content (different body parts), not merely different in noise or artifacts. Amended claim 1 expressly requires that the subsets differ from one another "only in noise, aliasing artifact, or other artifact." This feature is fundamentally different from Bengtsson's approach of creating subsets with different anatomical content.
, filed 6/30/2026, with respect to the rejection(s) of claim(s) 1 under 35 USC 102 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of 35 USC 103:
Claim(s) 1,5,6,7,10,11,12,18,20 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1), wherein HETEREN teaches a “simulated” projection via “unique training” and different noise:
[0053] It is noted that the modifications to the content in a specific digital volume performed in step 603 differs from the augmentation of projection images described above in conjunction with full data set 521. In step 603, a unique training target volume is generated from an existing target volume, whereas the above-described augmentation of projection images generates modified projection images that are not directly associated with any particular target volume, either real or simulated.
[0090] Generator function 1250 can be a neural net or other suitable machine learning model that is configured to generate new data instances based on a particular partial data reconstruction 524. For example, generator function 1250 can be configured to generate a synthesized reconstruction 1231. In addition, during GAN training phase 1230, the machine learning model of generator function 1250 is configured to improve its performance of generating synthesized reconstructions 1231 based on feedback 1233 from discriminator function 1232. For example, during GAN training phase 1230, generator function 1250 is configured to modify algorithm parameters 1251 so that discriminator function 1232 fails to detect reconstruction artifacts and/or other image artifacts in a particular synthesized reconstruction 1231 of a training target volume of interest. More specifically, through an iterative process included in GAN training phase 1230, algorithm parameters 1251 are modified. In this way, generator function 1250 can learn to generate synthesized reconstructions 1231 that appear to discriminator function 1232 to be free of reconstruction artifacts and/or other image artifacts. Generator function 1250 can then generate another synthesized reconstruction 1231 using the newly modified values for algorithm parameters 1251.
D. Bengtsson Does Not Disclose X-Ray Microscopy Imaging or CT Reconstruction from Partitioned Subsets.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 11, 2nd para: “X-ray microscopy imaging”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 11, 3rd para, 1st S: “generating CT training reconstructed volumes from partitioned projection subsets”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 11, 3rd para, 4th S: “tomographic3 reconstruction4”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). In contrast. claim 1 states alternatives:
“a computed tomography (CT)5 training reconstructed volume6” or in other words:
“a computed tomography (CT)…volume” or
“a…training…volume” or
“a…reconstructed volume” or
a reconstructed, training, CT volume or
a reconstructed, CT, training volume or
a training, CT, reconstructed volume or
a training and CT and reconstructed volume.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 11, 3rd para, last S: “raw”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 11, 4th para:
In sum, Bengtsson fails to disclose at least (1) the improvement selection related to throughput, de-noising, or artifact reduction; (2) the partitioning of projections into mutually exclusive subsets differing only in noise/artifacts; and (3) the generation of CT training reconstructed volumes from such partitioned subsets. Thus, Bengtsson cannot anticipate amended claim 1 or any claim depending therefrom. These distinctions apply equally to all claims rejected under 35 U.S.C. § 102, including claims 1, 5-7, 10-12, 18, and 20. Applicant respectfully requests withdrawal of the§ 102 rejections.
The examiner respectfully disagrees since BENGTSSON teaches claim 1’s:
--receiving7 (or knowledgably acquiring) an improvement selection89 (“to be trained on different anatomical regions, which improves overall learning10 of the CNN models” [0102] last S: likewise/similarly in view of the claimed “receiving an improvement selection”) related to a desired improvement being one of an improved throughput, de-noising or artifact reduction (or likewise “as input to pre-processing subsystem 210 of the CNN architecture… Image intensity values of all images may be truncated to a specified range (e.g., -1000 to 3000 Hounsfield Unit) to remove noise and possible artifacts” [0079] 1st & 3rd Ss);--.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 11, 4th para: “mutually exclusive subsets”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 11, 4th para: “from such partitioned subsets”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Claim Rejections - 35 U.S. C § 103
Claim(s) 1,5,6,7,10,11,12,18,20 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1):
Claim(s) 2 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of ZHU et al. (CN 111275128 A) with SEARCH machine translation:
Claim(s) 3 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Xu et al. (US 2018/0374245 A1):
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Xu et al. (US 2018/0374245 A1) as applied in claim 3 further in view of ALLMENDINGER et al. (US 2019/0130571 A1):
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of THOMAS (DE 10211485 A1) with SEARCH machine translation:
Claim(s) 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of DEY et al. (US 2011/0275933 A1):
Claim(s) 13,14,15 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Bryant et al. (US 5,751,910):
Claim(s) 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Wei et al. (Real-time tumor localization with single x-ray projection at arbitrary gantry angles using a convolutional neural network (CNN)):
Claim(s) 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Holt (US 12,579,718 B1):
Claim(s) 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of CHU et al. (US 2021/0073678 A1):
Accordingly, claim 46 withdrawn from consideration as being directed to a non-elected invention. See 37 CFR 1.142(b) and MPEP § 821.03.
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) 1,5,6,7,10,11,12,18,20 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1):
PNG
media_image4.png
754
548
media_image4.png
Greyscale
Re 1. (Currently Amended), BENGTSSON teaches A (“Downstream” [0105], last S: fig. 2C:260: Head, Chest, Abdomen VNet) method (fig. 7:700) comprising11 (likewise):
receiving training data of a sample (“received from one or more imaging systems 160” [0068]), wherein the training data is acquired by an X-ray imager (“to aim a motorized x-ray source” [0070]) using training parameters (or “learning parameters” [0065]);
receiving (or knowledgably acquiring) an improvement selection (“to be trained on different anatomical regions, which improves overall learning of the CNN models” [0102] last S: likewise/similarly in view of the claimed “receiving an improvement selection”) related to a desired improvement being one of an improved throughput, de-noising or artifact reduction (or likewise “as input to pre-processing subsystem 210 of the CNN architecture… Image intensity values of all images may be truncated to a specified range (e.g., -1000 to 3000 Hounsfield Unit) to remove noise and possible artifacts” [0079] 1st & 3rd Ss);
partitioning the training data into a plurality of (“different” [0102] last S) training subsets (or likewise “narrow…the imaging space12…for…trained on”, [0102] last S, regional parts: fig. 2C:260: “Head VNet”; “Chest VNet”) using the improvement (CNN) selection (via “Downstream processing may…use… select CNN models” [0102], penult S) ,13 each of the plurality14 (or likewise any one box-rectangle in fig. 2C:220) of training subsets including unique projections from the training data (or likewise “projections… include one or more training input elements 115a-b” [0061] 2nd & [0062] 1st S) ,15 such that no projection is present in more than one of the plurality of training subsets ,16 and the plurality of training subsets differing from one another only in noise, aliasing artifact, or other artifacts;
generating a computed tomography (CT) training reconstructed volume (“a three-dimensional image of the subject” [0070] 7th S) for at least two of the plurality of
training subsets;
training a neural network using each of the CT training reconstructed volumes (“for training the CNN models” [0071] last S: fig. 2A:215:”UNet”);
receiving additional imaging data (“one or more PET scans, CT scans, MRI scans, or any combinations thereof within a set of input image elements 115a-n.” [0059] 3rd S: fig. 2A:205: “Input Image Elements”), wherein i) the additional imaging data is of the sample acquired using additional parameters (or “one or more parameters” [0065] 5th S) different than the training parameters, or ii) the additional imaging data is of an additional sample acquired using the training parameters; and
generating a CT reconstructed volume using the additional imaging data and the trained neural network (“to generate a final masked image 235” [0083] 2nd S).
BENGTSSON does not teach the difference of claim 1 of:
( each of the plurality of training subsets including )17 unique (projections from the training data) ,18 such that no projection is present in more than one of (the plurality of training subsets) ,19 (and the plurality of training subsets) differing from one another only in noise, aliasing artifact, or other artifacts.
VAN HETEREN teach the difference of claim 1 of:
( each of the plurality of training subsets including )20 unique (projections from the training data) (or likewise “a unique21 training target volume...projection image simulation process” [0053] 2nd S & [0054] 1st S: fig. 6:604) ,22 such that no projection is present in more than one of (the plurality of training subsets) ,23 (and the plurality of training subsets) differing from one another only24 in noise, aliasing artifact, or other artifacts (or likewise “reconstruction artifacts and/or25 other26 image artifacts in a particular synthesized reconstruction 1231 of a training target volume of interest” [0090]: fig. 12:1231: “Synthesized Reconstruction”).
Since BENGTSSON teaches reconstructing “images” with “noisier” problems and known “combination” fix thereof with the possibility to “remove noise” via [0068][0069] [0079]:
[0068] At least part of the training input image elements 115a-d, the validation input image elements 115e-g and/or the unlabeled input image elements 115h-n may include or may have been derived from data collected using and received from one or more imaging systems 160. The imaging system 160 can include a system configured to collect image data (e.g., PET scans, CT scans, MRI scans, or any combinations thereof). The imaging system 160 may include a PET scanner and optionally a CT scanner and/or an MRI scanner. The PET scanner may be configured to detect photons (subatomic particles) emitted by a radionuclide in the organ or tissue being examined. The radionuclides used in PET scans may be made by attaching a radioactive atom to a chemical substance that is used naturally by the particular organ or tissue during its metabolic process. For example, in PET scans of the brain, a radioactive atom (e.g., radioactive fluorine such as .sup.18F) may be attached to glucose (blood sugar) to create FDG, because the brain uses glucose for its metabolism. Other radioactive tracers and/or substances may be used for image scanning, depending on the purpose of the scan. For example, if blood flow and perfusion of an organ or tissue is of interest, the radionuclide may be a type of radioactive oxygen, carbon, nitrogen, or gallium; if infectious diseases are of interest, a radioactive atom may be attached to sorbitol (e.g., fluorodeoxysorbitol (FDS)); and if oncology is of interest, a radioactive atom may be attached to misonidazole (e.g., fluoromisonidazole (FMISO)). The raw data collected by the PET scanner are a list of 'coincidence events' representing near-simultaneous detection (typically, within a window of 6 to 12 nanoseconds of each other) of annihilation photons by a pair of detectors. Each coincidence event represents a line in space connecting the two detectors along which the positron emission occurred (i.e., the line of response (LOR)). Coincidence events can be grouped into projection images, called sinograms. The sinograms are used in computer analysis to reconstruct two-dimensional images and three-dimensional images of metabolic processes in organ or tissue being examined. The two-dimensional PET images and/or the three-dimensional PET images may be included within the set of input image elements 115a-n. However, the sinograms collected in PET scanning is much poorer quality with respect to anatomical structures than CT or MRI scans, which can result in noisier images.
[0069] To overcome the deficiencies of PET scanning with respect to anatomical structures, PET scans are increasingly read alongside CT or MRI scans, with the combination (called "co registration") giving both detailed anatomic and metabolic information (i.e., what the structure is, and what it is doing biochemically). Because PET imaging is most useful in combination with separate anatomical imaging, such as CT or MRI, PET scanners are available with integrated high-end multi -detector-row CT or MRI scanners. The two scans can be performed in immediate sequence during the same session, with a particular subject not changing position between the two types of scans. This allows the two sets of images to be more precisely registered, so that areas of abnormality on the PET imaging can be more perfectly correlated with anatomy on the CT or MRI images.
[0079] The input image elements 205 are provided as input to pre-processing subsystem 210 of the CNN architecture, which generates standardized image data across the input image elements 205. Pre-processing may include selecting subsets of images or input image elements 205 for slices (e.g., coronal, axial, and sagittal slices) or regions of the body and performing geometric re-sampling (e.g., interpolating) of the subsets of images input image elements 205 in terms of uniform pixel spacing (e.g., 1.0 mm) and slice thickness (e.g., 2 mm). Image intensity values of all images may be truncated to a specified range (e.g., -1000 to 3000 Hounsfield Unit) to remove noise and possible artifacts. The standardization of the spacing, slice thickness, and units ensures that each pixel has a consistent area and each voxel has a consistent volume across all images of the input image elements 205. The output from the pre-processing subsystem 210 is subsets of standardized images for slices (e.g., coronal, axial, and sagittal slices) or regions of the body.
one of skill in the art would or could have done is refer to others regarding the known fix and to others in the event that possible artifacts show up in the images of CT,MRI,PET or combination thereof and thus make BENGTSSON’s be as VAN HETEREN’s seeing in the change good via VAN HETEREN [0090]: “free of reconstruction artifacts”:
[0090] Generator function 1250 can be a neural net or other suitable machine learning model that is configured to generate new data instances based on a particular partial data reconstruction 524. For example, generator function 1250 can be configured to generate a synthesized reconstruction 1231. In addition, during GAN training phase 1230, the machine learning model of generator function 1250 is configured to improve its performance of generating synthesized reconstructions 1231 based on feedback 1233 from discriminator function 1232. For example, during GAN training phase 1230, generator function 1250 is configured to modify algorithm parameters 1251 so that discriminator function 1232 fails to detect reconstruction artifacts and/or other image artifacts in a particular synthesized reconstruction 1231 of a training target volume of interest. More specifically, through an iterative process included in GAN training phase 1230, algorithm parameters 1251 are modified. In this way, generator function 1250 can learn to generate synthesized reconstructions 1231 that appear to discriminator function 1232 to be free of reconstruction artifacts and/or other image artifacts. Generator function 1250 can then generate another synthesized reconstruction 1231 using the newly modified values for algorithm parameters 1251.
via explicit, creative, routine, inferential Supreme court steps A,B,C:
A) Create a Dara Pre-Processing noise & artifacts removal program based on BENGTSSON’s fig. 2A and [0079]:
PNG
media_image5.png
625
987
media_image5.png
Greyscale
[0079] The input image elements 205 are provided as input to pre-processing subsystem 210 of the CNN architecture, which generates standardized image data across the input image elements 205. Pre-processing may include selecting subsets of images or input image elements 205 for slices (e.g., coronal, axial, and sagittal slices) or regions of the body and performing geometric re-sampling (e.g., interpolating) of the subsets of images input image elements 205 in terms of uniform pixel spacing (e.g., 1.0 mm) and slice thickness (e.g., 2 mm). Image intensity values of all images may be truncated to a specified range (e.g., -1000 to 3000 Hounsfield Unit) to remove noise and possible artifacts. The standardization of the spacing, slice thickness, and units ensures that each pixel has a consistent area and each voxel has a consistent volume across all images of the input image elements 205. The output from the pre-processing subsystem 210 is subsets of standardized images for slices (e.g., coronal, axial, and sagittal slices) or regions of the body.
A1) create the program using VAN HETEREN’s image generator of fig 13:
PNG
media_image6.png
1290
933
media_image6.png
Greyscale
B) input the noise removal program of VAN HETEREN at BENGTSSON’s fig. 2A:210
PNG
media_image7.png
1004
989
media_image7.png
Greyscale
C) see what happens (I foresee: generator function 1250 (i.e., BENGTSSON’s fig. 2A:210) can learn to generate synthesized reconstructions 1231 that appear to discriminator function 1232 to be free of reconstruction artifacts and/or other image artifacts. Generator function 1250 can then generate another synthesized reconstruction 1231 using the newly modified values for algorithm parameters 1251.)
Re 5. (Original), BENGSTTON discloses The method of claim 1, wherein partitioning the training data into the plurality of training subsets using the improvement selection includes partitioning the training data into a number of training subsets (via “split a region or body depicted in the scans into multiple anatomical regions, for example, three regions including the head-neck, chest, and abdomen-pelvis” [0102]), wherein the number of training subsets is selected (via “Downstream processing” [0102] penult S) using the improvement selection.
Re 6. (Original), BENGSTTON discloses The method of claim 5, wherein the training data includes imaging data for a plurality of acquisitions, and wherein partitioning the training data into the number of training subsets using the improvement selection includes associating (“sequential” [0061] 3rd S) imaging data for sequential acquisitions (fig. 1:160: “Imaging System(s)”) of the plurality of acquisitions to alternate training subsets (or “different anatomical regions, which improves overall learning of the CNN models.” [0102] last S: fig. 5: “Head VNet”) of the plurality of training subsets.
Re 7. (Original), BENGTSSON discloses The method of claim 6, wherein sequential acquisitions of the plurality of acquisitions are acquired at different angles (or different angle configurations” [0061] 7th S) with respect to the sample.
Re 10. (Original), BENGTSSON discloses The method of claim 1, further comprising receiving a (split) region of interest (ROI) selection (or “ROI”-“tumor segmentation” [0048] last S: fig. 2A: “2D Segmentation Mask”), wherein partitioning the training data into the plurality of training subsets includes using the ROI selection (to segment tumors: fig. 2A:235: “Final Masked Image”).
Re 11. (Original), BENGSTTON discloses The method of claim 1, further comprising:
receiving category identification information (“identification”-“images” [0071] 8th S) associated with the sample; and
retrieving a pre-trained neural network using the category identification information,
wherein the pre-trained (or “separately trained” [0059] 4th S) neural network is trained on a different sample (or “each”-“sample”-“voxel” [0048] 4th S & [0088] last S: fig. 5: “Head VNet”: voxel-cube),
wherein training the neural network includes further training (via “Further…used for training” [0061] last S) the pre-trained neural network, and
wherein generating the CT reconstructed volume using the additional imaging data and the trained neural network includes using the further trained pre-trained neural network (“for image and lesion metabolism analysis.” [0046] last S).
Re 12. (Original), BENGSTTON discloses The method of claim 1, further comprising:
applying imaging corrections (via “one or more labels that identify a ‘correct’ interpretation of a presence and/or severity of a tumor” [0063] 1st S: fig. 2C:255: “Component detection and region labeling”) to the training data prior to generating the CT training reconstructed volumes; and
applying (fig. 2C: arrows) the imaging corrections to the additional imaging data prior to generating the CT reconstructed volume.
Re 18. (Original), BENGSTTON discloses The method of claim 1, further comprising:
receiving category identification information (or “identification”-“images” [0070] 8th S): fig. 1:115a-n: “Input Images”) associated with the (voxel) sample; and
storing (via fig. 1:125: “Parameters”) the trained neural network in associated with the category identification information.
Re 20. (Original), BENGSTTON discloses A system comprising:
a control system (or “data processing system” [0009]) including one or more processors; and
a memory (or “ a non-transitory computer readable storage medium” [0023]) having stored thereon machine readable instructions;
wherein the control system is coupled to the memory, and the method of claim 1 is implemented when the machine executable instructions in the memory are executed by at least one of the one or more processors of the control system.
Claim(s) 2 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of ZHU et al. (CN 111275128 A) with SEARCH machine translation:
PNG
media_image8.png
754
554
media_image8.png
Greyscale
Re 2. (Original), BENGSTTON teaches The method of claim 1, wherein the additional imaging data is of the sample acquired using the additional parameters, and wherein the additional parameters are selected (or “Learned”27 [0065] last S) to achieve a greater throughput than the training parameters.
BENGSTTON does not teach the difference28 of claim 2 of:
greater throughput29 than.
ZHU teaches being slow (similar to the problem faced by applicants) and the difference of claim 2 of:
greater throughput (“of operation”) than (“the complex structure”, pg. 9, last txt blk).
Since BENGSTTON teaches a neural net, one of skill in the art of nets can make BENGSSTON’s be as ZHU’s seeing in the change “greater throughput of operation”, ZHU, pg. 9, last txt blk.
Claim(s) 3 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Xu et al. (US 2018/0374245 A1):
PNG
media_image9.png
754
554
media_image9.png
Greyscale
Re 3. (Original), BENGSSTON teaches The method of claim 1,
wherein the training parameters are associated with a training number (via “one or more training input image elements 115a-d” [0061] 2nd to last S) of projections (“called sinograms” [0068] 3rd to last S),
wherein the additional parameters are associated with an additional number (via said “one or more training input image elements 115a-d” [0061] 2nd to last S) of (sinogram) projections that is smaller than the training number of projections; and
wherein the method further comprises determining the additional number (via said “one or more training input image elements 115a-d” [0061] 2nd to last S) of (sinogram) projections using the training number of projections and the improvement selection.
BENGSTTON does not teach the difference of claim 3 of:
(an additional number of projections)30 that is smaller than (the training number of projections).
Xu teach the difference of claim 3 of:
(an additional number of projections)31 (“a subset”32) that is smaller than (“2D CBCT projection space images (step 944)” [0083] 12th S) (the training number of projections).
Since BENGSTTON teaches projection sinograms, one of skill in the art of projections can make BENGSTTON’s be as Xu’s seeing the change “reducing artifacts in 2D CBCT artifact contaminated projection space images in a projection space approach”, Xu [0083] 1st S.
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Xu et al. (US 2018/0374245 A1) as applied in claim 3 further in view of ALLMENDINGER et al. (US 2019/0130571 A1):
PNG
media_image10.png
753
790
media_image10.png
Greyscale
Re 4. (Original), BENGTSSON of the combination of BENGTSSON,Xu teaches The method of claim 3, wherein the improvement selection is an improvement factor (or improves-overall-learning regions as a factor via “improves overall learning”-“on33 different anatomical regions”) , and wherein determining the additional number of projections includes dividing (splitting) the training number34 (instead BENGTSSON of the combination of BENGTSSON,Xu teaches dividing/splitting a region which is not the same or similar as the claimed “training number”) of projections by the improvement factor.
BENGTSSON of the combination of BENGTSSON,Xu does not teach the difference of claim 4 of:
the training number35.
ALLMENDINGER teach the difference of claim 4 of:
the training (divided-output-image) number36 (via “The number of the training output images preferably amounts to a value that corresponds37 to the time of the motion period divided by the average recording time.“ [0136] 2nd S).
Since BENGTSSON of the combination of BENGTSSON,Xu teaches training, one of skill in the art of training can make BENGTSSON’s of the combination of BENGTSSON,Xu be as ALLMENDINGER’s seeing the change “especially advantageous if a plurality of training input images is present for each section of the motion phases of the object, i.e. for each training output image.”, ALLMENDINGER [0136] last S.
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of THOMAS (DE 10211485 A1) with SEARCH machine translation:
PNG
media_image11.png
753
790
media_image11.png
Greyscale
Re 8. (Original), BENGTSSON teaches The method of claim 7, wherein each sequential acquisition is angularly offset from a previous acquisition by an angle (configuration/inclination) determined using golden ratio angle determination techniques.
BENGTSSON does not teach the difference of claim 8 of:
angularly offset …
golden ration angle determination techniques.
THOMAS teach the difference of claim 8 of:
angularly offset (or “angle”-“offset”, pg. 4, 1st txt blk)…
golden ration angle determination techniques (via “the golden ration”, pg. 4, 1st txt blk).
Since BENGTSSON teaches an angle, one of skill in the art of angles can make BENGTSSON’s be as THOMAS’ seeing in the change “improvements over existing systems”, THOMAS, pg. 3, 2nd txt blk.
Claim(s) 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of DEY et al. (US 2011/0275933 A1):
PNG
media_image12.png
753
790
media_image12.png
Greyscale
Re 9. (Original), BENGTSSON teaches The method of claim 6, wherein (“voxel” [0089] last S) groups of sequential acquisitions of the plurality of acquisitions are acquired at common (“configuration” [0061] 7th S/inclination) angles, and wherein each sequential acquisition of a group of sequential acquisitions is associated (via fig. 1: signal lines) with a respective one of the plurality of training subsets.
BENGTSSON does not teach the difference of claim 9 of:
common (angles).
DEY teach the difference of claim 9 of:
(“the registration process will omit the corresponding projection image from sate 5 and instead use only the” [0066] penult S) common (angles).
Since BENGTSSON teaches projection and an angle, one of skill in the art of projection angles can make BENGTSSON’s be as DEY’s seeing that in the change “distortions can be overcome by using projection images in the reference state corresponding to the available angles in the given state”, DEY [0066] 4th S.
Claim(s) 13,14,15 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Bryant et al. (US 5,751,910):
PNG
media_image13.png
753
790
media_image13.png
Greyscale
Re 13. (Original), BENGTSSON teaches The method of claim 1, further comprising:
determining that the trained neural network is insufficient;
updating the improvement selection when38 the trained neural network is determined to be insufficient;
repartitioning the training data into an updated plurality of training subsets using the updated improvement selection;
generating an updated CT training reconstructed volume for the at least two of the updated plurality of training subsets;
retraining the neural network using each of the updated CT training reconstructed volumes; and
generating an updated CT reconstructed volume using the additional imaging data and the retrained neural network.
BENGTSSON does not teach the difference of claim 13 of:
(the trained neural network)39 is insufficient.
Bryant teach the difference of claim 13 of:
(the trained neural network)40 is insufficient (“training continues with the crisp input and fuzzy output vectors pair”, c.12,ll.20-25: fig.12: “HAS NETWORK REACHED SUFFICIENT PERFORMANCE?”).
Since BENGTSSON teaches a neural network, one of skill in the art of neural networks can make BENGTSSON’s be as Bryant’s seeing the change in the network being sufficient.
Re 14. (Original), BENGTSSON of the combination of BENGTSSON,Bryant teaches The method of claim 13, wherein determining that the trained neural network is insufficient includes:
presenting
i) cost function value information;
ii) the CT reconstructed volume (via the combination of BENGTSSON,Bryant after being determined sufficient for reconstruction);
iii) a trial CT reconstructed volume generated using the trained neural network and the training data; or
iv) any combination of i-iii; and
receiving user input (or “the plurality of PET scans and CT or MRI scans for the subject into a data processing system comprising the convolutional neural network architecture” [0009]: fig. 1:180: “User Device”) indicative that the trained neural network is insufficient (via the combination of BENGTSSON,Bryant).
Re 15. (Original), BENGTSSON of the combination of BENGTSSON,Bryant teaches The method of claim 13, wherein determining that the trained neural network (via “train the CNN models” , BENGTSSON [0065] 1st S) is insufficient is performed automatically (via the “flowcharts of FIGS. 10-12”, Bryant, c. 11, ll. 35-40) using a machine learning classifier (130 “used to train the CNN models”, BENGTSSON [0065] 1st S).
Claim(s) 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Wei et al. (Real-time tumor localization with single x-ray projection at arbitrary gantry angles using a convolutional neural network (CNN)):
PNG
media_image14.png
753
790
media_image14.png
Greyscale
Re 16. (Original), BENGTSSON teaches The method of claim 1, further comprising applying an angle-dependent weighting mask (or “relevant features”-“mask 220” [0082] 1st S) to the training data.
BENGTSSON does not teach the difference of claim 16 of:
angle-dependent (weighting mask).
Wei teach the difference of claim 16 of:
angle-dependent (“region of interest (ROI) mask”, pg. 2, 2nd para, 3rd S) (weighting mask).
Since BENGTSSON teaches a region of interest *ROI), one of skill in the art of ROI’s can make BENGTSSON’s be as Wei’s seeing the change “improve the accuracy of tumor localization”, Wei, pg. 2, 2nd para, last S.
Claim(s) 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of Holt (US 12,579,718 B1):
PNG
media_image15.png
753
790
media_image15.png
Greyscale
Re 17. (Original), BENGTSTTON teaches The method of claim 1, further comprising truncating each (“intensity value” [0079] 3rd S) of the CT training reconstructed volumes in a Z direction prior to training the neural network.
BENGTSTTON does not teach the difference of claim 17 of:
in a Z direction.
Holt teach the difference of claim 17 of:
(“an axially truncated object that is roughly constant in Z (for example a human torso or a rocket motor)” c.7,ll.40-45) in a Z direction.
Since BENGTSTTON teaches directions (“In some instances, the PET scans, CT scans, or MRI scans of the set of input image elements 115a-n may be divided into one or more parts (e.g., slices of the body: transverse scans, coronal scans, and sagittal scans and/or anatomical portions or regions: head-neck, chest and abdomen-pelvis) by an image detector such that each of the classifier subsystems 1 lOa-n may process a respective part of the PET scans, SPECT scans, CT scans, or MRI scans for training and deployment.” [0059] last S), one of skill in the art of directions can make BENGTSTTON’s be as Holt’s seeing the change “iteratively improving a guess or estimate (or set of guesses or estimates) for the reconstruction until an acceptable image is found. For example, in computed tomography (CT), such as cone beam computed tomography (CBCT), an image is reconstructed from projections of an object. CBCT is an imaging technique consisting of x-ray computed tomography where the x-rays are divergent in two dimensions, forming a cone (or pyramid) shaped beam.” Holt, c.2,ll.55-65.
Claim(s) 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over BENGTSSON et al. (WO 2020/190821 A1) in view of IDS cited VAN HETEREN et al. (US 2020/0034999 A1) as applied in claims 1,5,6,7,10,11,12,18,20 further in view of CHU et al. (US 2021/0073678 A1):
PNG
media_image16.png
753
790
media_image16.png
Greyscale
Re 19. (Original), BENGSTTON teaches The method of claim 18, further comprising:
transmitting the trained neural network (resulting in “images or scans that are input to one or more classifier subsystems are received from the provider system 170” [0072] penult S) via a network interface, wherein transmitting the trained neural network includes transmitting the category identification information (via “transmit the images or scans (e.g., along with a subject identifier and one or more labels) to the CNN system 105” [0072] last S);
receiving a collaboratively trained neural network via the network interface, wherein the collaboratively trained neural network is based on the trained neural network and one or more additional trained neural networks (fig. 1: 105: “Deep Convolutional Neural Network System”) associated with the category identification information; and
storing (via fig. 1:125: “Parameters”) the collaboratively trained neural network as a pre-trained neural network.
BENGTSTTON does not teach the difference of claim 19 of:
a network interface…
collaboratively (trained neural network)41…42the network interface…
collaboratively (trained neural network)…
collaboratively (trained neural network)…as a pre-trained.
CHU teach the difference of claim 19 of:
a network interface (“for connection to available mass storage(s), video adapter(s), and I/O interface(s) available on the networks” [0170] last S) …
collaboratively (“trained together using vertically partitioned training data, said vertically partitioned training data including other data samples each including plural features, and different subsets of said plural features are accessible to different ones of the plurality of private machine learning models.” [0163] last S) (trained neural network)43…44the network interface (“1050 for connecting the computing system to communication networks 1055.” [0167] last S: fig.10)…
collaboratively (“train a machine learning model, M.sub.FED, such that, during the training, the data D.sub.i of any data owner F.sub.i is not exposed to other data owners F.sub.k, k≠i.” [0052] 5th S) (trained neural network)…
collaboratively (“training the model using all data which is freely shared” [0052] last S) (trained neural network)…as a pre-trained (“machine learning model 712, 724-1, . . . , 724-k associated with its private machine learning model” [0122] 3rd S).
Since BENGTSTTON teaches a neural network, one of skill in the art of neural networks can make BENGTSTTON’s be as CHU’s seeing in the change “neural network node weights can be adjusted based on the feedback in a manner that would have made the model generate a prediction that resulted in an improved result, as measured by the feedback (loss).” [0089] last S.
Conclusion
The prior art “nearest to the subject matter defined in the claims” (MPEP 707.05) made of record and not relied upon is considered pertinent to applicant's disclosure.
The following table lists several references that are relevant to the subject matter claimed and disclosed in this Application. The references are not relied on by the Examiner, but are provided to assist the Applicant in responding to this Office action.
Citation
Relevance
BAI et al. (US 2021/0059625 A1)
BAI teaches training a neural network with projection that includes unique data via [0019] and figure 6:
PNG
media_image17.png
940
843
media_image17.png
Greyscale
[0019] In one approach described herein, a neural network is trained at least with projection data (which, again, includes unique spectral information), reference spectral data (both data from a same scan and a same CT scanner), and system information. Through training, the neural network learns how to produce spectral data from the non-spectral projection data and the system information. For example, the neural network learns how to produce spectral data corresponding to Compton scatter for high-Z, low density and low-Z, high density objects and photoelectric absorption for high-Z, low density and low-Z, high density objects. Further training can be performed to fine tune the parameters of the neural network such that the neural network can process training non-spectral data and produce spectral data within a given error of training spectral data. Once trained, the trained neural network can process non-spectral projection data from a scan of a patient and system information and produce spectral data for the patient with the neural network. As further described herein, the training data can include non-spectral volumetric image data.
as the closest to “training subsets including unique projections” of claim 1.
Bradley (Artifacts and quantitative biases in neutron tomography introduced by systematic and random errors)
Bradley teaches “projections” (para of pages 7,8) and “unique angles” in desription of figure 3 and future-“neural network-based reconstruction” (pg 31, penult S):
PNG
media_image18.png
592
888
media_image18.png
Greyscale
It is also desirable that a slice can be reconstructed fast and with moderate computational resources. In neutron tomography sample rotations are often performed over 360 instead of 180 degrees, which averages the geometric blur over a full sample rotation, improving image quality. Furthermore, a 360 degree CT is preferred in the case of strongly attenuating materials as beam hardening effects are distributed over a full 360 degree scan with reduced intensity, rather than a semicircle. In this study we collected projections from 0 to 360 degrees, at unique sample angles avoiding angle pairs separated by 180 degrees and with the number of projections satisfying the Nyquist-Shannon criterion. Thus, figure 3a illustrates the ideal sampling case with a rotation of more than 180 degrees and with the optimum number of projections. Figure 3b represents a case of undersampling in two respects: missing wedges of angles and reduced number of projections. Figure 3c also represents a case of undersampling, as the number of projections is lower than that given by the Nyquist Shannon criterion. Throughout the paper we will therefore refer to a full sampled data set as one which includes a 360 degrees angular sampling range, seen in figure 3a. It can be noted that the angular steps calculated for an uneven number of projections avoids undersampling, and that for this case a separate recording of the 180 degree position is useful for determining the exact tilt angle and center of rotation.
as the closest to “training subsets including unique projections” of claim 1.
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 DENNIS ROSARIO whose telephone number is (571)272-7397. The examiner can normally be reached Monday-Friday, 9AM-5PM 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, Henok Shiferaw can be reached at 571-272-4637. 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.
/DENNIS ROSARIO/Examiner, Art Unit 2676
/Henok Shiferaw/Supervisory Patent Examiner, Art Unit 2676
1 [0011] Embodiments of the present disclosure include a method that comprises
receiving training data of a sample. The training data is acquired by an X-ray imager using training parameters. The method further comprises receiving an improvement selection, such as a selection to improve throughput or a selection to improve image quality at the same throughput. The method further comprises partitioning the training data into a plurality of training subsets using the improvement selection (e.g., the throughput or image quality selection). The method further comprises training a neural network using each of the training subsets. Training the neural network includes generating de-noised) imaging data for each of the training subsets using the neural network. Training the neural network further includes evaluating the neural network using the improved imaging data. The method further comprises receiving additional imaging data (e.g., additional imaging data of the same sample, such as a different region of interest, or additional imaging data of a subsequent sample in the scan series acquired under similar imaging conditions). The additional imaging data is acquired by the X- ray imager using additional parameters. The method further comprises generating improved additional imaging data using the additional imaging data and the trained neural network. The method further comprises generating a CT reconstructed volume using the improved additional imaging data.
2 imaging: Medicine/Medical. the use of computerized axial tomography, sonography, or other specialized techniques and instruments to obtain pictures of the interior of the body, especially those including soft tissues. (Dictionary.com)
3 tomographic: a word derived from tomograph, wherein tomograph is defined: a machine for making an x-ray of a selected plane of the body. (Dictionary.com)
4 reconstruction: the act of reconstructing, rebuilding, or reassembling, or the state of being reconstructed. (Dictionary.com)
5 computed tomography: computerized axial tomography. CT, wherein computerized axial tomography is defined: Tomography in which computer analysis of a series of cross-sectional x-ray images made along a single axis of a bodily structure or tissue is used to construct a three-dimensional image of that structure. The technique is used in diagnostic studies of internal bodily structures, as in the detection of tumors or brain aneurysms, wherein tomography is defined: Any of several radiologic techniques for making detailed three-dimensional images of a plane section of a solid object, such as the body, while blurring out the images of other planes. (Dictionary.com)
6 volume: The amount of space occupied by a three-dimensional object or region of space. Volumes are expressed in cubic units. (Dictionary.com)
7 BROAD CLAIM LANGUAGE: -ing (of “receiving”): a suffix of nouns formed from verbs (receive), expressing the action of the verb (“receive”) or its result, product, material, etc. (the art of building; a new building; cotton wadding ), wherein etc. is defined: and others; and so forth; and so on (used to indicate that more of the same sort or class might have been mentioned, but for brevity have been omitted), wherein and is defined: (used to connect [Markush] alternatives). (Dictionary.com)
8 selection: an act or instance of selecting or the state of being selected; choice. (Dictionary.com)
9 noun phrase
10 learn: to acquire knowledge of or skill in by study, instruction, or experience. (Dictionary.com)
11 BROAD CLAIM LANGUAGE: -ing (of “comprising”): a suffix of nouns formed from verbs, expressing the action of the verb or its result, product, material, etc. (the art of building; a new building; cotton wadding ), wherein etc. is defined: and others; and so forth; and so on (used to indicate that more of the same sort or class might have been mentioned, but for brevity have been omitted), wherein so is defined: likewise or correspondingly; also; too. (Dictionary.com)
12 space: Mathematics A mathematical object, typically a set of sets, that is usually structured to define a range across which variables or other objects (such as a coordinate system) can be defined. (Dictionary.com)
13 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“a plurality of training subsets”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
14 plurality: state or fact of being plural, wherein plural is defined: being one of such a plurality. (Dictionary.com)
15 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“unique projections from the training data”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
16 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“a plurality of training subsets”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
17 (italics) represent claim limitations already taught
18 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“unique projections from the training data”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
19 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“a plurality of training subsets”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
20 (italics) represent claim limitations already taught
21 unique: existing as the only one or as the sole example; single; solitary in type or characteristics. (Dictionary.com)
22 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“unique projections from the training data”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
23 comma: the sign (,), a mark of punctuation used for indicating a division in a sentence, as in setting off a word, phrase, or clause [“a plurality of training subsets”], especially when such a division is accompanied by a slight pause or is to be noted in order to give order to the sequential elements of the sentence. (Dictionary.com)
24 only: without others or anything further; alone; solely; exclusively. (Dictionary.com)
25 or: (used to connect words, phrases, or clauses representing alternatives), wherein alternative is defined: a choice limited to one of two or more possibilities, as of things, propositions, or courses of action, the selection of which precludes any other possibility, wherein preclude is defined: to exclude or debar from something.
26 other: different or distinct from the one or ones already mentioned or implied. (Dictionary.com)
27 learn: to acquire knowledge of or skill in by study, instruction, or experience, wherein acquire is defined: to come into possession or ownership of; get as one's own, wherein possession is defined: the state of being possessed, wherein posses is defined: to seize or take, wherein take is defined: to pick from a number; select. (Dictionary.com)
28 THE CLAIMED INVENTION AS A WHOLE:The disclosed problem is:
[0004] While X-ray Microscopy Imaging provides best-in-class spatial resolution, it
suffers from relatively low throughput. This low throughput is especially problematic when using analytical image reconstruction algorithms. For example, to generate a reconstructed volume of a sample with suitably useful quality (e.g., low noise and low aliasing) using conventional techniques, it is necessary to acquire at least a minimum number of projections, each from various angular locations around the sample. As the number of required projections increase, the time required to acquire all of the imaging data increases
The disclosed solution is:
[0039] Certain aspects and features of the present disclosure provide a useful workflow
that enables a user with minimal or no knowledge of machine learning or neural network optimization to simply select a desired amount of improvement (e.g., throughput gain) and then train a neural network to achieve that desired improvement (e.g., that desired amount of gain). The workflow also permits users to more quickly identify useful throughput gain amounts through feedback and evaluation, which can further reduce overall time to acquire useful imaging data, reconstructed volumes, trained neural networks, and other data. Additionally, certain aspects and features of the present disclosure enable image quality improvements even when no throughput gain is desired. For example, a neural network trained using an improvement factor at or above 2 can also be applied on imaging data acquired using non- reduced throughput (e.g., throughput with a 1X gain) or sub-reduced throughput (e.g., throughput with a gain between 1X and the improvement factor of the neural network) to provide image quality improvements without or with, respectively, accompanying throughput improvements. For example, for a neural network trained on training data acquired at a throughput associated with 1600 projections and using an improvement factor of 2, that neural network can be applied to imaging data also acquired at a throughput associated with 1600 projections (instead of 800 projections) to improve the image quality of the resultant reconstructed volume. In another example, that same neural network can be applied to imaging data acquired at a throughput associated with fewer than 1600 projections to improve the image quality and overall throughput of the resultant reconstructed volume. While a neural network may be trained using a certain improvement factor, that neural network can be applied to any desired imaging data. For example, if a 4x improvement factor is selected based on training data acquired at a throughput associated with 1600 projections, the trained neural network can be applied on any suitable imaging data, such as imaging data acquired at 400 projections, 390projections, 410 projections, or any other suitable number of projections. In some cases, the trained neural network can be applied to imaging data having a number of projections that is equal to or approximately (e.g., within 1%, 2%, 3%, 4%, 5%, 6%, 7%,8%, 9%, or 10% of) the number of projections used to train the neural network divided by the improvement factor.
Indication of obviousness: the lack in claims 1 and 2 of the disclosed (fig. 3:316,318,320: feedback-evaluation loop 320):
“The workflow also permits users to more quickly identify useful throughput gain amounts through feedback and evaluation, which can further reduce overall time to acquire useful imaging data, reconstructed volumes, trained neural networks, and other data.”
29 throughput: the quantity or amount of raw material processed within a given time, especially the work done by an electronic computer in a given period of time. (Dictionary.com)
30 (italics) represent claim limitations already taught
31 (italics) represent claim limitations already taught
32 subset: Mathematics. a set consisting of elements of a given set (“2D CBCT projection space images (step 944)”) that can be the same as the given set or smaller. (Dictionary.com)
33 on: in connection, association, or cooperation with, wherein with is defined: characterized by or having, wherein have is defined: to possess; own; hold for use; contain, wherein contain is defined: to have as contents content or constituent parts; comprise; include, wherein include is defined: to contain as a subordinate element; involve as a factor: (Dictionary.com)
34 “the training number” is the object of “dividing”
35 “the training number” is the object of “dividing”
36 “the training number” is the object of “dividing”
37 correspond: to be similar or analogous; be equivalent in function, position, amount, etc. (usually followed byto ). (Dictionary.com)
38 contingent limitation: The claimed “the trained neural network is determined to be insufficient” is not a satisfied condition in claim 13; thus, the examiner need not show “evidence” (MPEP 2111.04 II CONTINGENT LIMITATIONS, last para, 5th S) for corresponding claim limitations (shown in italics above) of claim 13 in view of the art under the broadest reasonable interpretation of claim 13.
39 (italics) represent claim limitations already taught
40 (italics) represent claim limitations already taught
41 (italics) represent claim limitations already taught
42 ellipse (…) represent claim limitations already taught
43 (italics) represent claim limitations already taught
44 ellipse (…) represent claim limitations already taught
Prosecution Insights