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 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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 1-5, 10-16, 18, 20-25 & 74 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rodellar et al. (‘Image Processing and Machine Learning in Morphological Analysis of Blood Cells’) in view of Lillard (WO 2013/025418) Pepper et al. (WO 2020/104777).
Claim 1.
Rodellar et al. teaches:
analyzing a plurality of abnormal lymphoid cells obtained from patients having leukemia or lymphoma, including CLL (pp. 46-47; Fig.1);
receiving digital microscopic blood-cell images comprising pixels (pp. 47-48).
segmenting each image to identify nuclear pixels separately from the whole cell, cytoplasm, and background (pp. 48; Fig 2).
measuring nuclear size through geometric descriptors, including nuclear area, perimeter, and shape;
measuring nuclear intensity through color, intensity, and texture descriptors extracted from the nuclear region (pp. 48-50).;
and
using the measured morphological features as inputs to a machine-learning classifier that distinguishes malignant lymphoid-cell types, including CLL (pp. 50-52; Figs. 4-5)
Rodellar does not expressly teach measuring cellular density or treating the subject according to the resulting classification.
Lillard teaches digitizing stained tissue sections and automatically:
performing positive-pixel counting;
quantifying staining intensity per area;
detecting nuclear staining for individual nuclei;
quantifying nuclear intensity; and
determining the number of stained nuclei within the analyzed image area ([0167] & [0189]).
It would have been obvious to modify Rodellar et al. to determine cellular density from the segmented nuclear pixels, as taught by Lillard, because calculating the number of detected nuclei per analyzed area was a known image-analysis technique that would predictably provide an additional quantitative measure of tissue cellularity for cancer classification. See KSR International Co. v. Teleflex Inc, 550 U.S. 398, 417 (2007).
Pepper teaches:
classifying or staging CLL patients using the Binet classification system;
that CLL treatment varies according to staging and disease progression;
earlier intervention for aggressive disease and monitoring of more benign disease until treatment is appropriate; and
treating classified CLL patients using chemotherapy, radiotherapy, monoclonal-antibody therapy, bone-marrow transplantation, or DNA-damage-response inhibitors, optionally combined with DNA-damaging agents or procedures (‘Background’ pp. 1-2, ‘Methods’-CLL Patients, Figs. 1-7, claims 21-36).
Therefore, it would have been obvious to apply Pepper’s classification-responsive treatment to the CLL classification produced by Rodellar et al., as modified by Lillard et al., because doing so would have predictability matched treatment to the identified disease condition and converted the diagnostic classification into clinically useful patient management. See KSR, 550 U.S. at 417-18.
Under the broadest reasonable interpretation:
“a level of CLL, aCLL or RT’ is disjunctive and encompasses determining a CLL classification or stage;
“first,” “second” and “third treatment” do not require particular drugs, dosages, progressively intensive treatment, or the protocols asserted in Applicant’s remarks; and
the treatment steps are contingent upon their respective classifications.
Where the subject is classified as having CLL, the first treatment is required, but the aCLL and RT treatment branches are not required because their conditions have not occurred. Pepper’s CLL classification and corresponding treatment therefore satisfy the applicable treatment branch. See MPEP §2111.04(II); Ex parte Schulhauser, Appeal No. 2013-007847, pp. 9-10 (PTAB Apr. 28, 2016).
Accordingly, the combined teachings render claim 1 obvious.
Claim 2.
Rodellar et al. teaches:
segmenting a blood-cell image into nuclear and non-nuclear cellular regions and extracting measurements from the resulting regions of interest (pp. 47-48; Fig. 2)
Huang et al. teaches:
performing a first segmentation of biological units;
ranking or evaluating segmented units according to shape and scale;
selecting a subset of the segmented units;
and further processing the selected subsets ([0005]-[0006], [0030]-[0032], [0045]-[0047]; Fig. 5).
Wong teaches:
generating a nuclear mask for cell-by-cell analysis;
limiting each nuclear representation to pixels corresponding to the nucleus;
identifying touching or overlapping nuclei as causing unsuccessful or ambiguous segmentation; and
processing such nuclei using watershed segmentation and size-and location-based information ([0042], [0060]-[0078], [0122], [0125]-[0130], [0169]; Figs. 3-5 and 13)
It would have been obvious to exclude from quantitative analysis cells whose nuclear regions remained overlapping or unsuccessfully segmented because both Haung et al. and Wong et al. identify such cells as unreliable segmentation results. Excluding those cells would have predictably prevented inaccurate nuclear-size, intensity, and density measurements.
The pixels corresponding to the retained cells constitute the claimed fourth pixels, and the measurements would thereafter be calculated from that filtered population
ather than retain, cells whose nuclear portions overlap because overlapping objects produce unreliable nuclear-size, intensity, and density measurements. Filtering such objects was a predictable quality control alternative to attempting to resolve ever overlapping object.
The retained segmented pixels correspond to the claimed “fourth pixels” and Rodellar’s feature measurements would consequently be performed using the retained cells.
Therefore, claim 2 is rejected under §103.
Claim 3.
The claim 2 combination teaches the inherited limitations.
Wong et al. teaches:
the relative location two nuclei in an image;
the center of gravity of each nucleus; and
distances between nuclear centers of gravity [0020]
Wong et al. further teaches computing Euclidean distances between segmented nuclei ([0115], [0120]).
It would have been obvious to include intercellular distance in Rodellar’s feature set because spatial organization is a conventional morphological characteristic useful in characterizing a population of cells
Therefore, claim 3 is rejected under §103.
Claim 4.
Wong et al. expressly teaches relative locations and distances between two nuclei in an image [0020].
Wong et al. calculates Euclidean distances using the centers of gravity of the segmented nuclei ([0120], [0134]).
Using the nucleus as the positional reference for its corresponding cell would have been a predictable implementation because the nuclear region is already identified and segmented.
Therefore, claim 4 is rejected under §103.
Claim 5.
Wong et al. teaches determining a center or center of gravity for a segmented nucleus and calculating the distance between two nuclear centers of gravity ([0020], [0115], [0120], [0134], [0175]).
Under the broadest reasonable interpretation, the center of gravity of a segmented nucleus corresponding to the claimed centroid.
Wong et al. also teaches selecting nuclear associations based on the smallest calculated distance [0134].
Therefore, claim 5 is rejected under §103.
Claim 6.
Wong et al. teaches distance-based cell association and nearest-neighbor analysis, including K-nearest-neighbor classification.
Once distances between cells are available, selecting the minimum distance identifies the nearest neighbor.
Under KSR, using the nearest neighbor instead of all pairwise distances is a predictable use of known distance calculations and reduces computation while retaining local density
Claim 10.
Rodellar et al. teaches extracting intensity and color information from segmented blood-cell images using multiple color representations or channels.
Rodellar et al. further teaches calculating statistical descriptors of the pixel values, including average, distribution, and related intensity statistics (pp. 48-49).
A mean, median, standard deviation, histogram value or similar aggregate constitutes the claimed statistical value under its broadest reasonable interpretation.
Therefore, claim 10 is rejected under §103.
Claim 11.
Rodellar et al. and Wong et al. teach measuring the size or area of individual segmented nuclei ([0017], [0022], [0085]-[0087]).
Wong et al. teaches selecting nuclear features and using those features and using those features in a trained classifier. ([0023], [0090]-[0101], [0159]-[0161]); Fig. 16).
Comparing each measured nuclear size with a cutoff is a conventional threshold-classification technique.
Counting the cells satisfying the cutoff would have been the predictable result of applying that threshold to each cell in the filtered population.
Therefore, claim 11 is rejected under §103.
Claim 12.
Wong et al. teaches evaluating classification performance using numbers of cells assigned to respective classes and expresses performance as a ration ([0159]-[0161]; Fig. 16).
Once the cells are separated into cells within and outside a size classification, expressing their relative amounts as a ratio would have been an obvious mathematical normalization.
Using the number outside the classification as the denominator merely expresses the same underlying class counts in a convention relative form.
Therefore, claim 12 is rejected under §103.
Claim 13.
Wong teaches:
Selecting an optimal feature subset;
Evaluating the discrimination power of the features using a classifier; and
Optimizing the classifier using training data [0023].
Wong et al. describes training and testing using known cell-cycle classification and selecting features according to classification performance ([0090]-[0101], [0159]-[0161]; Fig 16).
Selecting a cutoff using an objective function based on quantities of known cell types would have been a conventional implementation of supervised classifier optimization.
Therefore, claim 13 is rejected under §103.
Claim 14.
Rodellar et al. teaches analyzing large blood-cell image datasets containing substantially more than 1,000 cells.
Lillard et al. teaches scanning and automatically analyzing entire digitized tissue sections, which inherently contain large cellular populations ([0167], [0189]).
It would have been obvious to analyze at least 1000 lymphoid cells to increase statistical reliability and reduce sampling error when calculating population-level morphological characteristics. The claimed numerical threshold is not shown to produce a critical or unexpected result.
Therefore, claim 14 is rejected under §103
Claim 15.
Rodellar et al. teaches separately segmenting nuclear and non-nuclear portions of each cell.
The respective pixel counts provide the areas of the nuclear and non-nuclear portions.
Huang et al. teaches ranking segmented cells according to shape and scale, comparing the rankings with a predetermined threshold, and selecting a subset for further processing ([0005]-[0009], [0030]-[0032], [0045]-[0047]; Fig 5).
It would have been obvious to use a nucleus-to-non-nucleus pixel-area ratio as a segmentation-quality metric because an abnormal ratio would identify an incomplete, merged, or otherwise unreliable cell segmentation.
Therefore, claim is rejected under §103.
Claim 16.
Rodellar et al. teaches receiving training images labeled with known blood-cell classifications, extracting nuclear-size and nuclear-intensity features, and training a machine-learning classifier.
Lillard et al. supplies cellular density as an additional measured image feature.
Wong et al. independently teaches:
extracting nuclear-size and nuclear-intensity features;
selecting features according to discriminatory performance; and
optimizing a classifier using labeled training data. ([0017], [0022]-[0023], [0085]-[0101], [0159]-[0161]; Fig 16.)
It would have been obvious to measure the same size, intensity, and density features in both the training images and the subject image because a trained model must use a consistent input-feature space.
Optimizing model parameters based on whether the predicted output matches the known training label is conventional supervised learning.
Therefore, claim 16 is rejected under §103.
Claim 18.
Rodellar et al. and Wong et al. teach training classifiers using labeled cell images and optimizing classification performance using training data. ([0023], [0090]-[0101], [0159]-[0161]).
Actually training the model with the recited plurality of training images is expressly taught or necessarily performed as part of the supervised-learning process.
Therefore, claim 18 is rejected under §103.
Claim 20.
Pepper teaches that CLL changes over time and that disease progression affects treatment selection and clinical management.
Wong et al. teaches that is analysis system:
Classifies biological states;
Tracks those states over time; and
Updated biological conclusions as the cells change ([0009], [0021]-[0024], [0078], [0103]-[0104]).
Wong et al. discloses also analyzes images recorded at consecutive time points and tracks the corresponding classified nuclei ([0019]-[0020], [0089]-[0095], [0120]-[0135]).
It would have been obvious to compare a current disease classification with a previous classification to determine progression because progression represents a change in classification or severity over time. The motivation would have been to determine whether the patient should remain on the existing treatment or receive a different treatment.
Therefore, claim 20 is rejected under §103.
Claim 21.
Rodellar et al. discloses digital blood-cell images having dimensions of approximately 360 x 363 pixels.
Such an image contains approximately 130,680 pixels and therefore satisfies the claimed minimum of 100,000 pixels.
Therefore, claim 21 is rejected under §103.
Claim 22.
Rodellar et al. teaches capturing blood-cell images using microscopy and a digital imaging system.
Lillard et al. teaches scanning microscope slides and automatically analyzing the resulting digital images ([0167], [0189]).
Pepper teaches obtaining and analyzing patient biological samples in connection with CLL.
Applying the known microscopy and digital-image acquisition techniques to a patient biopsy sample would have been a predictable use of established digital-pathology methods.
Claim 23.
Pepper teaches obtaining a biological sample from a patient for CLL evaluation.
Wong et al. expressly teaches that lymphocytes, T cells, B cells, and tumor cells may be obtained by drawing blood or by isolating cells from tissue obtained through biopsy [0052].
Obtaining the biopsy sample before performing microscopy is a necessary and conventional sample preparation step.
Therefore, claim 23 is rejected under §103.
Claim 24.
Lillard et al. teaches scanning larger tissue sections and analyzing selected anatomical regions ([0167], [0189]).
Huang et al. teaches dividing or partitioning an image into subsets of cells and applying further processing to a selected portion of the image ([0030]-[0032], [0045]-[0047]; Fig. 5)
It would have been obvious to select a region containing a representative cellular population before segmentation to:
avoid artifacts and empty regions;
reduce computational burden; and
obtain measurements representative of the biopsy sample.
Therefore, claim is rejected under §103
Claim 25.
Rodellar teaches deriving classifier parameters from measured morphological features and classifying a sample using learned criteria.
Wong et al. teaches selecting an optimal feature subset and evaluating the features using a classifier optimized with training data ([0023], [0090]-[0101], [0159]-[0161]).
Under the broadest reasonable interpretation, a learned threshold, decision boundary, or trained class criterion constitutes a reference value. Comparing a feature-derived parameter with that reference to determine a classification is the ordinary operation of the disclosed classifier.
Therefore, claim 25 is rejected under §103.
Claim 74.
Pepper teaches treating CLL and B-cell malignancies with chemotherapy, targeted therapeutic agents, combination therapy, and radiation therapy.
Under the broadest reasonable interpretation, claim 1 recites conditional treatment alternatives:
Chemotherapy when the classification is CLL;
Chemotherapy plus targeted therapy when the classification is aCLL;
And chemotherapy plus immunotherapy and radiation when the classification is RT.
When the subject is classified as CLL, only the corresponding first-treatment branch is affirmatively performed. The untriggered aCLL and RT branches do not require performance. See MPEP §2111.04 (II); Ex parte Schulhauser.
Pepper’s administration of chemotherapy to a patient having CLL therefore satisfies the operative CLL branch of claims 1 and 74.
Alternatively, Pepper’s progression-dependent treatment disclosure would have suggested using increasingly intensive combination regimens as the disease progresses.
Therefore claim 74 is rejected under §103.
Allowable Subject Matter
Claims 6-7, 9, 17 & 27-28 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
The prior art of record, considered individually and in combination, does not teach or reasonably suggest:
Spatial nearest-neighbor cell analysis;
Restricted four-feature classification;
Morphology-based Hodgkin/non-Hodgkin diagnosis with responsive treatment;
CNN implementation; or
Tile-based identification of regions having elevated mitotic activity.
These limitations, considered with the limitations of the respective parent claims, distinguish the claims from the prior art of record.
Response to Arguments
Applicant’s arguments concerning the §101 rejection are persuasive. As amended, claim 1 applies the diagnostic classification by administering a classification-dependent treatment to the subject. The judicial exception is therefore integrated into a practical application. Accordingly, the §101 rejection is withdrawn.
Applicant’s argument concerning anticipation by Rodellar et al. are also persuasive to the extent that Rodellar et al. alone does not expressly or inherently disclose determining cellular density across the plurality of lymphoid cells. Accordingly, the prior §102 rejection is withdrawn.
The arguments do not, however, overcome the new §103 rejection. Lillard et al. teaches image-based cellular-density assessment through nuclear detection, counting, positive -pixel analysis, and quantification per area, while Pepper teaches selecting treatment according to CLL state or progression. A person of ordinary skill would have combined these teachings with Rodellar’s blood-cell segmentation and morphological classification to obtain population-level density information and use the resulting classification to guide treatment, for the reasons set forth in the rejection.
Applicant’s arguments addressing Rodellar et al. individually are not persuasive against the combined teaching of Rodellar et al., Lillard et al., Pepper, and the additional references applied to the dependent claims. Nonobviousness cannot be established by attacking references individually where the rejection relies upon the combined disclosures.
Nevertheless, the further limitations of claims 6-7, 9, 17, 27-28 are not taught or suggested by the prior art combination. Those claims are therefore objected to but indicated as containing allowable subject matter if rewritten in independent form to include all intervening limitations.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICOLE F JOHNSON whose telephone number is (571)270-5040. The examiner can normally be reached Monday-Friday 8:00am-5:00pm 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, David Hamaoui can be reached at 571-270-5625. 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.
/NICOLE F JOHNSON/Primary Examiner, Art Unit 3796
Prosecution Insights