DETAILED ACTION
Response to Amendment
Claims 1-20 are pending. Claims 1-7 are amended directly or by dependency on an amended claim. Claims 8-20 are new.
Response to Arguments
Applicant’s arguments with respect to claim(s) 1-4, 6 and 7 have been considered but are moot because the new ground of rejection does not rely on the combination of references applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. Examiner notes the limitation “wherein the classifying of each of the LWD image logs comprises classifying a plurality of sections of the LWD image log into a single one of three quality categories” is interpreted as each/every segment has a single score, not that all the different sections are scored and then aggregated
Applicant’s arguments, see page 8, filed 21 May, 2026, with respect to the 35 USC 103 rejection of claim 5 along with accompanying amendments received on the same date have been fully considered and are persuasive. The 5 USC 103 rejection of claim 5 has been withdrawn.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1, 2, 4, 6, 7, and 9-11 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bisht et al. (US 20240062119 A1) in view of Souza et al. (“CNN Prediction Enhancement by Post-Processing for Hydrocarbon Detection in Seismic Images”) in view of Shebl et al. (US 20220207079 A1).
Regarding claim 1, Bisht et al. disclose a method for assessing quality of LWD (Logging While Drilling) image logs (borehole images, [0092], processing the field equipment data using a trained machine learning model to generate a quality score for the field equipment data; and outputting the quality score, abstract, ingest various types of data files e.g., CSV, LAS, DLIS, raster, document, logs etc. where the data engine can output quality metrics, [0154], ML model service that can provide for data agnostic implementation for generation of one or more types of quality metrics for various types of data e.g., log data, seismic survey data, [0196], wireline logging, directional drilling, etc., [0267]), comprising: performing processing of a plurality of LWD image logs (processing the field equipment data using a trained machine learning model to generate a quality score for the field equipment data; and outputting the quality score, abstract, a framework such as the TECHLOG framework can dynamically incorporate data as they are streamed directly from a wellsite for real-time processing and instantaneous analysis as a well is drilled, [0093], automatically processing the field equipment data using a trained machine learning model to generate a quality score for the field equipment data, [0133]); performing normalization of each pseudo-image of the LWD image logs (As to a data pre-processing component, as explained, one or more ML models may be utilized to automatically classify data, uncover data type, etc. As an example, pre-processing can include analysis of the outliers, null treatment, standardization, normalization of data, etc., [0181], As an example, a data pre-processing engine can be implemented that provides for data treatment, which can include analysis of outliers, null treatment, standardization, normalization of data, skewness treatment, etc. As an example, a base data pre-processing pipeline can be implemented to perform actions to convert raw data into processed data before feeding to a trained ANN classifier, [0261]); wherein the classifying of each of the LWD image logs comprises classifying a plurality of sections of the LWD image log into a single one of three quality categories using a trained neural network model for quality assessment of the LWD image log (processing the field equipment data using a trained machine learning model to generate a quality score for the field equipment data, [0133], a control action may be taken in response to a data metric being above, below or at a certain value (e.g., a threshold, etc.). For example, consider taking a sensor off-line where the sensor is generating data of poor quality, [0139], As an example, one or more trained ML models can be suitable for implementation in real-time workflows where streaming data from one or more sources can be assessed to output metrics (e.g., quality score, statues, etc.), [0151], a customizable color coding scheme can be implemented for highlighting data quality scores (e.g., on a scale such as 0 to 100), [0203], deep learning classifier to generate quality metrics (e.g., completeness, fairness, validity, accuracy, etc.)., [0248]) [as scores can be 0-100 this implies at a minimum low/medium/high quality, where low can be 0 high 100 and medium anything between].
Bisht et al. do not disclose subdividing the LWD image logs into smaller pseudo-images of the same size. While scores 0-100 implies at a minimum low/medium/high quality, another reference is added to better teach the classifying of each of the LWD image logs comprises classifying a plurality of sections of the LWD image log into a single one of three quality categories.
Souza et al. disclose a method for assessing the quality of LWD (Logging While Drilling) image logs, characterized in that it comprises: performing the processing of a plurality of LWD image logs (Each image generated for the training set was post-processed through reconstruction, thresholding - binarization and deblurring -, and outlier removal, abstract, The pre-processing stage can be organized into three major steps: cleaning, patch generation, and data augmentation, with their technicalities, part IIB); subdividing the LWD image logs into smaller pseudo-images of the same size (
PNG
media_image1.png
786
474
media_image1.png
Greyscale
, part IIB); and classifying the LWD image log according to its quality, which comprises classifying a plurality of sections of the LWD image log into quality categories, using a trained neural network model for quality assessment of the LWD image log (a simple binarization was applied to the images to leave them with two labels only, thus identifying a given pixel as ``lead'' or ``no-lead''. This way, the original annotations of the geological bounds were eliminated to purge all the no-interest areas of the image and maintain the hydrocarbon region as interpreted. Lead pixels were colored
in white, whereas no-lead pixels were colored in black, resulting in a binary mask as depicted in Fig. 3,
PNG
media_image2.png
46
462
media_image2.png
Greyscale
, collectible vs non collectible designation, see Fig. 4, part IIB) [collectible interpreted as high quality and non-collectible interpreted as low quality].
Bisht et al. and Souza et al. are in the same art of wellbore/seismic/petroleum images (Bisht et al., [0038], [0059], [0092], [0267]; Souza et al., part I). The combination of Souza et al. with Bisht et al. enables subdividing the LWD image logs into smaller pseudo-images of the same size. It would have been obvious at the time of filing to one of ordinary skill in the art to combine the division of Souza et al. with the invention of Bisht et al. as this was known at the time of filing, the combination would have predictable results, and as Souza et al. state “To put another way, a proper classification of leads was hindered due to partial comprehensions of the network while parsing the image patches. Here, we resort to U-net to improve the process of lead identification through patch-based semantic segmentation and postprocessing (image reconstruction, thresholding and outlier removal). With a pixel-wise scrolling, a much more accurate prediction of the potential hydrocarbon zones is delivered. Moreover, since U-net is tailored to label whole images, its architecture becomes a suitable choice to deal with this particular problem. As it will be seen, our outcomes overperform those previously obtained with other methods” (part I), indicating a large commercial benefit of the combination in the petrochemical field.
Bisht et al. do not disclose classifying of each of the LWD image logs comprises classifying a plurality of sections of the LWD image log into a single one of three quality categories.
Shebl et al. teach classifying of each of the LWD image logs comprises classifying a plurality of sections of the LWD image log into a single one of three quality categories using a trained neural network model for quality assessment of the LWD image log (“In examples, one or more steps of FIGS. 1 and 3 may be combined, for example, to implement a stacked neural net model of varying architectures to deliver one or more of: Classification of Reservoir Rock Textures (e.g. High, Medium, Low Reservoir Quality (RQ) Classifications) for example at the step s36, Pore Type Classification and Subjective Quantity for example at the step s40, Framework Grain Type Classification and Subjective Quantity for example at the step s40, Right order words text description of image classification and contents for example at the step s44 and s46, and Petrophysical Properties (Pc (drainage), Rel-K, SOR, FRF, etc.) for example at the step s52. It will also be appreciated that the steps described herein with respect to FIGS. 1 and 3 may be implemented in different orders as appropriate, not just in the order in which they are described herein”, [0082], “For example, the step s12 may classify a thin section according to reservoir rock quality, for example high medium or low quality based on image analysis as mentioned above. In the example of FIG. 10, finer grained dolostone for example as indicated in image 802 may be considered to be a lower reservoir rock quality than more coarsely grained dolostone as indicated in image 804. Also, referring to the example of FIG. 10, a thin section image 806 identified as comprising grainstone may be considered to relate to a high reservoir rock quality, a thin section image 808 identified as comprising packstone may be considered to relate to a medium reservoir rock quality, and a thin section image 810 identified as comprising wackestone may be considered to relate to a low reservoir rock quality”, [0111]) [Note on interpretation: Examiner interpreting as each/every segment has a single score, not that all the different sections are scored and then aggregated].
Bisht et al. and Souza et al. and Shebl et al. are in the same art of analyzing rocks (Bisht et al., [0070]; Souza et al., part I; Shebl et al., abstract). The combination of Shebl et al. with Bisht et al. and Souza et al. enables using three distinct quality categories and VGG16. It would have been obvious at the time of filing to one of ordinary skill in the art to combine the categories and VGG16 of Shebl et al. with the invention of Bisht et al. and Souza et al. as this was known at the time of filing, the combination would have predictable results, and as Shebl et al. state “The examples described herein may help improve rock description efficiency and accuracy, as well as helping accelerate Geomodel development” ([0028]) “the labelled image training database may be used to train convolution (CNN) and encoder-decoder convolution neural nets of varying architectures to a satisfactory accuracy” ([0129]) suggesting an efficiency and accuracy benefit when the inventions are combined.
Regarding claim 2, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Bisht et al. and Souza et al. further indicate processing of the plurality of LWD image logs comprises removing spurious points from the LWD image logs (Bisht et al., pre-processing, outlier treatment, [0185], “normal” and “abnormal” data, where those data points located far away from the normal data point space can be considered outliers and referred to as anomalies, [0260], data treatment service that can perform data adjustments, filtering, etc. Such an approach can include analysis of outliers, null treatment, completeness, standardization, normalization of data, etc., which may provide for conversion of raw data into a standard type of processed data with an improved quality score, [0281]; Souza et al., outlier removal, abstract, part I, part IIG3).
Regarding claim 4, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Souza et al. further indicate performing data augmentation (The pre-processing stage can be organized into three major steps: cleaning, patch generation, and data augmentation; to expand the image bank, data augmentation operations were applied to the patches, such as rotations and shearing part IIB).
Regarding claim 6, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Souza et al. further indicate the neural network model comprises a convolutional neural network architecture followed by a direct network (
PNG
media_image3.png
274
674
media_image3.png
Greyscale
).
Regarding claim 7, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 6. Shebl et al. further teach the convolutional neural network is a VGG16 network (
PNG
media_image4.png
494
452
media_image4.png
Greyscale
, At a step s12, the one or more rock samples are classified, using the classifying module 104, by applying convolutional neural network (CNN) processing to the input rock image data to identify one or more attributes of the input rock samples. For example, the image data may be classified according to one or more of mineralogy (such as dolomite vs calcite based on presence or absence of dolomite crystals), texture, and rock quality, [0057], At a step s34, trained convolution neural network (CNN) processing may be used to classify the rock samples, for example by using the classifying module 104. For example, at a step s36, the thin section rock samples may be classified according to carbonate texture and reservoir rock quality. In some examples, the thin section rock samples may be classified at the step s36 according to dolomite, and rock quality classes, [0071], In examples, the second attribute 204 relates to Limestone Modified Dunham (1971) texture such as whether the texture is classified as Grainstone, Packestone, Wackestone or other texture. In examples, the third attribute 206 relates to the rock quality, such as whether the thin section is classified as high, medium, or low rock quality, [0073], FIG. 20 shows an example of a convolution neural network VGG 16—Open Source VGG16 Deep Convolution Neural Net architecture for implementation by the classifying module 104, [0140]).
Regarding claim 9, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Bisht et al. further indicate the section is classified into the single one of the three quality categories based on which one of the three quality categories is found most in the section (Such categories can be used to measure degree in which data meet a predefined set of data quality demands, for example, percentage of data that are defect free, [0274]).
Regarding claim 10, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Bisht et al. further indicate the classifying of each of the LWD image logs occurs after the performing of the processing of the plurality of LWD image logs and after the performing of the normalization (As to a data pre-processing component, as explained, one or more ML models may be utilized to automatically classify data, uncover data type, etc. As an example, pre-processing can include analysis of the outliers, null treatment, standardization, normalization of data, etc. As an example, a pre-processing pipeline can include features to process raw and/or other data into a form or forms suitable for input to one or more ML models, whether for training, classification, prediction, etc., [0181], [0261], [0281]) [preprocessing indicates the normalization happens before the classification]; and the classifying occurs automatically during well drilling (by creating accurate production scenarios and, with the integration of precise models of the surface facilities and field operations, the INTERSECT framework can produce reliable results, which may be continuously updated by real-time data exchanges (e.g., from one or more types of data acquisition equipment in the field, [0046], As an example, a data quality management framework can provide an automated rule-based QC engine to identify and locate data issues, which may provide for changing and/or and automatically adjusting or synchronizing data according to one or more rules. In such an example, the QC engine can perform checks for data quality across different data quality measurement categories like content, completeness, consistency, uniqueness and validity, [0272], As explained, a ML model-based approach can assess data quality and provide for actions that may improve data quality. Such actions may be in the form of suggestions, available for selection and implementation by a user, or may be in the form of routines that can be performed automatically, [0291]).
Regarding claim 11, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Shebl et al. further indicate the three quality categories are the only quality categories available for classification (High, Medium, Low Reservoir Quality Classifications, [0082]).
Claim(s) 3 is/are rejected under 35 U.S.C. 103 as being unpatentable over Bisht et al. (US 20240062119 A1) and Souza et al. (“CNN Prediction Enhancement by Post-Processing for Hydrocarbon Detection in Seismic Images”) and Shebl et al. (US 20220207079 A1) as applied to claim 1 above, further in view of Fu et al. (“Deep learning based lithology classification of drill core images”).
Regarding claim 3, Bisht et al. and Souza et al. and Shebl et al. disclose the method according to claim 1. Bisht et al. and Souza et al. and Shebl et al. do not disclose performing the processing of the plurality of LWD image logs comprises defining a single classification window of 120 rows and 120 columns.
Fu et al. teach defining a single classification window of 120 rows and 120 columns (
PNG
media_image5.png
432
604
media_image5.png
Greyscale
, These structures usually contain impurities other than their own lithology, such as sediment particles, tree roots, etc. An example of the image is shown in Fig 3. In addition to the lithology that needs to be classified, it is one way to add an extra ‘garbage’ class to all lithology categories to absorb these non-core parts, part 2.4).
Bisht et al. and Souza et al. and Fu et al. are in the same art of wellbore/seismic/petroleum images (Bisht et al., [0038], [0059], [0092], [0267]; Souza et al., part I; Fu et al., part 2.3). The combination of Fu et al. with Bisht et al. and Souza et al. and Shebl et al. enables subdividing the LWD image logs into defining a single classification window of 120 rows and 120 columns. It would have been obvious at the time of filing to one of ordinary skill in the art to combine the 120x120 division of Fu et al. with the invention of Bisht et al. and Souza et al. and Shebl et al. as this was known at the time of filing, the combination would have predictable results, and as Fu et al. state “Drill core lithology is an important indicator reflecting the geological conditions of the drilling area. Traditional lithology identification usually relies on manual visual inspection, which is time-consuming and professionally demanding. In recent years, the rapid development of convolutional neural networks has provided an innovative way for the automatic prediction of drill core images. In this work, a core dataset containing a total of 10 common lithology categories in underground engineering was constructed… The test results show that the proposed method is optimal and effective for automatic lithology classification of borehole cores” and “The test results showed that larger crop sizes are generally better, include more lithology information, and have lower losses in training and validation” (part 2.2) indicating a large commercial benefit to the combination in the petrochemical field, and as it has been held that discovering an optimum value of a result effective variable involves only routine skill in the art In re Boesch, 617 F.2d 272, 205 USPQ 215 (CCPA 1980).
Allowable Subject Matter
Claims 5 and 8 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.
Claims 12-20 are allowed.
The following is an examiner’s statement of reasons for allowance: The following art is cited as relevant but not sufficient alone or in combination to disclose, teach or fairly suggest the subject matter of the allowed claims:
“Improving geological logging of drill holes using geochemical data and data analytics for mineral exploration in the Gawler Ranges, South Australia”: With data clustering, the number of classes needs to be estimated. Before performing data clustering,
two choices must be made: type of algorithm and number of clusters (classes). The best algorithm to choose depends on the structure of the data. The number of clusters to choose is generally problem-dependent unless the clusters are clearly separable, in which case criteria can be applied that determine the optimal number of clusters. Unfortunately, clearly separable clusters are rare in geological data.
PNG
media_image6.png
600
826
media_image6.png
Greyscale
Instead of using a larger sample size to overcome noisy results, samples can be composited mathematically using the continuous wavelet transform (CWT) to provide spatially continuous domains consisting of samples of similar composition. By using CWT for compositing similar rock units, boundary precision is retained because the size of the composited units is not
fixed, instead boundaries are determined by points of change in the signal. The application of CWT methods for extracting geological boundaries has been widely tested for wireline logging
in the petroleum industry.
PNG
media_image7.png
502
776
media_image7.png
Greyscale
“Improving Automated Geological Logging of Drill Holes by Incorporating Multiscale Spatial Methods”: Manually interpreting multivariate drill hole data is very time-consuming,
and different geologists will produce different results due to the subjective nature of
geological interpretation. Automated or semi-automated interpretation of numerical
drill hole data is required to reduce time and subjectivity of this process. However,
results from machine learning algorithms applied to drill holes, without reference to
spatial information, typically result in numerous small-scale units. These small-scale
units result not only from the presence of very small rock units, which may be below
the scale of interest, but also from misclassification. A novel method is proposed
that uses the continuous wavelet transform to identify geological boundaries and uses
wavelet coefficients to indicate boundary strength. The wavelet coefficient is a useful
measure of boundary strength because it reflects both wavelength and amplitude of
features in the signal. Thismeans that boundary strength is an indicator of the apparent
thickness of geological units and the amount of change occurring at each geological
boundary. For multivariate data, boundaries from multiple variables are combined and
multiscale domains are calculated using the combined boundary strengths. The method
is demonstrated using multi-element geochemical data from mineral exploration drill
holes. The method is fast, reduces misclassification, provides a choice of scales of
interpretation and results in hierarchical classification for large scales where domains
may contain more than one rock type.
PNG
media_image8.png
390
528
media_image8.png
Greyscale
Any comments considered necessary by applicant must be submitted no later than the payment of the issue fee and, to avoid processing delays, should preferably accompany the issue fee. Such submissions should be clearly labeled “Comments on Statement of Reasons for Allowance.”
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 MICHELLE M ENTEZARI HAUSMANN whose telephone number is (571)270-5084. The examiner can normally be reached 10-7 M-F.
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, Vincent M Rudolph can be reached at (571) 272-8243. 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.
/MICHELLE M ENTEZARI HAUSMANN/Primary Examiner, Art Unit 2671
Prosecution Insights