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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 02/05/2026 was filed. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Arguments
Applicant's arguments filed 02/04/2026 have been fully considered but they are not persuasive.
In regard to Response to Rejections- Claim 1, see Applicant’s Remarks pgs. 10-14, Applicant first argues Mao fails to teach “training a model to predict a future virus titer of a cell culture using one or more images of the cell culture. Mao is instead directed to "a virus plaque detection and segmentation method."” Examiner would like to point out that below Mao discusses using an image as training data to train a machine learning model, in Mao’s case that is an SVM classifier. Then the training image is used to help predict unlabeled patches in order to figure out labels. Examiner interprets this as using images to help train a machine learning model to predict viruses.
Specifically:
(Mao, pgs. 3191 “Step 2: The image was divided into small connected regions, called cells, and the histograms of gradient directions or edge orientations in each cell were calculated. Each pixel contributed a weighted vote for orientation. In the implementation, each cell size was 4×4 pixels, and the orientation (0-180 degree) was separated into 9 histogram bins equally. Step 3: The 2×2 cells were combined into a block whose size was 8x8 pixels and histogram normalization was performed on the block. Each patch was composed of 6x6 blocks and the size of patch was 48x48. The descriptor was the vector of all components of the normalized cell responses from all of the blocks in the patch. Finally, a 4464-dimentional vector, denoted as uk , described the feature of the k-th patch.” And pg. 3197, 4. Conclusion, paragraph 1, “After preprocessing the original image by the Wiener filter, the training patches were randomly selected from the images [wherein the input comprises one or more images], and the HOG features were extracted from the corresponding training patches. The set of training patches and their corresponding HOG features were used for training the initial linear SVM classifier, which was harnessed to scan the training images exhaustively to predict the labels for the unlabeled patches [training one or more machine learning models with the numeric representations to use an input to generate an output].”)
Next, Applicant argues that Masci “is also not related to predicting a future virus titer.” Applicant also argues that Masci fails to teach training a machine learning model. Examiner would like to point out, that the portion of the claim that Masci teaches does not discuss training a machine learning model. Masci is used to bring in the portion of the claim that has a readout of the cell culture in the final time, as seen below.
Specifically:
(Masci, pg. 271, Col. 2, paragraph 2, “Using the Celigo imager, the readout from the traditional assay can be reduced to 2 days post-infection [readout of the virus- treated cell culture at the final time tfinal;], since the Celigo imager can detect and count plaques [recording at least one numeric virus titer] that are harder to visualize by eye.” And pg. 272, Col. 1, paragraph 4, “With optimization of parameters, the Celigo can be used to recognize and count individual foci, thus aiding in the speed of the readout of the assay.”)
Next, Applicant argues that Neher “fails to teach or suggest training a machine learning model to predict the future virus titer of a cell culture by processing an image that depicts the unknown cell culture.“ Examiner would like to point out that Neher discusses training models that are of a tree structure which is interpreted as a machine learning model. Neher then goes on to describe that these models are trained to predict a titer, using unknown data, as seen below in more detail.
Specifically:
(Neher, pg. E1702, Col. 2, paragraph 4, “We evaluated the performance of the models in predicting HI titers for different influenza lineages A(H3N2), A(H1N1)pdm09, B/Yam, and B/Vic [a prediction of a future virus titer of the second cell culture.]. We trained the models on 90% of the data and used the remaining measurements to validate the models as in ref. 5. The number of viruses, number of antisera, HI measurements, etc., for each lineage are provided as Table S1.” And paragraph 6, “Both models predicted titers for viruses not part of the training set [an unknown virus population] to an accuracy of approximately 0.75 log2 titer levels (Table 1 and Fig. 2B).”)
In regard to Ponomarev, Applicant argues it is directed to object identification through segmentation. Examiner would like to point out that in Ponomarev, the training set obtained is of images of viral particals which are being interpreted as the viral-treated cells.
Specifically:
(Ponomarev, paragraph 0021, “For example, it could be utilized for the rapid and automated quantitation of image features that are of interest to the pathologist such as counts of viral particles [plurality of images of virus-treated cell cultures] and abnormal cells in cancer screening and disease diagnosis.” And paragraph 0087, “The image dataset contains an electronic representation of an image [obtaining a training set in a form of a plurality of images of virus-treated cell cultures from a plurality of experiments]. The image dataset may be a database, a computer file, or an array of data in computer memory. The image may be a 2-dimensional or 3-dimensional image. Additionally, the image may be a time-sliced image representing the state of the image at a particular time sequence [at one or more time points from a start time to to a final time tfinal].”)
Lastly, regarding claim 15, see Applicant’s Remarks pg. 15, Applicant states that Zhou does not teach “wherein the training step (4) comprises minimizing an error between a prediction of the future virus titer provided by the one or more machine learning models and a ground truth associated with the at least one numeric virus titer readout of the first virus-treated cell culture at the final time tfinai." Examiner would like to point out that Zhou discusses achieving the best performance, examiner interpreting this as minimizing error.
Specifically:
(Zhou, pg. 90941, Col. 2, paragraph 2, “With the exception of one team, all teams use DCNNs as part of the processing pipeline [provided by the one or more machine learning models]. For the first task, the best performing method achieves a quadratic-weighted Cohen's kappa score of k D 0:567, 95% CI [0.464, 0.671] between the predicted scores and the ground truth [comprises minimizing an error between a prediction of the final virus titer and a ground truth]. For the second task, the predictions of the top method have a Spearman's correlation coefficient of r D 0:617, 95% CI [0.581, 0.651] with the ground truth. [the at least one numeric virus titer readout of the virus-treated cell culture at the final time tfinal., second task being interpreted as the final time]”)
Therefore, the 35 USC 103 is maintained.
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 non-obviousness.
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.
Claims 1-2, 6, 17, 19, 24 and 31-32 are rejected under 35 U.S.C. 103 as being unpatentable over Mao et al (Detection and segmentation of virus plaque using HOG and SVM: Toward automatic plaque assay, "Mao") published 2014, in view of Masci et al (Integration of Fluorescence Detection and Image-Based Automated Counting Increases Speed, Sensitivity, and Robustness of Plaque Assays, "Masci") published Sept. 2019, Neher et al (Prediction, dynamics, and visualization of antigenic phenotypes of seasonal influenza viruses, “Neher”) published Mar. 7, 2016 and in further view of Ponomarev et al (US Published Patent Application No. 20040114800, "Ponomarev") published Jun. 17, 2004.
In regard to claim 1 and analogous claims 24 and 31, Mao teaches A method comprising the steps of: (1) obtaining a training set in a form of a plurality of images of virus-treated cell cultures (Mao, pg. 3188, paragraph 3, “In this paper, an effective and novel method for the detection and segmentation of plaques is presented by means of histogram of oriented gradient (HOG) features [14,15] and support vector machine (SVM) [16]. This method is comprised of the training and the testing stage. In the training stage, the images are pre-processed for noise suppression. Then, the initial training data [obtaining a training set in a form of a plurality of images of virus-treated cell cultures], is set up by sampling the positive training patches (with a plaque at their centers) and negative training patches (plaque-free) from the training images [A method for training a machine learning model to predict a virus titer from an image or a sequence of images], and extracting the HOG features from each training patch.”)
(3) processing the images in the training set to acquire a numeric representation of each image; and (Mao, pg. 3190, 2.1 Pre-processing image [processing the images in the training set] and extracting HOG feature, paragraph 2, “The HOG, a popular feature descriptor, was first used to calculate the occurrence of gradient direction in the local patch of an image. It captured edge and gradient structures, which are characteristic of local shapes, while it performed photometric and geometric transformations [14]. In this study, the image was denoted as I(x,y) with an image size of M x N, where and y were the pixel position. [acquire a numeric representation of each image;, x and y are being interpreted as number representation of the image]”)
4) training one or more machine learning models with the numeric representations to use one or more images of a second cell culture containing an unknown virus population to generate (Mao, pgs. 3191 “Step 2: The image was divided into small connected regions, called cells, and the histograms of gradient directions or edge orientations in each cell were calculated. Each pixel contributed a weighted vote for orientation. In the implementation, each cell size was 4×4 pixels, and the orientation (0-180 degree) was separated into 9 histogram bins equally. Step 3: The 2×2 cells were combined into a block whose size was 8x8 pixels and histogram normalization was performed on the block. Each patch was composed of 6x6 blocks and the size of patch was 48x48. The descriptor was the vector of all components of the normalized cell responses from all of the blocks in the patch. Finally, a 4464-dimentional vector, denoted as uk , described the feature of the k-th patch.” And pg. 3197, 4. Conclusion, paragraph 1, “After preprocessing the original image by the Wiener filter, the training patches were randomly selected from the images [wherein the input comprises one or more images], and the HOG features were extracted from the corresponding training patches. The set of training patches and their corresponding HOG features were used for training the initial linear SVM classifier, which was harnessed to scan the training images exhaustively to predict the labels for the unlabeled patches [training one or more machine learning models with the numeric representations to use an input to generate an output].”)
However, Mao does not explicitly teach from a plurality of experiments at one or more time points from a start time to to a final time tfinal;
(2) for each experiment, recording at least one numeric virus titer readout of the virus- treated cell culture at the final time tfinal;
containing an unknown virus population to generate a prediction of a future virus titer of the second cell culture.
Masci teaches (2) for each experiment, recording at least one numeric virus titer readout of the virus- treated cell culture at the final time tfinal; (Masci, pg. 271, Col. 2, paragraph 2, “Using the Celigo imager, the readout from the traditional assay can be reduced to 2 days post-infection [readout of the virus- treated cell culture at the final time tfinal;], since the Celigo imager can detect and count plaques [recording at least one numeric virus titer] that are harder to visualize by eye.” And pg. 272, Col. 1, paragraph 4, “With optimization of parameters, the Celigo can be used to recognize and count individual foci, thus aiding in the speed of the readout of the assay.”)
Mao and Masci are related to the same field of endeavor (i.e. plaque assay). In view of the teachings of Masci, it would have been obvious for a person with ordinary skill in the art to apply the teachings of Masci to Mao before the effective filing date of the claimed invention in order to provide an efficient production process. (Masci, pg. 270, Col. 2, “Because optimal rAAV production can be dependent on the concentration of helper virus used, helper virus infectious titer is one of the critical factors in establishing an efficient and highyielding rAAV production process.”)
However, Mao and Masci do not explicitly teach from a plurality of experiments at one or more time points from a start time to to a final time tfinal;
containing an unknown virus population to generate a prediction of a future virus titer of the second cell culture.
Neher teaches containing an unknown virus population to generate a prediction of a future virus titer of the second cell culture. (Neher, pg. E1702, Col. 2, paragraph 4, “We evaluated the performance of the models in predicting HI titers for different influenza lineages A(H3N2), A(H1N1)pdm09, B/Yam, and B/Vic [a prediction of a future virus titer of the second cell culture.]. We trained the models on 90% of the data and used the remaining measurements to validate the models as in ref. 5. The number of viruses, number of antisera, HI measurements, etc., for each lineage are provided as Table S1.” And paragraph 6, “Both models predicted titers for viruses not part of the training set [an unknown virus population] to an accuracy of approximately 0.75 log2 titer levels (Table 1 and Fig. 2B).”)
Mao, Masci and Neher are related to the same field of endeavor (i.e. image detection). In view of the teachings of Neher, it would have been obvious for a person with ordinary skill in the art to apply the teachings of Neher to Mao and Masci before the effective filing date of the claimed invention in order to utilize model to predict virus titers with better accuracy. (Neher, pg. E1706, Col. 2, paragraph 1, “Both of these models predict titers with similar or better accuracy than cartographic approaches (4).”)
Mao, Masci and Neher do not explicitly teach at one or more time points from a start time to to a final time tfinal. Mao teaches using a database composed of various images of plaques to create training sets, see Section 3 but does not expressly disclose that the training images are at one or more time points from a start time to to a final time tfinal.
Ponomarev teaches (1) obtaining a training set in a form of a plurality of images of virus-treated cell cultures from a plurality of experiments at one or more time points from a start time to to a final time tfinal; (Ponomarev, paragraph 0021, “For example, it could be utilized for the rapid and automated quantitation of image features that are of interest to the pathologist such as counts of viral particles [plurality of images of virus-treated cell cultures] and abnormal cells in cancer screening and disease diagnosis.” And paragraph 0087, “The image dataset contains an electronic representation of an image [obtaining a training set in a form of a plurality of images of virus-treated cell cultures from a plurality of experiments]. The image dataset may be a database, a computer file, or an array of data in computer memory. The image may be a 2-dimensional or 3-dimensional image. Additionally, the image may be a time-sliced image representing the state of the image at a particular time sequence [at one or more time points from a start time to to a final time tfinal].”)
Mao, Masci, Neher and Ponomarev are related to the same field of endeavor (i.e. image detection). In view of the teachings of Ponomarev, it would have been obvious for a person with ordinary skill in the art to apply the teachings of Ponomarev to Mao, Masci and Neher before the effective filing date of the claimed invention in order to allow for segmentation of homogenous or inhomogeneous objects. (Ponomarev, Abstract, “The system and method allows for segmenting homogenous or inhomogeneous objects of rather uniform dimensions and geometry in 2-dimensional or 3-dimensional images.”)
In regard to claim 24, the claim recites similar limitations as corresponding claim 1, and is rejected for similar reasons as claim 1 using similar teachings and rationale.
Masci further teaches An analytical instrument, comprising: a system configured to hold one or more plates containing a cell culture and a virus sample; an integrated imaging system (Masci, pg. 271, Col. 1, paragraph 3, “The Celigo imaging system [an integrated imaging system] from Nexcelom [An analytical instrument] was evaluated for its ability to image and accurately count plaques for the helper virus used during process development.” And pg. 270, Col. 2, paragraph 3, “In general, cells are seeded in multi-well plates to achieve a confluent monolayer. The following day, cells are inoculated with diluted viral samples for a specific amount of time (dependent on the helper virus being tittered [to hold one or more plates containing a cell culture and a virus sample]).”)
In regard to claim 31, the claim recites similar limitations as corresponding claim 1, and is rejected for similar reasons as claim 1 using similar teachings and rationale.
A non-transitory computer readable medium storing instructions that when executed by a processing unit associated with an analytical instrument comprises an integrated imaging system (Ponomarev, paragraph 0135, “For example, in a networked environment, such as the Internet, or an intranet, a client application may be utilized to submit an image dataset to a central server for processing to determine segmented objects in the image.”)
In regard to claim 2 and analogous claim 19, Mao, Masci, Neher and Ponomarev teach the method of claim 1 and claim 18.
Masci further teaches wherein the at least one numeric virus titer readout comprises (1) a number of infective particles or a number of infective particles per unit volume, (2) a Tissue Culture Infective Dose 50 % Assay, or (3) the readout from a focus-forminq assay. (Masci, pg. 271, Col. 2, Discussion, “50% tissue culture infectious dose (TCID50) is another commonly used assay [(2) a Tissue Culture Infective Dose 50 % Assay] to detect infectious viral titer. TCID50 is an endpoint dilution assay that determines what dilution of a viral sample is needed to infect 50% of inoculated cells. After infection, infectious titer is measured by qPCR.”)
Mao and Masci are combinable for the same rationale as set forth above with respect to claim 1.
In regard to claim 6, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
Masci further teaches wherein the training set comprises (1) label-free light microscopy images, (2) fluorescence images of the cell culture labelled with a fluorescent marker, or (3) immunohistochemistry images of the cell culture labeled with a chromogenic detection system. (Masci, pg. 271, Col. 2, paragraph 1, “We tested whether fluorescence imaging using a fluorescent-labeled [fluorescence images of the cell culture labelled with a fluorescent marker] antibody could be used as an improved method of detection in this assay. As shown in Figure 3A, plaques stained with the fluorescent-labeled antibody, imaged, and counted using the Celigo imager were similarly visible after 3 days, or 72 hours postinfection. Importantly, both the fluorescent detection method and the traditional HRP-based method resulted in similar counts of plaques 3 days post-infection.”)
Mao and Masci are combinable for the same rationale as set forth above with respect to claim 1.
In regard to claim 17, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
Mao further teaches wherein the one or more time points comprise at least two time points in step (1), and wherein a period between the time points is less than or equal to 60 minutes. (Mao, pg. 3195, paragraph 5, “Given a trained classifier, the running time from scanning the testing image using sliding-window to outputting quantitative and qualitative results [at least two time points in step (1)] was less than 30 seconds. [less than or equal to 60 minutes.]”)
In regard to claim 32, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
Mao further teaches wherein processing the images in the training set comprises processing the images in the training set to acquire the numeric representation of each of the images that are captured at a time before tfinal. (Mao, pg. 3190, 2.1 Pre-processing image and extracting HOG feature, “The HOG, a popular feature descriptor, was first used to calculate the occurrence of gradient direction in the local patch of an image. It captured edge and gradient structures, which are characteristic of local shapes, while it performed photometric and geometric transformations [14]. In this study, the image was denoted as I(x,y) with an image size of M x N , where x and y were the pixel position [the images in the training set to acquire the numeric representation of each of the images]. The HOG features were extracted in the following steps [15]:”)
In regard to claim 33, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
Neher further teaches wherein the prediction of the future virus titer is a prediction of a number of infective particles, a number of infective particles per unit volume, or a readout of a Tissue Culture Infective Dose 50% Assay. (Neher, pg. E1702, Col. 2, paragraph 4, “We evaluated the performance of the models in predicting HI titers for different influenza lineages A(H3N2), A(H1N1)pdm09, B/Yam, and B/Vic [a prediction of a future virus titer]. We trained the models on 90% of the data and used the remaining measurements to validate the models as in ref. 5. The number of viruses, number of antisera, HI measurements, etc. [number of infective particles], for each lineage are provided as Table S1.” And paragraph 6, “Both models predicted titers for viruses not part of the training set to an accuracy of approximately 0.75 log2 titer levels (Table 1 and Fig. 2B).”)
Mao and Neher are combinable for the same rationale as set forth above with respect to claim 1.
Claims 9-10, 14-16, 18, 20-23, 25-26 and 30 are rejected under 35 U.S.C. 103 as being unpatentable over Mao, in view of Masci, Neher, Ponomarev and in even further view of Zhou et al (A Comprehensive Review for Breast Histopathology Image Analysis Using Classical and Deep Neural Networks, "Zhou") published May 27, 2020.
In regard to claim 9, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
However, Mao, Masci, Neher and Ponomarev do not explicitly teach wherein the processing step (3) comprises passing the images through a convolutional neural network (CNN) to acquire an intermediate data representation of the images.
Zhou teaches wherein the processing step (3) comprises passing the images through a convolutional neural network (CNN) to acquire an intermediate data representation of the images. (Zhou, pg. 90935, Col. 2, paragraph 2, “In [52], the classification of breast cancer histopathological images [passing the images through] by a Convolutional Neural Network (CNN) [a convolutional neural network (CNN)] independent of magnification is proposed… Finally, the average recognition rate of the single-task CNN model in the benign/malignant classification task [acquire an intermediate data representation of the images, classification task being interpreted as the intermediate representation] is 83:25%.”)
Mao, Masci, Neher, Ponomarev and Zhou are related to the same field of endeavor (i.e. plaque assay). In view of the teachings of Zhou, it would have been obvious for a person with ordinary skill in the art to apply the teachings of Zhou to Mao, Masci, Neher and Ponomarev before the effective filing date of the claimed invention in order to improve the accuracy of image analysis. (Zhou, pg. 90931, Abstract, “To improve the accuracy and objectivity of Breast Histopathological Image Analysis (BHIA), Artificial Neural Network (ANN) approaches are widely used in the segmentation and classification tasks of breast histopathological images.”)
In regard to claim 10 and analogous claim 26, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
Mao further teaches b) calculating a cell-by-cell numeric description of each cell; and (Mao, pg. 3191, Step 2, paragraph 1, “The image was divided into small connected regions, called cells, and the histograms of gradient directions or edge orientations in each cell were calculated. Each pixel contributed a weighted vote for orientation. In the implementation, each cell size was 4×4 pixels, and the orientation (0-180 degree) was separated into 9 histogram bins equally.”)
However, Mao, Masci, Neher and Ponomarev do not explicitly teach a) segmenting individual cells from the images;
c) aggregating the numeric descriptions over all cells.
Zhou teaches a) segmenting individual cells from the images; (Zhou, pg. 90943, Col. 2, paragraph 1, “In [139], a CNN based model with three hidden layers is built to segment the breast cancer cell nucleus in histopathological images.”)
c) aggregating the numeric descriptions over all cells. (Zhou, pg. 90934, Col. 2, paragraph 2, “scheme based on histopathological images is proposed. First edge, texture and intensity features are extracted. Then based on each of the extracted features, an ANN classifier is designed, respectively. Thirdly, an ensemble learning approach, namely ``random subspace ensemble'', is use to select and aggregate these classifiers for even better classification performance.”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 14 and analogous claim 30, Mao, Masci, Neher and Ponomarev teach the method of claim 1 and claim 24.
However, Mao, Masci, Neher and Ponomarev do not explicitly teach wherein the machine learning model comprises one of: a partial least squares linear model, an artificial neural network, a Gaussian process regression, and a neural ordinary differential equation model.
Zhou teaches wherein the machine learning model comprises one of: a partial least squares linear model, an artificial neural network, a Gaussian process regression, and a neural ordinary differential equation model. (Zhou, pg. 90932, Col. 2, B. Motivation of Our Review Paper, “This paper focuses on the work of ANNs [an artificial neural network, ANN = artificial neural network] in the image analysis of breast histopathology. A comprehensive overview of techniques for image analysis of breast histopathology using classical neural networks and deep neural networks is presented.”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 15, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
However, Mao, Masci, Neher and Ponomarev do not explicitly teach wherein the training step (4) comprises minimizing an error between a prediction of the future virus titer provided by the one or more machine learning models and a ground truth associated with the at least one numeric virus titer readout of the virus-treated cell culture at the final time tfinal.
Zhou teaches wherein the training step (4) comprises minimizing an error between a prediction of the final virus titer provided by the one or more machine learning models and a ground truth associated with the at least one numeric virus titer readout of the virus-treated cell culture at the final time tfinal. (Zhou, pg. 90941, Col. 2, paragraph 2, “With the exception of one team, all teams use DCNNs as part of the processing pipeline [provided by the one or more machine learning models]. For the first task, the best performing method achieves a quadratic-weighted Cohen's kappa score of k D 0:567, 95% CI [0.464, 0.671] between the predicted scores and the ground truth [comprises minimizing an error between a prediction of the final virus titer and a ground truth]. For the second task, the predictions of the top method have a Spearman's correlation coefficient of r D 0:617, 95% CI [0.581, 0.651] with the ground truth. [the at least one numeric virus titer readout of the virus-treated cell culture at the final time tfinal., second task being interpreted as the final time]”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 16, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
However, Mao, Masci, Neher and Ponomarev do not explicitly teach further comprising a step of repeating steps (1) - (4) for different classes of viruses, different cell types, or different machine learning models for each time point.
Zhou teaches further comprising a step of repeating steps (1) - (4) for different classes of viruses, different cell types, or different machine learning models for each time point. (Zhou, pg. 90939, Col. 1, paragraph 1, “The Grand Challenge on Breast Cancer Histology images (BACH) is co-organized with the 15th International Conference on Image Analysis and Recognition (ICIAR 2018) [10]. There are two goals in this challenge. Part A of the challenge consists of automatically classifying H&E-stained breast histology microscopy images in four classes: Normal, benign, in-situ carcinoma and invasive carcinoma. [different cell types, examiner interprets that with each different cell they will repeat the steps]”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 18, Mao, Masci, Neher and Ponomarev teach the method of claim 1.
Mao further teaches a) obtaining the one or more images, wherein the one or more images comprise a time sequence of images of the second cell culture containing the unknown virus population; (Mao, pgs. 3191 “Step 2: The image was divided into small connected regions, called cells, and the histograms of gradient directions or edge orientations in each cell were calculated. Each pixel contributed a weighted vote for orientation. In the implementation, each cell size was 4×4 pixels, and the orientation (0-180 degree) was separated into 9 histogram bins equally. Step 3: The 2×2 cells were combined into a block whose size was 8x8 pixels and histogram normalization was performed on the block. Each patch was composed of 6x6 blocks and the size of patch was 48x48 [obtaining the one or more images, wherein the one or more images comprise a time sequence of images of the second cell culture]. The descriptor was the vector of all components of the normalized cell responses from all of the blocks in the patch. Finally, a 4464-dimentional vector, denoted as uk , described the feature of the k-th patch.” And pg. 3197, 4. Conclusion, paragraph 1, “After preprocessing the original image by the Wiener filter, the training patches were randomly selected from the images, and the HOG features were extracted from the corresponding training patches [containing the unknown virus population]. The set of training patches and their corresponding HOG features were used for training the initial linear SVM classifier, which was harnessed to scan the training images exhaustively to predict the labels for the unlabeled patches.”)
c) making the prediction with the one or more trained machine learning models. (Mao, pgs. 3192-3193, 2.3 Detecting plaques, “The sliding-window technique was utilized to search exhaustively for positive patches. A fixed-size rectangular window (48×48) was used to scan the image with the stride of 12 rows. The HOG features were extracted from the window and inputted into the learnt classifier, and then a prediction [making a prediction with the one or more trained machine learning models] whether the patch was positive was returned.”)
However, Mao, Masci, Neher and Ponomarev do not explicitly teach b) supplying a numeric representation of the time sequence of images obtained in step a) to the one or more machine learning models; and
Zhou teaches b) supplying a numeric representation of the time sequence of images obtained in step a) to the one or more machine learning models; and (Zhou, pg. 90935, Col. 2, paragraph 2, “Finally, the average recognition rate of the single-task CNN model in the benign/malignant classification task is 83:25%. The average recognition rate of the multi-task CNN model in the benign/malignant classification task is 82:13% and the average recognition rate in the magnification estimation task is 80.10%. [supplying a numeric representation of the time sequence of images obtained in step a) to the one or more machine learning models]”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 20, Mao, Masci, Neher, Ponomarev and Zhou teach the method of claim 18.
Masci further teaches wherein the time sequence of images obtained in step a) are obtained in an instrument holding one or more culture plates containing the second cell culture and having an integral imaging system. (Masci, pg. 270, Col. 2, paragraph 4, “Here, we describe the implementation of the Celigo imaging system from Nexcelom into the traditional plaque assay workflow to automatically identify, image, and count plaques [are obtained in an instrument holding one or more culture plates containing the second cell culture and having an integral imaging system.]. Automated plaque counting is not only consistent with manual plaque counting—it also improves the assay robustness by eliminating human bias.”)
Mao and Masci are combinable for the same rationale as set forth above with respect to claim 1.
In regard to claim 21, Mao, Masci, Neher, Ponomarev and Zhou teach the method of claim 18.
Masci further teaches wherein the imaging system comprises a fluorescence imaging system. (Masci, pg. 270, Col. 2, paragraph 4, “Here, we describe the implementation of the Celigo imaging system [fluorescence imaging system, examiner researched that this is a fluorescence imaging system] from Nexcelom into the traditional plaque assay workflow to automatically identify, image, and count plaques… We also integrate fluorescence detection of plaques…”)
Mao and Masci are combinable for the same rationale as set forth above with respect to claim 1.
In regard to claim 22, Mao, Masci, Neher, Ponomarev and Zhou teach the method of claim 18.
Zhou further teaches wherein the second cell culture further comprises specialist media aiding in imaging of the second cell culture. (Zhou, pg. 90932, Col.1, paragraph 1, “Hematoxylin and Eosin (H&E) staining approach is the most common method. The hematoxylin dyes [specialist media aiding in imaging of the second cell culture] the nuclei a dark purple color and the eosin dyes other structures (cytoplasm, stroma, etc.) a pink color.”)
Mao, Masci, Neher and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 23, Mao, Masci, Neher, Ponomarev and Zhou teach the method of claim 22.
Zhou further teaches wherein the specialist media further comprises at least one of: reagents that react to a release of cell contents, reagents that aggregate on viral antigens, fluorogenic dyes, and reagents used for early detection of cytopathic effects such as live versus dead cell detection, activation of apoptosis and autophagy pathways, cell cycle, and oxidative stress. (Masci, pg. 270, Col. 2, paragraph 4, “We also integrate fluorescence detection [fluorogenic dyes] of plaques…”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
In regard to claim 25, Mao, Masci, Neher, Ponomarev and Zhou teach the analytical instrument of claim 24.
Mao further teaches (1) obtaining a training set in a form of a plurality of images of virus-treated cell cultures from a plurality of experiments at one or more time points from a start time to to a final time tfinal; (Mao, pg. 3188, paragraph 3, “In this paper, an effective and novel method for the detection and segmentation of plaques is presented by means of histogram of oriented gradient (HOG) features [14,15] and support vector machine (SVM) [16]. This method is comprised of the training and the testing stage. In the training stage, the images are pre-processed for noise suppression. Then, the initial training data [obtaining a training set in a form of a plurality of images of virus-treated cell cultures from a plurality of experiments at one or more time points from a start time to to a final time tfinal], is set up by sampling the positive training patches (with a plaque at their centers) and negative training patches (plaque-free) from the training images [A method for training a machine learning model to predict a virus titer from an image or a sequence of images], and extracting the HOG features from each training patch.”)
(3) processing the images in the training set to acquire a numeric representation of each
image; and (Mao, pg. 3190, 2.1 Pre-processing image and extracting HOG feature, paragraph 2, “The HOG, a popular feature descriptor, was first used to calculate the occurrence of gradient direction in the local patch of an image. It captured edge and gradient structures, which are characteristic of local shapes, while it performed photometric and geometric transformations [14]. In this study, the image was denoted as I(x,y) with an image size of M x N, where and y were the pixel position. [acquire a numeric representation of each image;], x and y are being interpreted as number representation of the image]”)
However, Mao does not explicitly teach (2) for each experiment, recording at least one numeric virus titer readout of the virus- treated cell culture at the final time tfinal;
(4) training one or more machine learning models to make a prediction of a future virus titer on the numeric representations, wherein the training comprises minimizing an error between the model prediction of a future virus titer and a ground truth.
Masci further teaches (2) for each experiment, recording at least one numeric virus titer readout of the virus- treated cell culture at the final time tfinal; (Masci, pg. 271, Col. 2, paragraph 2, “Using the Celigo imager, the readout from the traditional assay can be reduced to 2 days post-infection [readout of the virus- treated cell culture at the final time tfinal;], since the Celigo imager can detect and count plaques [recording at least one numeric virus titer] that are harder to visualize by eye.” And pg. 272, Col. 1, paragraph 4, “With optimization of parameters, the Celigo can be used to recognize and count individual foci, thus aiding in the speed of the readout of the assay.”)
Mao and Masci are combinable for the same rationale as set forth above with respect to claim 1.
However, Mao and Masci do not explicitly teach (4) training one or more machine learning models to make a prediction of a future virus titer on the numeric representations, wherein the training comprises minimizing an error between the model prediction of a future virus titer and a ground truth.
Zhou further teaches (4) training one or more machine learning models to make a prediction of a future virus titer on the numeric representations, wherein the training comprises minimizing an error between the model prediction of a future virus titer and a ground truth. (Zhou, pg. 90941, Col. 2, paragraph 2, “With the exception of one team, all teams use DCNNs as part of the processing pipeline [provided by the one or more machine learning models]. For the first task, the best performing method achieves a quadratic-weighted Cohen's kappa score of k D 0:567, 95% CI [0.464, 0.671] between the predicted scores and the ground truth [comprises minimizing an error between a prediction of the final virus titer and a ground truth]. For the second task, the predictions of the top method have a Spearman's correlation coefficient of r D 0:617, 95% CI [0.581, 0.651] with the ground truth. [the at least one numeric virus titer readout of the virus-treated cell culture at the final time tfinal., second task being interpreted as the final time]”)
Mao, Masci, Neher, Ponomarev and Zhou are combinable for the same rationale as set forth above with respect to claim 9.
Claims 11 and 27 are rejected under 35 U.S.C. 103 as being unpatentable over Mao, in view of Masci, Neher, Ponomarev, Zhou as applied to respective claims 10 and 26 and in even further view of Logan et al (A flow cytometric granularity assay for the quantification of infectious virus, "Logan") published Apr. 16, 2019.
In regard to claim 11 and analogous claim 27, Mao, Masci, Neher, Ponomarev and Zhou teach the method of claim 10.
However, Mao, Masci, Neher, Ponomarev and Zhou do not explicitly teach further comprising a step of either (1) filtering out cells not infected by the virus, (2) filtering out dead cells, or (3) filtering out dead cells that did not die from a virus infection.
Logan teaches further comprising a step of either (1) filtering out cells not infected by the virus, (2) filtering out dead cells, or (3) filtering out dead cells that did not die from a virus infection. (Logan, pg. 7092, Col. 2, 2.7 Data visualization and analysis, paragraph 2, “To remove debris and dead cells, [filtering out dead cells] events with FSC less than 300 and 3 x 10^6 for each cytometer were removed from the data, respectively.”)
Mao, Masci, Neher, Ponomarev Zhou and Logan are related to the same field of endeavor (i.e. plaque assay). In view of the teachings of Logan, it would have been obvious for a person with ordinary skill in the art to apply the teachings of Logan to Mao, Masci, Neher, Ponomarev and Zhou before the effective filing date of the claimed invention in order to provide a simple and rapid process of assays. (Logan, pg. 7098, 4. Discussion, “The method presented here provides a simple, rapid, and high-throughput flow cytometry-based assay for the enumeration of infectious virus to support continuous process monitoring and process development of viral vaccines.”)
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 SKYLAR K VANWORMER whose telephone number is (703)756-1571. The examiner can normally be reached M-F 6:00am to 3:00 pm.
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, Usmaan Saeed can be reached on (571) 272-4046. 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.
/S.K.V./Examiner, Art Unit 2146 /USMAAN SAEED/Supervisory Patent Examiner, Art Unit 2146
Prosecution Insights