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 .
DETAILED ACTION
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.
Claims 1-6, 8-12, 14-27, 29-33, and 35-42 are rejected under 35 U.S.C. 103 as being unpatentable over Ferrantelli (Pub No. 20210118134) in view of Patias (Patias, P., Grivas, T.B., Kaspiris, A. et al. A review of the trunk surface metrics used as Scoliosis and other deformities evaluation indices. Scoliosis 5, 12 (2010). https://doi.org/10.1186/1748-7161-5-12 (Year: 2010)).
As per claim 1, Ferrantelli teaches the claimed:
1. A method of determining a degree of spinal curvature, the method being executed by a processor, the method comprising: for each segment of a plurality of segments of a torso of a subject: (Ferrantelli [0014]: “In another embodiment, there is provided non-transitory computer readable storage medium having stored therein a program to be executable by a processor for use of machine learning in computer-assisted anatomical prediction. The program causes the processor to execute identifying with a processor parameters in a plurality of training images to generate a training dataset, the training dataset having data linking the parameters to respective training images, training at least one machine learning algorithm based on the parameters in the training dataset and validating the trained machine learning algorithm, identifying with the processor digitized points on a plurality of anatomical landmarks in a radiographic image of a person displayed on a digital touch screen by determining linear anatomical dimensions of at least a portion of a body of the person in the displayed image using the validated machine learning algorithm and a scale factor for the displayed image, and making an anatomical prediction of the person based on the determined linear anatomical dimensions and a known morphological relationship.” Ferrantelli [0011]: “Disclosed embodiments use machine learning to provide several advantages including: (1) enhanced ability to compare a subject's spine to normal from digital x-ray, CT and/or MRI images of the spine and pelvis. In practice, these areas may fall under the following x-ray views: AP and lateral cervical in neutral position as well as with flexion and extension postures, AP Open Mouth Cervical, Nasium, Vertex, AP and lateral thoracic with flexion and extension, and AP and lateral lumbar films with or without flexion and extension postures, AP or PA Ferguson sacral base views along with Modified Ferguson projections as well as femur head x-ray views in order to obtaining true anatomical leg length inequality, AP and lateral full spine and flexion/extension stress views, (2) allowing a subject's x-rays to be available in and compared to a multitude of records in a master database, (3) enhanced ability to compare a subject's subluxations at each level to normal and the percentage loss or gain from normal calculated automatically measuring all segmental relative rotational angles and the global absolute rotational angles in degrees along with all translational distances in, for example, centimeters, millimeters or percentages of translation relative to the vertebrae below along with corresponding angular measurements both intersegmental and global angulations of all segments, (4) allowing for superimposition of normal and/or ideal spinal curves (using any color) and highlights George's line (subject's spinal position) on their digital x-rays, (5) applying and overlaying a normal and/or ideal spinal models on digital images, (6) comparing a subject's pre- and post-care films with percentage improvement at each level of their spine, or conversely their worsening if they had a traumatic event such as a motor vehicle crash, and (7) objectively measuring spinal instability crucial for calculating impairment ratings.”).
Ferrantelli alone does not explicitly teach the remaining claim limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
determining at least one volume of the segment;
and determining the degree of spinal curvature of the subject based on the volumes of the respective segments. (Patias pg. 13 col. 2: “ The ISIS system uses the Imbalance, Lateral Asymmetry and volumetric asymmetry Indices (Fig. 19) as indices in the Coronal plane, while the QSIS system uses a series of angles and distances (Fig. 20) for the same reason.” The volumetric asymmetry is using the volume of a portion of the spine to determine the angle, as the volume will change/increase if the segment deviates from its proper alignment. Patias pg. 13 col. 2-pg 14 co. 1: “Major deformity indices measured on the Transverse plane. 1. Angle of Trunk Rotation (ATR, or ATI - Angle of Trunk Inclination) [62], 2. CTAS [63, 64], 3. Torso centroid line, Principal axis orientation, Back surface rotation, Envelope indices, Half-area indices, Quarter-area indices [60], 4. Rib prominence, Flank prominence [27], 5. ISIS2 TA (Transverse Asymmetry index), VA (Volume Asymmetry index) and HS (Hump Severity index) [30, 33, 34], 6. Suzuki Hump Sum (SHS) [68], 7. Deformity in the Axial Plane Index (DAPI) [59], 8. QSIS indices in the Transverse plane [55], 9. Y1, ASY2, ASY3 [47] (see also SOSORT Conclusion 3.4 and 4).” This could be used for any segment of the spine. Patias teaches analyzing different portions of a spine. Volumetric asymmetry could be measured for different segments.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the determination of spinal curvature by volume of spine segments as taught by Patias with the system of Ferrantelli in order to use the deviation of a portion of a spine from its aligned position to determine the affect the curvature has on the 3D shape of the body.
As per claim 22, this claim is similar in scope to limitations recited in claim 1, and thus is rejected under the same rationale.
As per claim 2, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
2. The method of claim 1, wherein the at least one volume is a volume of a substantially planar segment oriented in a transverse plane of the subject. (Patias fig. 22: “Major deformity indices measured on the Transverse plane. 1. Angle of Trunk Rotation (ATR, or ATI - Angle of Trunk Inclination) [62], 2. CTAS [63, 64], 3. Torso centroid line, Principal axis orientation, Back surface rotation, Envelope indices, Half-area indices, Quarter-area indices [60], 4. Rib prominence, Flank prominence [27], 5. ISIS2 TA (Transverse Asymmetry index), VA (Volume Asymmetry index) and HS (Hump Severity index) [30, 33, 34], 6. Suzuki Hump Sum (SHS) [68], 7. Deformity in the Axial Plane Index (DAPI) [59], 8. QSIS indices in the Transverse plane [55], 9. Y1, ASY2, ASY3 [47] (see also SOSORT Conclusion 3.4 and 4).” Figs. 22-23 show transverse planes through a segment of the spine to compare it to the geometry of the plane.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the comparison of a segment of a curved spine to a transverse plane as taught by Patias with the system of Ferrantelli in order to compare the curvature to an established plane in a coordinate system to establish its curvature, especially one that deviates sideways from the midline of the body.
As per claims 18 and 23, these claims are similar in scope to limitations recited in claim 2, and thus are rejected under the same rationale.
As per claim 3, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
3. The method of claim 2, wherein the at least one volume includes four volumes of the segment in respective quadrants of the transverse plane. (Patias Fig. 3 shows quadrant of the plane. The segment intersecting with the transverse plane has volume in each, and the proportion of the volume in each quadrant shows the curvature.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the quadrants of a transverse plane to measure asymmetric of a spine section as taught by Patias with the system of Ferrantelli in order to measure asymmetry of a spine from a midline in an established coordinate system.
As per claim 24, this claim is similar in scope to limitations recited in claim 3, and thus is rejected under the same rationale.
As per claim 4, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
4. The method of claim 3, wherein the respective quadrants are defined relative to a center of a segment located inferior to a spine of the subject. (Patias fig. 11 depicts a coordinate system with the origin inferior to the spine. The X-Z plane of that system is the transverse plane. The pelvis area can be a segment of the torso.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the center of a transverse plane defined relative to the bottom of a spine as taught by Patias with the system of Ferrantelli in order to measure a whole curved spine relative to a coordinate system whose origin is below it see how the spine segments deviate sideways.
As per claims 20, 25, and 41, these claims are similar in scope to limitations recited in claim 4, and thus are rejected under the same rationale.
As per claim 5, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
5. The method of claim 4, wherein determining the degree of spinal curvature comprises determining a Cobb angle. (Patias “The Cobb Angle”: “The degree of curvature in the coronal plane is radiographically measured according to the method of Cobb [16]. The Cobb angle, which is considered the golden standard, is the angle between lines drawn along the upper end plate of the most tilted vertebrae above the curve's apex and the lower end plate of the most tilted vertebrae below the apex. While Cobb angle is the accepted standard for measuring scoliosis on radiographs [17, 18], it has some important limitations [10, 18]:”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the Cobb angle to measure spine curvature as taught by Patias with the system of Ferrantelli in order to use a standard measure of scoliosis severity.
As per claim 26, this claim is similar in scope to limitations recited in claim 5, and thus is rejected under the same rationale.
As per claim 6, Ferrantelli teaches the claimed:
6. The method of claim 5, wherein determining the degree of spinal curvature includes using a convolutional neural network. (Ferrantelli teaches determining the rotational angles of spine segments. Ferrantelli [0011]: “Disclosed embodiments use machine learning to provide several advantages including: (1) enhanced ability to compare a subject's spine to normal from digital x-ray, CT and/or MRI images of the spine and pelvis. In practice, these areas may fall under the following x-ray views: AP and lateral cervical in neutral position as well as with flexion and extension postures, AP Open Mouth Cervical, Nasium, Vertex, AP and lateral thoracic with flexion and extension, and AP and lateral lumbar films with or without flexion and extension postures, AP or PA Ferguson sacral base views along with Modified Ferguson projections as well as femur head x-ray views in order to obtaining true anatomical leg length inequality, AP and lateral full spine and flexion/extension stress views, (2) allowing a subject's x-rays to be available in and compared to a multitude of records in a master database, (3) enhanced ability to compare a subject's subluxations at each level to normal and the percentage loss or gain from normal calculated automatically measuring all segmental relative rotational angles and the global absolute rotational angles in degrees along with all translational distances in, for example, centimeters, millimeters or percentages of translation relative to the vertebrae below along with corresponding angular measurements both intersegmental and global angulations of all segments, (4) allowing for superimposition of normal and/or ideal spinal curves (using any color) and highlights George's line (subject's spinal position) on their digital x-rays, (5) applying and overlaying a normal and/or ideal spinal models on digital images, (6) comparing a subject's pre- and post-care films with percentage improvement at each level of their spine, or conversely their worsening if they had a traumatic event such as a motor vehicle crash, and (7) objectively measuring spinal instability crucial for calculating impairment ratings.” Ferrantelli teaches using a convolutional model to measure this. Ferrantelli [0108]: “In a CNN-based model, a set of filters are used to extract features from images using convolution operation. Training of the CNN is done using a training dataset containing images and landmark points, which determines the trained values of the parameters/weights of the neural network. FIG. 13 depicts a CNN architecture for learning landmark points. As seen in FIG. 13, the CNN includes multiple layers. A convolutional layer may include 8 128×128 kernels feeding into 2×2 pooling-layer. The pooling layer then feeds into another convolutional layer including 24, 48×48 kernels feeding into 2×2 pooling-layer. Further layers include fully-connected layers 1×256.”).
As per claims 12, 27 and 33, these claims are similar in scope to limitations recited in claim 6, and thus are rejected under the same rationale.
As per claim 8, Ferrantelli teaches the claimed:
8. The method of claim 7, further comprising determining the plurality of segments, wherein said determining the plurality of segments comprises: obtaining a plurality of images of the torso of the subject; determining a shape of the torso based on the plurality of images; and dividing the shape of the torso into the plurality of segments. (Ferrantelli [0096]: “How effectively a machine learning algorithm can be trained may be related to how well the data is classified or labeled before it is used in a training procedure. Classifiers play an important role in the analysis of radiographic images of the human body. In embodiments, classifiers are used to classify the body dimensions such as, for example, body features, lengths, widths, etc., based on the relevant extracted body portions from the images. To develop a procedure for identifying images or videos as belonging to particular classes or categories (or for any classification or pattern recognition task), supervised learning technology may be based on decision trees, on logical rules, or on other mathematical techniques such as linear discriminant methods (including perceptrons, support vector machines, and related variants), nearest neighbor methods, Bayesian inference, neural networks, etc.” Ferrantelli [0011]: “Disclosed embodiments use machine learning to provide several advantages including: (1) enhanced ability to compare a subject's spine to normal from digital x-ray, CT and/or MRI images of the spine and pelvis. In practice, these areas may fall under the following x-ray views: AP and lateral cervical in neutral position as well as with flexion and extension postures, AP Open Mouth Cervical, Nasium, Vertex, AP and lateral thoracic with flexion and extension, and AP and lateral lumbar films with or without flexion and extension postures, AP or PA Ferguson sacral base views along with Modified Ferguson projections as well as femur head x-ray views in order to obtaining true anatomical leg length inequality, AP and lateral full spine and flexion/extension stress views, (2) allowing a subject's x-rays to be available in and compared to a multitude of records in a master database, (3) enhanced ability to compare a subject's subluxations at each level to normal and the percentage loss or gain from normal calculated automatically measuring all segmental relative rotational angles and the global absolute rotational angles in degrees along with all translational distances in, for example, centimeters, millimeters or percentages of translation relative to the vertebrae below along with corresponding angular measurements both intersegmental and global angulations of all segments, (4) allowing for superimposition of normal and/or ideal spinal curves (using any color) and highlights George's line (subject's spinal position) on their digital x-rays, (5) applying and overlaying a normal and/or ideal spinal models on digital images, (6) comparing a subject's pre- and post-care films with percentage improvement at each level of their spine, or conversely their worsening if they had a traumatic event such as a motor vehicle crash, and (7) objectively measuring spinal instability crucial for calculating impairment ratings.”)
As per claims 29 and 35, these claims are similar in scope to limitations recited in claim 8, and thus are rejected under the same rationale.
As per claim 9, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
9. The method of claim 8, wherein at least some of the plurality of images contain at least one reference marker of a plurality of reference markers, and determining the shape of the torso includes determining a position of at least one point on the torso with respect to the plurality of reference markers (Ferrantelli teaches to apply reference points to the training images of spines being analyzed. Ferrantelli [0108]: “In a CNN-based model, a set of filters are used to extract features from images using convolution operation. Training of the CNN is done using a training dataset containing images and landmark points, which determines the trained values of the parameters/weights of the neural network. FIG. 13 depicts a CNN architecture for learning landmark points. As seen in FIG. 13, the CNN includes multiple layers. A convolutional layer may include 8 128×128 kernels feeding into 2×2 pooling-layer. The pooling layer then feeds into another convolutional layer including 24, 48×48 kernels feeding into 2×2 pooling-layer. Further layers include fully-connected layers 1×256.” Likewise, Patias fig. 13 shows a torse with several reference markers. Beside it is a mesh of the shape formed by the markers. Ferrantelli in the rejection to claim 8 above teaches the plurality of images used to determine segments of the spine.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the reference markers to establish a shape of a torso as taught by Patias with the system of Ferrantelli in order to represent the shape of a torso through a simple and computer-readable structure and allow them to be analyzed in the same way the training images are analyzed.
As per claims 15, 30, and 36, these claims are similar in scope to limitations recited in claim 9, and thus are rejected under the same rationale.
As per claim 10, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
10. The method of claim 9, wherein determining the shape of the torso comprises generating a three-dimensional (3D) model of the torso. (Patias fig. 13 shows a primitive 3D mesh of points on the spine. Fig. 25 shows a more complex 3D model of the spine.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the generation of a 3D model of the torso as taught by Patias with the system of Ferrantelli in order to model a torso and emphasize the most important details visually without using photorealistic detail.
As per claims 17, 31, and 38, these claims are similar in scope to limitations recited in claim 10, and thus are rejected under the same rationale.
As per claim 11, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
11. A method of determining a degree of spinal curvature, the method being executed by a processor, the method comprising:
receiving a plurality of images of at least a torso of a subject;
determining a plurality of segments of the torso; (Ferrantalli [0011]: “Disclosed embodiments use machine learning to provide several advantages including: (1) enhanced ability to compare a subject's spine to normal from digital x-ray, CT and/or MRI images of the spine and pelvis. In practice, these areas may fall under the following x-ray views: AP and lateral cervical in neutral position as well as with flexion and extension postures, AP Open Mouth Cervical, Nasium, Vertex, AP and lateral thoracic with flexion and extension, and AP and lateral lumbar films with or without flexion and extension postures, AP or PA Ferguson sacral base views along with Modified Ferguson projections as well as femur head x-ray views in order to obtaining true anatomical leg length inequality, AP and lateral full spine and flexion/extension stress views, (2) allowing a subject's x-rays to be available in and compared to a multitude of records in a master database, (3) enhanced ability to compare a subject's subluxations at each level to normal and the percentage loss or gain from normal calculated automatically measuring all segmental relative rotational angles and the global absolute rotational angles in degrees along with all translational distances in, for example, centimeters, millimeters or percentages of translation relative to the vertebrae below along with corresponding angular measurements both intersegmental and global angulations of all segments, (4) allowing for superimposition of normal and/or ideal spinal curves (using any color) and highlights George's line (subject's spinal position) on their digital x-rays, (5) applying and overlaying a normal and/or ideal spinal models on digital images, (6) comparing a subject's pre- and post-care films with percentage improvement at each level of their spine, or conversely their worsening if they had a traumatic event such as a motor vehicle crash, and (7) objectively measuring spinal instability crucial for calculating impairment ratings.”).
determining respective volumes of the plurality of segments of the torso based on the plurality of images;
determining the degree of spinal curvature of the subject based on respective shares of the respective volumes of the segments; (Patias pg. 13 column 2: “ The ISIS system uses the Imbalance, Lateral Asymmetry and volumetric asymmetry Indices (Fig. 19) as indices in the Coronal plane, while the QSIS system uses a series of angles and distances (Fig. 20) for the same reason.”).
and outputting the determined degree of spinal curvature. (Ferrantelli fig. 10 shows outputting an angle of a spine in an image.).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the measure of volumetric asymmetry to measure spinal curvature as taught by Patias with the system of Ferrantelli in order to analyze a spine by observing its increase volume being displaced and being in a different alignment that increase the space it occupies.
As per claim 32, this claim is similar in scope to limitations recited in claim 11, and thus is rejected under the same rationale.
As per claim 16, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Patias teaches the claimed:
16. The method of 15, wherein determining the position of the at least one point on the torso includes generating a depth map of a plurality of points on the torso. (Patias fig. 9 shows a 3D model of a back with points mapped onto it. The model shows depth of different portions of the spine to show curvature.).
As per claim 37, this claim is similar in scope to limitations recited in claim 16, and thus is rejected under the same rationale.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the depth map based on points assigned to a spine model as taught by Patias with the system of Ferrantelli in order to analyze a spine by assignment points to portions of a spine and show its misaligned shape in 3 dimensions including depth.
As per claim 21, Ferrantelli teaches the claimed:
21. The method of claim 20, wherein the plurality of images include at least one video recording. (Ferrantelli [0098]: “To apply standard approaches to supervised learning, the media segments (body dimensions) in both the training set and the test set must be represented in terms of numbers derived from the images, i.e., features. The relationship between features extracted for the purposes of supervised learning and the content of an image or video has an important impact on the success of the approach.”).
As per claim 42, this claim is similar in scope to limitations recited in claim 16, and thus is rejected under the same rationale.
Claims 7, 13, 28, 34 are rejected under 35 U.S.C. 103 as being unpatentable over Ferrantelli in view of Patias and further in view of Prasad (Pub No. US 20210327063 A1) and further in view of Li (WO 2020257592 A1).
As per claim 7, Ferrantelli alone does not explicitly teach the claimed limitations.
However, Ferrantelli in combination with Prasad and Li teaches the claimed:
7. The method of claim 6, wherein the convolutional neural network includes two convolutional layers, a flatten layer, and two dense layers, using an Adam optimizer. (Ferrantelli [0107]-[0109]: “[0107] The neural network may be a deep convolutional neural network. The neural network may be a deep neural network that comprises an output layer and one or more hidden layers. In embodiments, training the neural network may include: training the output layer by minimizing a loss function given the optimal set of assignments, and training the hidden layers through a back propagation algorithm. The deep neural network may be a Convolutional Neural Network (CNN).
[0108] In a CNN-based model, a set of filters are used to extract features from images using convolution operation. Training of the CNN is done using a training dataset containing images and landmark points, which determines the trained values of the parameters/weights of the neural network. FIG. 13 depicts a CNN architecture for learning landmark points. As seen in FIG. 13, the CNN includes multiple layers. A convolutional layer may include 8 128×128 kernels feeding into 2×2 pooling-layer. The pooling layer then feeds into another convolutional layer including 24, 48×48 kernels feeding into 2×2 pooling-layer. Further layers include fully-connected layers 1×256.
[0109] In some CNN models, the numbers of the CNN layers and fully connected layers may vary. In some network architectures, residual pass or feedbacks may be used to avoid a conventional problem of gradient vanishing in training the network weights. The network may be built using any suitable computer language such as, for example, Python or C++. Deep learning toolboxes such as TensorFlow, Caffe, Keras, Torch, Theano, CoreML, and the like, may be used in implementing the network. These toolboxes are used for training the weights and parameters of the network. In some embodiments, custom-made implementation of CNN and deep learning algorithms on special computers with Graphical Processing Units (GPUs) are used for training, inference, or both. The inference is referred to as the stage in which a trained model is used to infer/predict the testing samples. The weights of a trained model are stored in a computer disk and then used for inference. Different optimizers such as the Adam optimization algorithm, and gradient descent may be used for training the weights and parameters of the networks. In training the networks, hyperparameters may be tuned to achieve higher recognition and detection accuracies. In the training phase, the network may be exposed to the training data through several epochs. An epoch is defined as an entire dataset being passed only once both forward and backward through the neural network.” Ferrantelli teaches multiple layers. Prasad, likewise, teaches analyzing the curvature of a spine. Prasad [0003]: “The spine is the most complex anatomical structure in the entire human body. It is made up of twenty-six irregular bones connected in such a way that flexible curved structure results. The vertebral column is about 70 centimeters long in an average adult and features seven major divisions. Seven vertebrae found in the neck region constitute the cervical vertebrae, the next twelve comprise the thoracic vertebrae, and the five vertebrae supporting the lower back are the lumbar vertebrae. The sacrum, which is inferior to these vertebrae, articulates with the hip bones of the pelvis. The tiny coccyx terminates the entire vertebral column. Intervertebral discs act as shock absorbers and allow the spine to extend. These are the thickest in the lumbar and cervical regions, enhancing the flexibility in these regions. The degeneration of intervertebral discs is a relatively common phenomenon, with aging due to wear and tear, and is the major cause of back pain. Degenerative lumbar spine disease includes spondylosis (arthritic) and degenerative disc disease of the lumbar spine with or without neuronal compression or spinal instability.” Prasad teaches a dense and a flatten layer to a neural network. Prasad [0046]: “Output of the regressor 445 is down sampled at down sampling layer 455, followed by a dense and flatten layer 456 and batch normalization layers 457 and 458. Collectively, the down sampling layer 455, dense and flatten layer 456, and batch normalization layers 457, 458 comprise aggregation and spatial transformation layers 459. Aggregation at 459 consists of final geodesic displacement vectors that are corrected for spine regions alone. Additionally, the spatial transformations of each of the displacement vectors of the functional volume are stored in a matrix for mapping between the resampled output of both functional and non-functional volumes. The aggregation and re-sampling takes place simultaneously.” Li teaches using two dense layers for a neural network analyzing anatomical structures, along with many other varying layers. Li abstract: “A computer-executed method implementing a deep learning technique is carried out to perform on canine thoracic radiographic images an automated diagnosis of left atrial enlargement as an early sign of myxomatous mitral valve insufficiency.” Li [0026]: “The VGG framework was applied using the deep learning package Keras (version 2.3.0). The model structure was composed of 13 convolution layers, 5 pooling layers, 1 drop out layer, and 2 dense layers, with a total of 7,861 ,032 parameters. The following model parameters were used for training: 32 batch size, 100 epochs, and 0.0001 learning rate. The same padding technique was used in the model to improve the use of pixels on the edge of the image. The Adam optimization algorithm and 0.01 kernel regularizer were used.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the dense and flatten layers in a deep learning network as taught by Prasad with the system of Ferrantelli in order to incorporate those types of layers into the neural network of Ferrantelli.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use 2 dense layers together as taught by Li with the system of Ferrantelli in order to have a hidden layer and an output layer among to two layers.
As per claims 13, 28 and 34, these claims are similar in scope to limitations recited in claim 7, and thus are rejected under the same rationale.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to THOMAS JOHN FOSTER whose telephone number is (571)272-5053. The examiner can normally be reached Mon, Fri 8:30-6. Tues-Thurs 7:30-5.
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, Daniel Hajnik can be reached at 571-272-7642. 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.
/THOMAS JOHN FOSTER/Examiner, Art Unit 2616
/HAI TAO SUN/Primary Examiner, Art Unit 2616
Prosecution Insights