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 .
Status of the Claims
Claims 1-20, as originally filed, are currently pending and have been considered below.
Claim Rejections - 35 USC § 102
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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 1-6, 8, 9 and 14-20 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Yan, Ke, et al. "SAM: Self-supervised Learning of Pixel-wise Anatomical Embeddings in Radiological Images." arXiv preprint arXiv:2012.02383 (2020), hereinafter, “Yan”.
As per claim 1, Yan discloses a method, comprising:
determining, by a system comprising a processor (Yan, pages 5-6, D. Application: computed using the convolutional operation on GPU),
positions of target features within a target medical image of an anatomical region of a subject based on reference spatial relationships between the target features as defined in reference spatial information, and based on matching, by the system, reference pixel features respectively associated with the target features with corresponding subsets of pixel features of the target medical image (Yan, Abstract, Self-supervised Anatomical eMbedding (SAM) … Using SAM, one can label any point of interest on a template image and then locate the same body part in other images by simple nearest neighbor searching; Yan, pages 1-2, I. Introduction, develop a universal algorithm that learns from unlabeled radiological images to detect arbitrary points of interest. It can generate embeddings on each image pixel to encode its anatomical context information, so that the same body part in different images express similar embeddings and can be retrieved by simple nearest neighbor searching ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features),
wherein the reference pixel features comprise template image pixel features extracted from labeled versions of the target features as included in a template medical image depicting the anatomical region of a reference subject (Yan, pages 1-2, I. Introduction, SAM can be trained easily and applied to various tasks. A natural application is “one-shot” landmark detection with one labeled template image);
generating, by the system, label information for the target features identifying the target features and their positions (Yan, page 5, Fig. 3: The inference procedure of SAM for anatomical point matching. Red dots are the point of interest or matched point; Yan, pages 5-6, D. Application: Anatomical Point Matching, Fig. 3 depicts the inference procedure of SAM for anatomical point matching. To locate a certain point of interest, we first need to label it on a template image (also known as an atlas [47] or a support image [32]). Then, given an unlabeled query image, we compute the global and local embedding tensors. After extracting the anchor embedding vectors from the point of interest of the template, we compute the similarity maps Sg, Sl between the anchor vectors and the query tensors … Finally, we upsample Sg, Sl to the size of the original image, and find the peak of Sg + Sl as the matched anatomical point); and
associating, by the system, the label information with the target medical image (Yan, page 5, Fig. 3: The inference procedure of SAM for anatomical point matching. Red dots are the point of interest or matched point; Yan, pages 5-6, D. Application: Anatomical Point Matching, Fig. 3 depicts the inference procedure of SAM for anatomical point matching. To locate a certain point of interest, we first need to label it on a template image (also known as an atlas [47] or a support image [32]). Then, given an unlabeled query image, we compute the global and local embedding tensors. After extracting the anchor embedding vectors from the point of interest of the template, we compute the similarity maps Sg, Sl between the anchor vectors and the query tensors … Finally, we upsample Sg, Sl to the size of the original image, and find the peak of Sg + Sl as the matched anatomical point).
As per claim 2, Yan discloses the method of claim 1, wherein the pixel features of the target image and the reference pixel features respectively comprise extracted pixel features respectively extracted from the target medical image and the template medical image using one or more feature extraction models (Yan, page 1, Figure 1: Self-supervised anatomical embedding (SAM) and its application of anatomical location matching; Inference, Template image, Arbitrary input point, Embedding, extraction and matching; Query image 1, Query image 2; Yan, page 3, A. Coarse-to-Fine Network Architecture, As shown in Fig. 2, in training, we first take an unlabeled CT volume V , then crop two 3D patches with random location and size, and resize them to the same shape ... The patches are sent to a fully-convolutional network to extract pixel-wise embeddings; Yan, page 11, Table VII: Comparison of different feature extraction algorithms).
As per claim 3, Yan discloses the method of claim 1, further comprising: performing, by the system, the determining, the generating, and the associating, for a plurality of different target medical images depicting the anatomical region of respective different subjects, wherein the performing the determining for the plurality of the different target medical images comprises employing the reference pixel features (Yan, page 1, Figure 1, depicting multiple query images from one template image; Yan, page 11, A. Universal Anatomical Point Matching in CT, To show that SAM can be used to detect arbitrary anatomical locations, we randomly select a point in a template CT image, and then use SAM to find its matched point in a query image from another patient. Examples are demonstrated in Fig. 9. SAM can accurately find the matched anatomical location in the query image despite significant inter-subject variability, organ deformation, and contrast changes).
As per claim 4, Yan discloses the method of claim 1, wherein the reference pixel features respectively represent the target features and one or more adjacent features to the target features (Yan, page 3, A. Coarse-to-Fine Network Architecture, As shown in Fig. 2, in training, we first take an unlabeled CT volume V, then crop two 3D patches with random location and size, and resize them to the same shape ... The patches are sent to a fully-convolutional network to extract pixel-wise embeddings. To learn universal anatomical embeddings, on the one hand, SAM needs to memorize the 3D contextual appearance of each body part so its embedding is distinguishable from all other body parts. On the other hand, it needs to encode local information to differentiate adjacent structures).
As per claim 5, Yan discloses the method of claim 1, wherein the determining comprises determining the positions of respective ones of the target features individually and comprises determining a position of a current target feature based on a previously determined position of a previously localized target feature within the target medical image (Yan, page 7, C. Qualitative Results, As shown in Fig. 5, SAM is able to accurately detect various landmarks in different body parts with only one template image ... For X-ray images, SAM can locate the landmarks in the presence of body rotation, deformation, and metal prostheses ... SAM can match a variety of lesions effectively in follow-up CTs, see Fig. 6. In (a), it locates a tiny kidney lesion despite the different noise levels in two images. In (b), SAM detects the lesion in “follow-up 2” even though the shape of its surrounding organ has changed. In (c), the algorithm successfully differentiates the true matched lesion from an adjacent one in the dashed yellow circle. (d) is an interesting case where the lesion disappeared in “follow-up 3”, possibly due to surgery. SAM can locate the correct position even if the lesion no longer exists, showing that it did not simply match the lesion texture, but has learned the body part from the anatomical context).
As per claim 6, Yan discloses the method of claim 5, wherein determining the positions comprises determining the positions of the respective ones of the target features individually in accordance with an order tailored based on the anatomical region (Yan, page 7, C. Qualitative Results, As shown in Fig. 5, SAM is able to accurately detect various landmarks in different body parts with only one template image ... For X-ray images, SAM can locate the landmarks in the presence of body rotation, deformation, and metal prostheses ... SAM can match a variety of lesions effectively in follow-up CTs, see Fig. 6. In (a), it locates a tiny kidney lesion despite the different noise levels in two images. In (b), SAM detects the lesion in “follow-up 2” even though the shape of its surrounding organ has changed. In (c), the algorithm successfully differentiates the true matched lesion from an adjacent one in the dashed yellow circle. (d) is an interesting case where the lesion disappeared in “follow-up 3”, possibly due to surgery. SAM can locate the correct position even if the lesion no longer exists, showing that it did not simply match the lesion texture, but has learned the body part from the anatomical context).
As per claim 8, Yan discloses the method of claim 5, wherein determining the positions of the respective ones of the target features individually comprises restricting, by the system, respective regions of the pixel features of the target medical image searched in association with performing the matching based on respective positions of previously localized target features within the target medical image (Yan, pages 1-2, I. Introduction, we propose Self-supervised Anatomical eMbedding (SAM) ... We ... propose a pixel-level contrastive learning framework to differentiate pixels of varying body parts ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features).
As per claim 9, Yan discloses the method of claim 8, wherein the restricting comprises excluding pixel regions of the target medical image corresponding to the respective positions of the previously localized target features from the respective regions (Yan, pages 1-2, I. Introduction, we propose Self-supervised Anatomical eMbedding (SAM) ... We ... propose a pixel-level contrastive learning framework to differentiate pixels of varying body parts ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features).
As per claim 14, Yan discloses a system, comprising:
at least one memory that stores computer-executable components; and at least one processor that executes the computer-executable components stored in the at least one memory (Yan, pages 5-6, D. Application: computed using the convolutional operation on GPU), wherein the computer-executable components comprise:
a feature localization component that determines positions of target features within a target medical image of an anatomical region of a subject based on reference spatial relationships between the target features as defined in reference spatial information, and based on matching reference pixel features respectively associated with the target features with corresponding subsets of pixel features of the target medical image (Yan, Abstract, Self-supervised Anatomical eMbedding (SAM) … Using SAM, one can label any point of interest on a template image and then locate the same body part in other images by simple nearest neighbor searching; Yan, pages 1-2, I. Introduction, develop a universal algorithm that learns from unlabeled radiological images to detect arbitrary points of interest. It can generate embeddings on each image pixel to encode its anatomical context information, so that the same body part in different images express similar embeddings and can be retrieved by simple nearest neighbor searching ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features),
wherein the reference pixel features comprise template image pixel features extracted from labeled versions of the target features as included in a template medical image depicting a corresponding anatomical region of a reference subject (Yan, pages 1-2, I. Introduction, SAM can be trained easily and applied to various tasks. A natural application is “one-shot” landmark detection with one labeled template image); and
a labeling component that generates label information for the target features identifying the target features and their positions (Yan, page 5, Fig. 3: The inference procedure of SAM for anatomical point matching. Red dots are the point of interest or matched point; Yan, pages 5-6, D. Application: Anatomical Point Matching, Fig. 3 depicts the inference procedure of SAM for anatomical point matching. To locate a certain point of interest, we first need to label it on a template image (also known as an atlas [47] or a support image [32]). Then, given an unlabeled query image, we compute the global and local embedding tensors. After extracting the anchor embedding vectors from the point of interest of the template, we compute the similarity maps Sg, Sl between the anchor vectors and the query tensors … Finally, we upsample Sg, Sl to the size of the original image, and find the peak of Sg + Sl as the matched anatomical point) and
associates the label information with the target medical image (Yan, page 5, Fig. 3: The inference procedure of SAM for anatomical point matching. Red dots are the point of interest or matched point; Yan, pages 5-6, D. Application: Anatomical Point Matching, Fig. 3 depicts the inference procedure of SAM for anatomical point matching. To locate a certain point of interest, we first need to label it on a template image (also known as an atlas [47] or a support image [32]). Then, given an unlabeled query image, we compute the global and local embedding tensors. After extracting the anchor embedding vectors from the point of interest of the template, we compute the similarity maps Sg, Sl between the anchor vectors and the query tensors … Finally, we upsample Sg, Sl to the size of the original image, and find the peak of Sg + Sl as the matched anatomical point).
As per claim 15, Yan discloses the system of claim 14, wherein the pixel features of the target image and the reference pixel features respectively comprise extracted pixel features respectively extracted from the target medical image and the template medical image using one or more feature extraction models (Yan, page 1, Figure 1: Self-supervised anatomical embedding (SAM) and its application of anatomical location matching; Inference, Template image, Arbitrary input point, Embedding, extraction and matching; Query image 1, Query image 2; Yan, page 3, A. Coarse-to-Fine Network Architecture, As shown in Fig. 2, in training, we first take an unlabeled CT volume V , then crop two 3D patches with random location and size, and resize them to the same shape ... The patches are sent to a fully-convolutional network to extract pixel-wise embeddings; Yan, page 11, Table VII: Comparison of different feature extraction algorithms).
As per claim 16, Yan discloses the system of claim 14, wherein the feature localization component further determines the positions of the target features within additional target medical images corresponding to the target medical image based on the reference spatial relationships and based on performing the matching for the additional target medical images using the reference pixel features, wherein the additional target medical images respectively depict the anatomical region of different subjects (Yan, page 1, Figure 1, depicting multiple query images from one template image; Yan, page 11, A. Universal Anatomical Point Matching in CT, To show that SAM can be used to detect arbitrary anatomical locations, we randomly select a point in a template CT image, and then use SAM to find its matched point in a query image from another patient. Examples are demonstrated in Fig. 9. SAM can accurately find the matched anatomical location in the query image despite significant inter-subject variability, organ deformation, and contrast changes).
As per claim 17, Yan discloses the system of claim 14, wherein the feature localization component determines the positions of respective ones of the target features individually and determines a position of a current target feature based on a previously determined position of a previously localized target feature within the target medical image (Yan, page 7, C. Qualitative Results, As shown in Fig. 5, SAM is able to accurately detect various landmarks in different body parts with only one template image ... For X-ray images, SAM can locate the landmarks in the presence of body rotation, deformation, and metal prostheses ... SAM can match a variety of lesions effectively in follow-up CTs, see Fig. 6. In (a), it locates a tiny kidney lesion despite the different noise levels in two images. In (b), SAM detects the lesion in “follow-up 2” even though the shape of its surrounding organ has changed. In (c), the algorithm successfully differentiates the true matched lesion from an adjacent one in the dashed yellow circle. (d) is an interesting case where the lesion disappeared in “follow-up 3”, possibly due to surgery. SAM can locate the correct position even if the lesion no longer exists, showing that it did not simply match the lesion texture, but has learned the body part from the anatomical context).
As per claim 18, Yan discloses the system of claim 17, wherein the feature localization component restricts respective regions of the pixel features of the target medical image searched in association with performing the matching based on respective positions of previously localized target features within the target image (Yan, pages 1-2, I. Introduction, we propose Self-supervised Anatomical eMbedding (SAM) ... We ... propose a pixel-level contrastive learning framework to differentiate pixels of varying body parts ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features).
As per claim 19, Yan discloses the system of claim 17, wherein the feature localization component restricts respective regions of the pixel features of the target image searched in association with performing the matching based on respective positions of previously localized target features within the target medical image and the reference spatial relationships (Yan, pages 1-2, I. Introduction, we propose Self-supervised Anatomical eMbedding (SAM) ... We ... propose a pixel-level contrastive learning framework to differentiate pixels of varying body parts ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features).
As per claim 20, Yan discloses a non-transitory machine-readable storage medium, comprising executable instructions that, when executed by a processor (Yan, pages 5-6, D. Application: computed using the convolutional operation on GPU), facilitate performance of operations, comprising:
determining positions of target features within a target medical image of an anatomical region of a subject based on reference spatial relationships between the target features as defined in reference spatial information, and based on matching, by the system, reference pixel features respectively associated with the target features with corresponding subsets of pixel features of the target medical image (Yan, Abstract, Self-supervised Anatomical eMbedding (SAM) … Using SAM, one can label any point of interest on a template image and then locate the same body part in other images by simple nearest neighbor searching; Yan, pages 1-2, I. Introduction, develop a universal algorithm that learns from unlabeled radiological images to detect arbitrary points of interest. It can generate embeddings on each image pixel to encode its anatomical context information, so that the same body part in different images express similar embeddings and can be retrieved by simple nearest neighbor searching ... To cover both global and local information, we design a coarse-to-fine architecture with two-level embeddings. The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features),
wherein the reference pixel features comprise template image pixel features extracted from labeled versions of the target features as included in a template medical image depicting the anatomical region of a reference subject (Yan, pages 1-2, I. Introduction, SAM can be trained easily and applied to various tasks. A natural application is “one-shot” landmark detection with one labeled template image);
generating label information for the target features identifying the target features and their positions (Yan, page 5, Fig. 3: The inference procedure of SAM for anatomical point matching. Red dots are the point of interest or matched point; Yan, pages 5-6, D. Application: Anatomical Point Matching, Fig. 3 depicts the inference procedure of SAM for anatomical point matching. To locate a certain point of interest, we first need to label it on a template image (also known as an atlas [47] or a support image [32]). Then, given an unlabeled query image, we compute the global and local embedding tensors. After extracting the anchor embedding vectors from the point of interest of the template, we compute the similarity maps Sg, Sl between the anchor vectors and the query tensors … Finally, we upsample Sg, Sl to the size of the original image, and find the peak of Sg + Sl as the matched anatomical point); and
associating the label information with the target medical image (Yan, page 5, Fig. 3: The inference procedure of SAM for anatomical point matching. Red dots are the point of interest or matched point; Yan, pages 5-6, D. Application: Anatomical Point Matching, Fig. 3 depicts the inference procedure of SAM for anatomical point matching. To locate a certain point of interest, we first need to label it on a template image (also known as an atlas [47] or a support image [32]). Then, given an unlabeled query image, we compute the global and local embedding tensors. After extracting the anchor embedding vectors from the point of interest of the template, we compute the similarity maps Sg, Sl between the anchor vectors and the query tensors … Finally, we upsample Sg, Sl to the size of the original image, and find the peak of Sg + Sl as the matched anatomical point).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 7 and 10-13 is/are rejected under 35 U.S.C. 103 as being unpatentable over Yan, Ke, et al. "SAM: Self-supervised Learning of Pixel-wise Anatomical Embeddings in Radiological Images." arXiv preprint arXiv:2012.02383 (2020), hereinafter, “Yan” as applied to claims 6 and 8 above, and further in view of Ullmann, Eugénie, et al. "Automatic labeling of vertebral levels using a robust template‐based approach." International journal of biomedical imaging 2014.1 (2014): 719520, hereinafter, “Ullmann”.
As per claim 7, Yan discloses the method of claim 6, and (Yan, page 3, A. Coarse-to-Fine Network Architecture, To learn universal anatomical embeddings, on the one hand, SAM needs to memorize the 3D contextual appearance of each body part so its embedding is distinguishable from all other body parts … it needs to encode local information to differentiate adjacent structures with similar appearance for accurate localization) but does not explicitly disclose the following limitations as further recited however Ullmann discloses wherein the anatomical region comprises a region of a spine, wherein the target features comprise vertebrae included in the region of the spine, wherein the order corresponds to a natural sequential order of the vertebrae such that the current target feature and the previously localized target feature comprise adjacent vertebrae, and wherein reference spatial relationships comprise spatial relationships between pairs of adjacent vertebrae (Ullmann, Abstract, The robustness of the disk detection is improved by using a template of vertebral distance; Ullmann, page 3, 2.4.3. Detection of Disks. The algorithm detects disks one after another by analyzing the spine intensity profile towards the caudal direction. The starting point is the C2-C3 disk (or the disk indicated by the user). For each disk to be detected, its probabilistic location is first estimated using the template of intervertebral disk distances ... all disks locations are found and then projected back to the spinal cord centerline).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Ullmann and Yan because they are in the same field of endeavor. One skilled in the art would have been motivated to include the vertebrae template of Ullmann in the system of Yan as one of the memorized and recognized body parts of Yan (Yan, page 3, A. Coarse-to-Fine Network Architecture).
As per claim 10, Yan discloses the method of claim 8, and (Yan, pages 1-2, I. Introduction, we propose Self-supervised Anatomical eMbedding (SAM) ... We ... propose a pixel-level contrastive learning framework to differentiate pixels of varying body parts ... The global embedding is trained to distinguish every body part on a coarse scale, helping the local embedding to focus on a smaller region to differentiate with finer features) but does not explicitly disclose the following limitation as further recited however Ullmann discloses wherein the restricting comprises defining, by the system, a search region for the current target feature based on a reference spatial relationship between the current target feature and the previously localized target feature, and constraining, by the system, the pixel features of the target imaged searched to a portion of the pixel features included in the search region in association with matching a subset of the reference pixel features associated with the current target feature to a corresponding subset of the pixel features of the target medical image (Ullmann, Abstract, The robustness of the disk detection is improved by using a template of vertebral distance; Ullmann, page 3, 2.4.3. Detection of Disks. The algorithm detects disks one after another by analyzing the spine intensity profile towards the caudal direction. The starting point is the C2-C3 disk (or the disk indicated by the user). For each disk to be detected, its probabilistic location is first estimated using the template of intervertebral disk distances ... all disks locations are found and then projected back to the spinal cord centerline. The estimation of the probabilistic location of the disk uses the template of generic distance. After each new disk detection, the vertebral level is known and the distance between the current disk and the previous disk can be iteratively estimated … adjusted location is the series of previous locations found by the algorithm, generic distance (𝑖) is the distance between the disk 𝑖 and the disk (𝑖 + 1) obtained from the template, and ratio (𝑖) adjusts the generic distances ... it iteratively scales the template. The variable ratio is updated for each new peak detection).
It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine the teachings of Ullmann and Yan because they are in the same field of endeavor. One skilled in the art would have been motivated to substitute the distance measure of Ullmann for the “smaller region” as taught by Yan as an alternate means to localize the target features of interest (Yan, pages 1-2, I. Introduction; Ullmann, page 3, 2.4.3. Detection of Disks).
As per claim 11, Yan and Ullmann disclose the method of claim 10. Ullmann discloses wherein defining the search region comprises defining the search region based on a known orientation of the current target feature relative to the previously localized target feature and at least one of, a reference distance between the current target feature and the previously localized target feature, or a reference angle between the current target feature and the previously localized target feature (Ullmann, Abstract, The robustness of the disk detection is improved by using a template of vertebral distance; Ullmann, page 3, 2.4.3. Detection of Disks. The algorithm detects disks one after another by analyzing the spine intensity profile towards the caudal direction. The starting point is the C2-C3 disk (or the disk indicated by the user). For each disk to be detected, its probabilistic location is first estimated using the template of intervertebral disk distances). The motivation would be the same as above in claim 10.
As per claim 12, Yan and Ullmann disclose the method of claim 10. Ullmann discloses wherein the determining the position of the current target feature comprises iteratively adjusting a size or position of the search region and iteratively performing the matching until the position of the current target feature determined based on the matching satisfies a defined spatial accuracy criterion (Ullmann, page 3, 2.4.3. Detection of Disks. After each new disk detection, the vertebral level is known and the distance between the current disk and the previous disk can be iteratively estimated … adjusted location is the series of previous locations found by the algorithm, generic distance (𝑖) is the distance between the disk 𝑖 and the disk (𝑖 + 1) obtained from the template, and ratio (𝑖) adjusts the generic distances ... it iteratively scales the template. The variable ratio is updated for each new peak detection). The motivation would be the same as above in claim 10.
As per claim 13, Yan and Ullmann disclose the method of claim 12. Ullmann discloses wherein the iteratively adjusting comprises iteratively increasing the size of the search region, and wherein the defined spatial accuracy criterion comprises an acceptable distance or an acceptable angle between the position and the previously determined position of the previously localized target feature (Ullmann, pages 3-4, 2.4.3. Detection of Disks. The algorithm detects disks one after another by analyzing the spine intensity profile towards the caudal direction ... For each disk to be detected, its probabilistic location is first estimated using the template of intervertebral disk distances ... The range of peak searching is within 20% of generic distance (𝑖)∗ ratio (𝑖). The value of 20% has been set according to the value of standard deviation of intervertebral distance). The motivation would be the same as above in claim 10.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TRACY MANGIALASCHI whose telephone number is (571)270-5189. The examiner can normally be reached M-F, 9:30AM TO 6:00PM.
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, Vu Le can be reached at (571) 272-7332. 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.
/TRACY MANGIALASCHI/Examiner, Art Unit 2668