# US 12,014,833 B2 · System and method for selecting artificially fertilized embryos

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,014,833 B2
- Document ID: US12014833B2
- Status: Granted
- Kind: B2
- Application number: 17/266,300
- Filed: 2019-08-06
- Published: 2021-09-02
- Granted: 2024-06-18
- Earliest claimed: 2018-08-07
- Assignees: Cornell University; Yale University
- Applicants: Cornell University; Yale University
- Inventors: Nikica Zaninovic; Olivier Elemento; Iman Hajirasouliha; Pegah Khosravi; Jonas Malmsten; Zev Rosenwaks; Qiansheng Zhan; Ehsan Kazemi
- Web page: https://patentgrep.com/patent/US12014833B2

## CPC

- G16H 50/30: for calculating health indices; for individual health risk assessment
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 3/096: Transfer learning
- G06T 7/0012: {Biomedical image inspection}
- G06V 10/454: {Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]}
- G06V 10/82: using neural networks
- G06V 20/69: Microscopic objects, e.g. biological cells or cellular parts
- G06V 20/698: {Matching; Classification}
- G16H 10/60: for patient-specific data, e.g. for electronic patient records
- G16H 30/20: for handling medical images, e.g. DICOM, HL7 or PACS
- G16H 30/40: for processing medical images, e.g. editing
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- G16H 50/70: for mining of medical data, e.g. analysing previous cases of other patients
- G16H 70/60: relating to pathologies
- G06T 2207/20036: Morphological image processing
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30044: Fetus; Embryo
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A method for classifying human blastocysts includes obtaining images of a set of artificially fertilized (AF) embryos incubating in an incubator. A morphological quality of the AF embryos is determined based on a classification of the images by a convolutional neural network trained using images of pre-classified embryos. Each of the AF embryos is graded based on the morphological quality. A probability that a given graded AF embryo will result in a successful pregnancy after the given AF embryo is implanted in a gestating female is computed for each of the AF embryos from the set based on a grade of the given AF embryo and clinical parameters associated with the gestating female. One or more graded AF embryos to be recommended to be implanted in the gestating female from the set are selected based on the probability of successful pregnancy. An identity of each of the selected graded AF embryos and a number of the selected graded AF embryos to be potentially implanted in the gestating female is then output based on an outcome desired by the gestating female following implantation.

## Claims

### Claim 1

A method for classifying human blastocysts, comprising:
obtaining images of a set of artificially fertilized (AF) embryos incubating in an incubator;
determining a morphological quality of the AF embryos based on a classification of the images by a convolutional neural network trained using images of pre-classified embryos and grading each of the AF embryos based on the morphological quality, wherein the convolutional neural network is trained using the pre-classified embryos and clinical data associated with the pre-classified embryos, the clinical data associated with the pre-classified embryos comprising an age of a donor female providing eggs that were fertilized to obtain the pre-classified embryos and a presence of clinical conditions associated with the donor female, the clinical conditions including diabetes and obesity;
computing a probability that a given graded AF embryo will result in a successful pregnancy after the given AF embryo is implanted in a gestating female for each of the AF embryos from the set based on a grade of the given AF embryo and clinical parameters associated with the gestating female;
selecting graded AF embryos to be recommended to be implanted in the gestating female from the set based on the probability of successful pregnancy and an outcome desired by the gestating female following implantation and outputting an identity and number of the selected AF embryos.

### Claim 2 (depends on 1)

The method of claim 1, wherein the images are obtained at one or more predetermined points in time after fertilization.

### Claim 3 (depends on 1)

The method of claim 1, wherein the images include images of multiple planes of the AF embryos.

### Claim 4 (depends on 1)

The method of claim 1, wherein the selected AF embryos provide the highest likelihood of live birth following implantation of the selected AF embryos in the gestating female.

### Claim 5 (depends on 1)

The method of claim 1, wherein the morphological quality of the AF embryos is classified as good, fair and poor.

### Claim 6 (depends on 1)

The method of claim 1, wherein the images of pre-classified embryos used for training the convolutional neural network include images of pre-classified embryos taken at two different times post-fertilization and at seven different planes across each of the pre-classified embryos.

### Claim 7 (depends on 6)

The method of claim 6, wherein the images of pre-classified embryos are classified as good, fair and poor.

### Claim 8 (depends on 1)

The method of claim 1, wherein the images of the set of AF embryos are unprocessed bright field images using a Hoffman modulation contrast objective.

### Claim 9 (depends on 1)

The method of claim 1, wherein the clinical parameters associated with the gestating female include an age of the gestating female and a presence of clinical conditions associated with the gestating female, obesity.

### Claim 10

A system for classifying human blastocysts, comprising;
an input module configured to receive images of a set of artificially fertilized (AF) embryos incubating in an incubator;
a morphological quality evaluation module configured to determine a morphological quality of the AF embryos based on a classification of the images by a deep neural network trained using images of pre-classified embryos, wherein the deep neural network is trained using the pre-classified embryos and clinical data associated with the pre-classified embryos, the clinical data associated with the pre-classified embryos comprising an age of a donor female providing eggs that were fertilized to obtain the pre-classified embryos and a presence of clinical conditions associated with the donor female, the clinical conditions including diabetes and obesity;
a pregnancy success computing module configured to compute a probability that a given AF embryo will result in a successful pregnancy after the given AF embryo is implanted in a gestating female for each of the AF embryos from the set based on the morphological quality of the AF embryos and clinical parameters associated with the gestating female; and
an embryo selection module configured to select one or more of the AF embryos from the set based on the probability of successful pregnancy, and to output an identity of each of the selected AF embryos and a number of the selected AF embryos to be recommended to be implanted in the gestating female based on a desired outcome input into the embryo selection module following implantation.

### Claim 11 (depends on 10)

The system of claim 10, wherein the deep neural network is a convolutional neural network and the images of the pre-classified embryos are unprocessed bright field micrographs of the pre-classified embryos.

### Claim 12 (depends on 10)

The system of claim 10, wherein the images of the AF embryos and the images of the pre-classified embryos are monochrome images.

### Claim 13 (depends on 10)

The system of claim 10, wherein the embryo selection module is configured select AF embryos that provide a highest likelihood of live birth following implantation of the AF embryos in the gestating female.

### Claim 14 (depends on 10)

The system of claim 10, wherein the number of the selected AF embryos to be implanted into the gestating female is determined by the embryo selection module based on ab a combination of the number and the identity of selected AF embryos which will result in a highest likelihood of live birth following implantation of the AF embryos in the gestating female.

### Claim 15 (depends on 10)

The system of claim 10, wherein the clinical parameters associated with the gestating female include an age of the female and a presence of clinical conditions including diabetes and obesity.

### Claim 16

A non-transitory computer readable medium storing a computer readable program of classifying human blastocysts, comprising;
computer readable instructions to determine a morphological quality of artificially fertilized (AF) embryos based on a classification of images of the AF embryos by a convolutional neural network trained using images of pre-classified embryos, wherein the convolutional neural network is trained using the pre-classified embryos and clinical data associated with the pre-classified embryos, the clinical data associated with the pre-classified embryos comprising an age of a donor female providing eggs that were fertilized to obtain the pre-classified embryos and a presence of clinical conditions associated with the donor female, the clinical conditions including diabetes and obesity;
computer readable instructions to compute a probability that a given AF embryo will result in a successful pregnancy after the given AF embryo is implanted in a gestating female for each of the AF embryos, the probability being computed based on the morphological quality of the given AF embryo and clinical parameters associated with the gestating female; and
computer readable instructions to select one or more of the AF embryos based on the morphological quality of the AF embryos and clinical parameters associated with the gestating female, and provide an identity and number of AF embryos that will provide the highest likelihood of live birth following implantation in the gestating female.

### Claim 17 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein the clinical parameters associated with the gestating female include an age and a presence of clinical conditions including diabetes and obesity.

### Claim 18 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein the number of the selected AF embryos to be implanted into the gestating female is determined by the embryo selection module based on a combination of the number and identity of selected AF embryos which will result in a highest likelihood of live birth following implantation of the AF embryos in the gestating female.

### Claim 19 (depends on 16)

The non-transitory computer readable medium of claim 16, wherein the convolutional neural network is trained using the images of the pre-classified embryos and clinical data associated with the pre-classified embryos.