# US 12,602,912 B2 · Memory capacity determination system in learning of cell images and memory capacity determination method in learning of cell images

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

- Number: US 12,602,912 B2
- Document ID: US12602912B2
- Status: Granted
- Kind: B2
- Application number: 18/565,884
- Filed: 2022-09-08
- Published: 2024-08-08
- Granted: 2026-04-14
- Earliest claimed: 2021-09-13
- Assignees: Shimadzu Corporation
- Applicants: Shimadzu Corporation
- Inventors: Hiroaki Tsushima; Shuhei Yamamoto; Takeshi Ono; Ryuji Sawada
- Web page: https://patentgrep.com/patent/US12602912B2

## CPC

- G06V 10/776: Validation; Performance evaluation
- G06F 11/30: Monitoring
- G06F 11/32: with visual {or acoustical} indication of the functioning of the machine
- G06F 12/02: Addressing or allocation; Relocation (program address sequencing G06F9/00; arrangements for selecting an address in a digital store G11C8/00)
- G06N 20/00: Machine learning
- G06T 7/00: Image analysis
- G06V 10/945: {User interactive design; Environments; Toolboxes}
- G06V 20/69: Microscopic objects, e.g. biological cells or cellular parts
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A memory capacity determination system (200) in learning of cell images (80) includes a learning processor (10) including a first processor (10a) configured to execute processes of training a learning model (21), and a memory (10b); a selector (45) configured to select between a training mode of training the learning model, and a validation mode of validating whether a capacity of the memory becomes insufficient; and a determiner (12d) configured to determine whether the capacity of the memory becomes insufficient in the verification mode; and a display (121) configured to give a notice based on a first determination result (32).

## Claims

### Claim 1

A memory capacity determination system in learning of cell images comprising:
a learning processor including a processor configured to execute a predetermined number of processes of training a learning model by using cell images under a first learning condition(s), and a memory configured to be used as a working area for executing processes of training the learning model;
a selector configured to select between a training mode of training the learning model by executing processes of training by using the learning processor, and a validation mode of validating whether a capacity of the memory becomes insufficient in a process(es) of training by using the learning processor;
a determiner configured to determine whether the capacity of the memory becomes insufficient in the process(es) of training the learning model in the verification mode; and
an informer configured to give a notice based on a determination result of the determiner.

### Claim 2 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1, wherein the learning processor is configured to execute, in the verification mode, the process(es) of training the number of which is smaller than the number of processes of training in the learning mode.

### Claim 3 (depends on 2)

The memory capacity determination system in learning of cell images according to claim 2, wherein the learning processor is configured to execute a single process of training in the verification mode.

### Claim 4 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1 further comprising an image size increaser configured to increase a size of the cell image, wherein
the learning processor is configured to execute, in the verification mode, the process(es) of training by using the cell image the size of which is increased by the image size increaser.

### Claim 5 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1 further comprising a storage configured to store a second learning condition(s) under which the process(es) of training was/were executed in the verification mode, and a determination result(s) corresponding to the second learning condition(s), wherein
the determiner is configured to determine whether the same condition(s) as the first learning condition(s) are stored as the second learning condition(s) in the storage;
the learning processor is configured not to execute the process(es) of training in the verification mode if the same condition(s) as the first learning condition(s) is/are stored as the second learning condition(s) in the storage; and
the informer is configured to give, if the same condition(s) as the first learning condition(s) is/are stored as the second learning condition(s) in the storage, the determination result(s) corresponding to the second learning condition(s) that is/are same as the first learning condition(s).

### Claim 6 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1, wherein the informer is configured to give information that urges a user to reduce the number of the cell images in the first learning condition(s) if the capacity of the memory becomes insufficient.

### Claim 7 (depends on 6)

The memory capacity determination system in learning of cell images according to claim 6, wherein the informer is configured to give additional information that urges the user to reduce the size of the cell image in the first learning condition(s) if the capacity of the memory becomes insufficient.

### Claim 8 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1 further comprising an estimator configured to use the memory to execute, when the process(es) of training is/are executed by the learning processor in the verification mode, processing of estimation of the cell image by using a learned model that has learned in parallel to the process(es) of training executed by the learning processor.

### Claim 9 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1 further comprising a learning-condition changer configured to change the first learning condition, or one or more of the first learning conditions, wherein
the learning processor is configured to repeatedly execute the process(es) of training in the validation mode while the learning condition changer changes the first learning condition, or the one or more of the first learning conditions; and
the determiner is configured to acquire an optimal learning condition(s) for preventing that the capacity of the memory becomes insufficient as an optimal first learning condition(s) based on the determination result(s) of the determiner.

### Claim 10 (depends on 9)

The memory capacity determination system in learning of cell images according to claim 9, wherein
the learning condition changer is configured to change the number of the cell images in the first learning condition(s); and
the determiner is configured to acquire a maximum number of the cell images for preventing that the capacity of the memory becomes insufficient as the optimal first learning condition(s).

### Claim 11 (depends on 1)

The memory capacity determination system in learning of cell images according to claim 1 further comprising a learning condition register configured to register the first learning conditions that include a type of learning model, the number of the cell images used to train the learning model, a size of the cell image, and the number of identification classes into which the learning model learns.

### Claim 12

A memory capacity determination method in learning of cell images comprising:
a step of executing a predetermined number of processes of training a learning model by using cell images under a first learning condition(s);
a step of selecting between a training mode of training the learning model by executing processes of training the learning model, and a validation mode of validating whether a capacity of a memory used as a working area becomes insufficient in a process(es) of training the learning model;
a step of determining whether the capacity of the memory becomes insufficient in the process(es) of training the learning model in the verification mode; and
a step of giving a result of the determination whether the capacity of the memory becomes insufficient.