Cell image analysis method and cell image analyzer
A cell image analysis method includes converting a first image into a label image by performing a segmentation process to identify a region of a cell that has already started differentiation and a region of an undifferentiated cell in the first image, acquiring a shape feature amount from the label image, and determining whether or not a cell colony includes a colony region that is a candidate for a search target based on the shape feature amount and a determination criterion.
1. A cell image analysis method comprising:
acquiring a first image of a cell colony including a cell having differentiation potential;
converting the first image into a label image by performing a segmentation process to identify a colony region of a cell that has already started differentiation and a colony region of an undifferentiated cell in the cell colony in the first image;
acquiring a numerical value of a shape feature which is a numerical value that expresses a shape of the cell colony in the label image from the label image;
receiving an input regarding a colony region of a search target from a user using a computer;
setting a determination criterion for the numerical value of the shape feature based on the user's input; and
determining whether or not the cell colony includes a colony region that is a candidate for the search target based on the numerical value of the shape feature and the determination criterion,
wherein in the receiving of the input regarding the colony region of the search target, an input is received, from the user, of selection information as to whether or not the cell colony in an image group of the cell colony including cells having differentiation potential includes a desired colony region as a target for picking which is an extraction operation of the cell,
wherein the setting of the determination criterion includes setting the determination criterion based on the received selection information,
wherein the setting of the determination criterion includes creating a first trained model that has acquired the determination criterion for the numerical value of the shape feature by machine learning using the numerical value of the shape feature acquired from the label image of the first image as input data and the selection information as teaching data, and
wherein the determining of whether or not the cell colony includes the colony region that is the candidate for the search target includes inputting the numerical value of the shape feature acquired from the label image of the first image to the first trained model to generate a determination result.
2. The cell image analysis method according to claim 1 , wherein the receiving of the input of the selection information includes allowing the user to specify the colony region in the first image or the label image of the first image or allowing the user to pick the colony region.
3. The cell image analysis method according to claim 1 , wherein the numerical value of the shape feature includes the numerical value of the shape feature related to at least one of i) an entire region of the cell colony included in the label image, ii) the colony region of the undifferentiated cell included in the cell colony, or iii) the colony region of the cell that has started differentiation included in the cell colony.
4. The cell image analysis method according to claim 3 , wherein the numerical value of the shape feature includes at least one of i) an area of a region, ii) a contour length of the region, iii) a degree of circularity of the region, iv) an aspect ratio of a minimum circumscribed rectangle of the region, or v) an area ratio of the colony region to the entire region of the cell colony.
5. The cell image analysis method according to claim 1 , wherein the converting of the first image into the label image includes generating the label image by a second trained model configured to assign a segmentation result label to the colony region, using a microscope image of the cell colony as input data.
6. A cell picking method utilizing the cell image analysis method according to claim 1 , comprising:
setting picking coordinates of the colony region determined to be desired by the user based on a determination result as to whether or not the cell colony includes the colony region that is the candidate for the search target; and
picking a cell at the picking coordinates from a culture vessel.
7. A cell image analyzer comprising:
a storage configured to allow a microscope image of a cell colony including a cell having differentiation potential to be input thereto;
a processor configured to convert the microscope image into a label image by performing a segmentation process to identify a colony region of a cell that has already started differentiation and a colony region of an undifferentiated cell in the cell colony in the microscope image; and
an input configured to receive an input, from a user, regarding a colony region of a search target,
wherein the processor is further configured to determine whether or not the cell colony included in the microscope image includes the colony region that is a candidate for the search target,
wherein the processor is further configured to acquire a numerical value of a shape feature which is a numerical value that express a shape of the cell colony in the label image from the label image and determine a colony region based on the numerical value of the shape feature and a determination criterion for the numerical value of the shape feature set based on a user's input,
wherein the input is configured to receive an input, from the user, of selection information as to whether or not the cell colony in an image group of the cell colony including cells having differentiation potential includes a desired colony region as a target for picking which is an extraction operation of the cell,
wherein the processor is further configured to set the determination criterion based on the received selection information,
wherein the processor is further configured to create a first trained model that has acquired the determination criterion for the numerical value of the shape feature by machine learning using the numerical value of the shape feature acquired from the label image of the microscope image as input data and the selection information as teaching data, and
wherein the processor is further configured to input the numerical value of the shape feature acquired from the label image of the microscope image to the first trained model to generate a determination result, in determining whether or not the cell colony includes the colony region.
8. The cell image analysis method according to claim 1 , wherein
in the step of converting the first image into a label image, the first image is converted into the label image including a label region of the identified region among at least two label regions, the colony region of the undifferentiated cell and the colony region of the cell that has started differentiation,
in the step of determining whether or not the cell colony includes the colony region that is the candidate for the search target, for each of the cell colonies in the first image whether or not the colony regions that are candidates for the search target are included, based on the numerical values of the shape feature of the label regions obtained from the label image and the determination criterion input by the user by using the computer.
9. The cell image analysis method according to claim 1 , wherein
in the step of converting the first image into a label image, the first image is converted into the label image including a label region of the identified region among at least two label regions, the colony region of the undifferentiated cell and the colony region of the cell that has started differentiation, and
in the step of determining whether or not the cell colony includes the colony region that is the candidate for the search target, for each of the cell colonies in the first image whether or not the colony regions that are candidates for the search target are included, based on the numerical values of the shape feature of the label regions obtained from the label image and the determination criterion input by the user by using the computer.