Classification models for analyzing a sample
Apparatus and methods are described including analyzing one or more microscopic images of the blood sample using a machine-learning classifier. An entity within the one or more microscopic images is identified using a first classification model, and a first estimated concentration of the entity within the sample is determined, based upon the entity as identified using the first classification model. The entity is identified within the one or more microscopic images using a second classification model, and a second estimated concentration of the entity within the sample is determined, based upon the entity as identified using the second classification model. The first and second estimated concentrations are compared to each other, and, in response to the comparison, a hybrid classification model that is a hybrid of the first and second classification models is used. Other applications are also described.
1 . A method comprising:
analyzing one or more microscopic images of the blood sample using a machine-learning classifier, the analyzing comprising:
identifying platelets within the one or more microscopic images using a first classification model;
determining a first estimated concentration of the platelets within the sample, based upon the platelets as identified using the first classification model;
identifying the platelets within the one or more microscopic images using a second classification model;
determining a second estimated concentration of the platelets within the sample, based upon the platelets as identified using the second classification model;
comparing the first and second estimated concentrations to each other;
based on the comparison, determining that at least one of the estimated concentrations is close to a threshold platelet-concentration value that is of clinical relevance; and
using a hybrid classification model that is a hybrid of the first and second classification models, in response thereto.
2 . The method according to claim 1 , wherein determining that at least one of the estimated concentrations is close to the threshold platelet-concentration value that is of clinical relevance comprises determining that the first estimated concentration is less than the threshold platelet-concentration value and the second estimated concentration is greater than the threshold platelet-concentration value.
3 . Apparatus comprising:
a microscope configured to acquire one or more microscopic images of the blood sample;
an output device; and
at least one computer processor configured to:
analyze the one or more microscopic images of the blood sample using a machine-learning classifier, the analyzing comprising:
identifying platelets within the one or more microscopic images using a first classification model,
determining a first estimated concentration of the platelets within the sample, based upon the platelets as identified using the first classification model,
identifying the platelets within the one or more microscopic images using a second classification model,
determining a second estimated concentration of the platelets within the sample, based upon the platelets as identified using the second classification model,
comparing the first and second estimated concentrations to each other, and
based on the comparison, determining that at least one of the estimated concentrations is close to a threshold platelet-concentration value that is of clinical relevance; and
in response thereto, using a hybrid classification model that is a hybrid of the first and second classification models, and
generate an output on the output device based upon analyzing the one or more microscopic images of the blood sample using the machine-learning classifier.
4 . The apparatus according to claim 3 , wherein the computer processor is configured to determine that at least one of the estimated concentrations is close to the threshold platelet-concentration value that is of clinical relevance by determining that the first estimated concentration is less than the threshold platelet-concentration value and the second estimated concentration is greater than the threshold platelet-concentration value.