Method for processing an image of a sample comprising biological particles

A method for characterizing biological particles of a sample, the characterization assigning a class to each particle, each class being representative of a property of the particle, the method comprising:

• a) measuring a characteristic of at least one particle of the sample;
• b) processing the characteristic of each particle using an artificial intelligence algorithm;
• c) based on the processing, characterizing the particles of the sample, so as to assign a class to each particle;
• wherein the artificial intelligence algorithm has been trained beforehand by learning from label proportions, with at least one training sample, comprising training particles, the training comprising:
• (i) determining said characteristic of each training particle;
• (ii) determining a proportion of training particles belonging to the same class;
• (iii) annotating each training particle on the basis of the proportion determined in step (ii);
• (iv) using the characteristics of the training particles, determined in (i), respectively annotated based on the annotations resulting from (iii), as training data for the artificial intelligence algorithm;
• wherein:
• steps (i) to (iii) are carried out for various training samples;
• each step (ii) comprises determining a proportion of training particles, belonging to the same class, respectively in each training sample;
• the respective annotations of the training particles of one and the same training sample are identical and known, so that the learning from label proportions is a supervised learning.