Automatic identification method and system for bone marrow cell image in continuous maturation stage, and medium

Abstract

The invention discloses a continuous mature stage bone marrow cell image automatic identification method and system, and a medium. The method mainly comprises the following steps: obtaining a data set conforming to a specification; constructing a dense connection type convolutional neural network model for automatic identification of bone marrow cells through transfer learning; carrying out size normalization on the single-cell image data, and dividing a data set after image size normalization; finely adjusting hyper-parameters of the training method, performing structural parameter training on the constructed model through fine-adjusted hyper-parameter training, obtaining an optimal structural parameter model, and introducing multiple types of random data augmentation in training; and carrying out bone marrow cell recognition effect test and evaluation on the model by utilizing multi-fold cross validation. According to the invention, automatic identification or classification of the bone marrow cells in the continuous maturation stage can be realized, the identification effect is good, and the performance and accuracy of automatic identification of the bone marrow cells in the continuous maturation stage can be improved.

Description

Technical field [0001] The present invention relates to the field of computer vision and depth learning, in particular, an automatic identification method of continuous maturation stage bone marrow cell images based on migration learning and dense connection type depth convolutional neural network. Background technique [0002] In the current medical and biological research, bone marrow cell classification is an important technique. Myel cell morphological examination is the earliest application of bone marrow cell identification and classification, which is of great significance in the classification diagnosis of major diseases such as blood tumors, and is also one of the most basic and most important bone marrow cells. The classification of bone marrow cells welcomes the diagnosis of a variety of malignant hematology, including: leukemia, bone marrow tumor, regenerative disabilities, anemia, etc., is an essential part of a variety of malignant disease diagnosis and treatment ef...

AI Technical Summary

Problems solved by technology

[0016] The technical problem to be solved by the present invention is to provide a method, system and medium for automatic recognition of bone marrow cell images in the continuous maturation stage in view of the shortcomings of the current automatic recognition technology of bone marrow cells. Considering the continuity and complexity of the differences in the characteristics of bone marrow cells at different stages, considering that the number of various types of bone marrow cells in the human body usually varies greatly, and considering the fact that bone marrow cells are densely distributed and it is difficult to ensure the consistency of the size of single-cell images, improve the automatic recognition of bone marrow cells performance, improve the accuracy of bone marrow cell identification

Images