Cervical cancer cell image recognizing algorithm based on convolutional neural network

Abstract

The invention discloses a cervical cancer cell image recognizing algorithm based on a convolutional neural network. In a traditional method, due to precise dividing and manual feature extracting and selecting of cytoplasms or cell nucleuses, once features with dividing errors or features not accurately extracted are generated, the cell recognizing accuracy can be reduced. By automatically extracting deep features in cell images through the convolutional neural network, a recognizing task is completed. The recognizing algorithm includes the steps of roughly dividing a cervical cell TCT image; secondly, inputting the roughly-divided image into a convolutional neural network model for training; thirdly, storing the trained network model to obtain a final recognizing result. The cervical cancer cell image recognizing algorithm is different from the traditional recognizing method and has a good recognizing effect in actual application.

Description

Technical field [0001] The invention relates to the technical field of medical image processing, in particular to a cervical cancer cell picture recognition algorithm based on a convolutional neural network. Background technique [0002] In the past decades, the identification of cervical cancer cells mainly includes three steps: cell (cytoplasm and nucleus) segmentation, feature extraction and cell classification. Cell segmentation algorithms include threshold-based segmentation methods and edge-based segmentation methods. Cell feature extraction generally extracts manual features that describe morphological and chromatin features, and then uses feature selection or dimensionality reduction algorithms to organize the resulting features for classification. Cell classification generally uses classifiers based on Bayesian rules or support vector machines for classification. [0003] However, due to the serious overlap of cells, the existing segmentation algorithms cannot accurately...

AI Technical Summary

Problems solved by technology

The result of the classifier is based on the statistics of the features. If the statistics of the features are wrong, the results will not be very reliable.

Images