A Classification Method for Cell Image Recognition Based on Transform Domain Features and CNN

Abstract

The invention discloses a method for cell image recognition and classification based on transform domain features and CNN. The CNN neural network is set to include an input layer, a hidden layer and an output layer, wherein the input layer includes three channels of 72×72 neurons, and the hidden layer There are three convolutional layers, three pooling layers and two fully connected layers, and the cell image recognition and classification method includes the following steps: S10: Design a CNN input layer model, and fuse cell image transformation domain features with original image data; S20: Design the CNN hidden layer and output layer models, and input images to train the CNN model. The method of the present invention can more effectively train CNN model parameters and classify cell images when the number of training sets is not enough to train conventional CNN models. The accuracy of image recognition diagnosis is improved.

Description

technical field [0001] The invention relates to the field of medical and health diagnosis, in particular to a cell image recognition and classification method based on transform domain features and CNN (Convolutional Neural Network, Convolutional Neural Network). Background technique [0002] With the development of science and technology, medical imaging technology is widely used in the diagnosis and treatment of clinical diseases. With the help of medical imaging, doctors can more accurately and timely locate and assist in the identification of diseased parts before diagnosis, which facilitates further disease diagnosis and treatment. X-ray, B-ultrasound, and CT all use medical imaging technology. Cell image processing is an important branch of medical imaging. Due to the complexity of cell images, the quality of film production varies. At present, it mainly relies on manual image reading. Due to the visual fatigue caused by long-term observation of doctors and the differe...

AI Technical Summary

Problems solved by technology

Cell image processing is an important branch of medical imaging. Due to the complexity of cell images, the quality of film production varies. At present, it mainly relies on manual image reading. Due to the visual fatigue caused by long-term observation of doctors and the different levels of doctors' clinical experience and pathological analysis First, the diagnosis of diseases is often subject to the subjective influence of doctors, and the final diagnosis results often have high misdiagnosis. To improve these problems, in addition to improving the production technology, the introduction of computer image recognition and diagnosis technology for automatic analysis and processing has always been an image problem. It can deal with hot spots and difficulties in the field, and has certain applications in the medical field. However, due to the influence of image lighting and light intensity in the existing cell image processing technology, the robustness of cell classification is not enough. Computer image recognition Insufficient diagnostic accuracy

Images