Cancer cell detection device based on sliding window

Abstract

The invention discloses a cancer cell detection device based on a sliding window. The cancer cell detection device based on the sliding window comprises a computer, and the computer is programmed to execute the following steps of collecting image data through a cancer cell sampling needle buried in a patient body; after the collection is finished, carrying out fluorescence dyeing treatment and imaging on the data collected on the needle to obtain an image for detection; processing and judging image data; and marking the position of the cancer cell in the original image by using a square frameaccording to the position information obtained in the judgment process, and displaying a final detection result. According to the invention, the sliding window is adopted to intercept the image, so that the calculation complexity caused by inputting the whole image into a network is avoided, the parameter adjustment problem required by preprocessing and feature extraction of the image is also avoided, the operation is accelerated, the accuracy is improved, and the method has universality.

Description

technical field [0001] The invention relates to the field of cancer cell detection, in particular to a cancer cell detection device based on a sliding window. Background technique [0002] The high incidence and fatality rate of cancer has become the enemy of human life and health. In 2018 alone, there were 18.1 million new cancer cases and 9.6 million cancer deaths worldwide. Therefore, being able to detect the existence of tumor cells in time has become an important means of preventing and controlling cancer. With the development of various detection technologies, the detection methods of cancer cell images are also emerging one after another, and people are constantly seeking breakthroughs in the accuracy and speed of detection. [0003] From classic image processing methods such as threshold segmentation, K-means clustering, gray-scale co-occurrence matrix, and snake model to the currently popular machine learning methods such as deep neural networks, they have all been...

AI Technical Summary

Problems solved by technology

[0008] 1. CN201811068427 Cancer cell identification method based on the improved U-net convolutional neural network model: directly send the entire image to the network for training, the convolution process is time-consuming and inefficient

[0009] 2. CN201510489742 Medical Image Segmentation Method and Cancer Cell Identification Method Based on Defined Circle HSV Color Space: Convert to HSV Color Space to Segment Target and Background. This method is not sensitive to weak color information, and it is difficult to find the depth of cancer cell images. Hierarchical features, so it will cause more misjudgments and misjudgments, and the accuracy rate is not high

[0010] 3. CN201810793772 A method for re-identification and diagnosis of cervical cancer cells: preprocessing and feature extraction are required before network judgment, and parameter adjustment in this process is too cumbersome

Then use the extracted features to train the classifier, which is less efficient

And using traditional methods for preprocessing will affect the detection accuracy

Images