Metaphase split-phase chromosome region searching method and device

Abstract

The invention provides a metaphase split-phase chromosome region searching method and device, and relates to the technical field of medical images. The metaphase split-phase chromosome region searching method comprises the following steps: acquiring a split-phase cell image; inputting the clipped division-stage cell image into a deep convolutional network; automatically completing a feature extraction process by the network, and further fusing the features of different levels; generating mid-metaphase chromosome region prediction results at multiple levels. According to the technical scheme provided by the invention, the efficiency and accuracy of searching for a metaphase chromosome region can be improved, and the method can be suitable for images in different shooting environments.

Description

technical field [0001] The present application relates to the technical field of medical images, in particular to a method and device for searching chromosome regions in metaphase. Background technique [0002] Chromosomal karyotype analysis is a common and important medical clinical diagnosis method. According to the chart results of karyotype analysis, chromosomal variation can be observed, thereby diagnosing various genetic diseases. When performing chromosome karyotype analysis, it is usually necessary to screen out a certain number of metaphase chromosome region images from a large number of chromosome cell images. If this process is all done manually by a geneticist, it will take a lot of time, and it will be easily interfered by the doctor's subjective judgment, which will affect the quality and accuracy of the karyotype analysis results. [0003] Aiming at various problems in the process of searching for chromosome regions in metaphase, many automatic searching met...

AI Technical Summary

Problems solved by technology

Existing metaphase chromosome search methods are not truly automated and involve some manual intervention

In the thresholding segmentation process of images in the detection stage, parameters such as thresholds need to be set, and the adaptability to image data from different sources is poor

The selection stage relies on the manual feature extraction process, the quality of the classification results depends on the effectiveness of the feature design, and there are certain subjective factors.

Even if there are methods that try to use deep learning and other methods to improve the classification process in the selection stage, it is still necessary to classify each candidate region by a deep network, and the execution efficiency is low

Moreover, these methods divide the detection stage and the selection stage into two independent processes, ignoring the connection between the two, which affects the quality of the detection results to a certain extent.

Images