Mammary gland lesion area detection method based on deep learning and transfer learning

Abstract

The invention provides a mammary gland lesion area detection method based on deep learning and transfer learning. The method comprises: preparation and amplification of a training set and a test set;according to lump position information marked by a doctor in the breast data set, extracting an available lump image and normalizing the size of the available lump image into a size of 100 * 100 pixels as a positive sample; according to the invention, the AlexNet network is used to train the parameter model of the classification model of the natural image on the ImageNet data set; training and transfer learning are carried out on a specific breast image data set, so that the binary classification problem of the convolutional neural network on a small-scale breast data set can be successfully solved, a lesion area in the breast image can be identified, and the prediction effect on the breast lesion is improved.

Description

technical field [0001] The invention is a method for detecting breast lesion areas based on deep learning and transfer learning, which belongs to the medical field. Background technique [0002] In the prior art, breast cancer is a common malignant tumor, and early diagnosis and treatment are the key to reducing breast cancer mortality. Lesions in breast images include masses, calcifications, bilateral asymmetry, and structural distortions. Masses and calcification clusters are the most common imaging signs of breast cancer. Therefore, automatic detection of masses and calcifications has also become a computer-aided method. There are two main aspects of the diagnostic system. Among them, due to factors such as blurred edges, various shapes, and low contrast with surrounding tissues, tumors have always been a major difficulty in computer-aided detection. Therefore, if not emphasized, the lesion area in this patent all refers to lump. [0003] The idea of ​​deep learning or...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that only the gray value of the image block is used for analysis, and the texture, edge, shape and other information of the image are not considered, which will affect the detection rate of the tumor to a certain extent, and will also cause high false positives. Rate

The shortcomings of this method are: training a convolutional neural network from scratch requires a large amount of labeled data and the complexity of the parameter adjustment process is high, and it is not applicable to small data sets, especially in the extreme lack of medically labeled information. in the case of medical images

Images