Abstract
In this paper, we investigate the classification of microscopic tumours using full digital mammography images. Firstly, to address the shortcomings of traditional image segmentation methods, two different deep learning methods are designed to achieve the segmentation of uterine fibroids. The deep lab model is used to optimize the lesion edge detailed information by using the void convolution algorithm and fully connected CRF, and the two semantic segmentation networks are compared to obtain the best results. The Mask RCNN case segmentation model is used to effectively extract features through the ResNet structure, combined with the RPN network to achieve effective use and fusion of features, and continuously optimize the network training to achieve a fine segmentation of the lesion area, and demonstrate the accuracy and feasibility of the two models in medical image segmentation. Histopathology was used to obtain ER, PR, HER scores, and Ki-67 percentage values for all patients. The Kaplan-Meier method was used for survival estimation, the Log-rank test was used for single-factor analysis, and Cox proportional risk regression was used for multifactor analysis. The prognostic value of each factor was calculated, as well as the factors affecting progression-free survival. This study was done to compare the imaging characteristics and diagnostic value of mammography and colour Doppler ultrasonography in nonspecific mastitis, improve the understanding of the imaging characteristics of nonspecific mastitis in these two examinations, improve the accuracy of the diagnosis of this type of disease, improve the ability of distinguishing it from breast cancer, and reduce the rate of misdiagnosis.
Highlights
A nonspecific mastitis is a group of chronic inflammatory diseases of the breast that do not occur during lactation and are not associated with bacterial infections [1]
We review the research progress of deep learning in medical imaging and discuss the opportunities and challenges of incorporating deep learning into future medical imaging [10]. e deep learning method based on convolutional neural networks has won an overwhelming victory in the ImageNet International Liquid Scale Visual Recognition Challenge (ILSVRC), and, for the first time, the error rate of the deep learning method is lower than that of human observation
Medical imaging differs from other aspects of medicine in that almost all the primary data and reports used for imaging are digital and these data are suitable for analysis by deep learning algorithms [12]. e potential applications of deep learning in medical imaging have become evident, and, in this paper, we will outline the applications of deep learning in medical imaging [13]
Summary
A nonspecific mastitis is a group of chronic inflammatory diseases of the breast that do not occur during lactation and are not associated with bacterial infections [1]. Acquisition error rate: since the performance of deep learning algorithms for image classification has been improving and impressive results have been achieved in other areas [11]. Common architectures used in medical imaging are based on AlexNet and VGG, which have fewer network layers and weights, and Wang et al used a scratch-trained model for assessing the presence of Alzheimer’s disease in cranial MRI-based deep learning [16]. E deep learning-based target detection algorithm can detect, localize, and classify the lesions on the mammography images with high accuracy, which provides radiologists with an auxiliary diagnosis for lesion identification and classification and makes a preliminary exploration for the further application of deep learning in medical image lesion detection Each fine-tuned classification model was evaluated for classification of breast density in a small dataset (4000 images) and in the original dataset (18152 images) to obtain the corresponding classification accuracy, and the classification performance of the model was classified as BI-RADS 4A, BI-RADS 3, and BI-RADS 2 for lesion assessment using the subject’s working characteristic curve and area under the curve. e evaluation classification of lesions as BI-RADS 4B, BI-RADS 4C, and BI-RADS 5 was set to be inconsistent with the pathology control if the lesions were considered to have a low probability of malignancy, a high probability of malignancy, or a high suspicion of malignancy. ree comparison groups were used to evaluate the agreement between the classification and the pathological findings. e first group consisted of X-ray alone and ultrasonography; the second group consisted of X-ray alone and both tests in combination, and the third group consisted of ultrasonography alone and both tests in combination. e compliance rates of the three groups were compared using the x-test, and P value was
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
