AUTOMATION OF ULTRASOUND BREAST CANCER IMAGES CLASSIFICATION USING DEEP NEURAL NETWORKS

The article is devoted to the analysis of ways to improve medical systems, which are used to diagnose breast diseases by ultrasound images. Prospects for the use of neural networks in this area are considered. An analysis of neural network architectures that can be used to automate the classification of ultrasound images carried out. The architecture of optimized EfficientNet B1 networks was chosen. Training and testing of the model showed 81.26% of correct answers acording to test results. A program with a graphical user interface in the LabVIEW environment has been created for easy analysis of ultrasound images. The development reduces the influence of subjective factors in diagnosis and improves the overall effectiveness of diagnostic method.