| Autor: |
Patil, Chandrashekhar H., Naik, Sachin, Patil, Kalpesh, Sathe, Saloni, Bhingradiya, Meet |
| Předmět: |
|
| Zdroj: |
Grenze International Journal of Engineering & Technology (GIJET); Jun2024, Vol. 10 Issue 2, Part 2, p1183-1189, 7p |
| Abstrakt: |
one of the most important tasks in medical diagnostics is the identification of breast cancer using ultrasound imaging. In this work, ultrasound pictures are classified using a Convolutional Neural Network (CNN) technique into three classes: benign (non-hazardous), malignant (dangerous), and normal (no traces). The collection of data includes 1587 photos, of which 26.53% are malignant, 56.71% benign, and 16.76% normal cases. The CNN model makes use of transfer learning, extracting features using a pre-trained VGG16 architecture. Dropout regularisation, thick layers, and global average pooling are added for fine-tuning. Techniques for augmenting data are used during training to increase the resilience of the model. The model achieves a test accuracy of 85.49%, demonstrating its efficacy in breast cancer classification. Furthermore, confusion matrices are generated to visualize classification performance for each class, revealing insights into model behaviour. This study underscores the potential of CNN-based approaches in aiding medical professionals in accurate breast cancer detection through ultrasound imagery. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
