Holographic Microwave Image Classification Using a Convolutional Neural Network.

Area of Science:

• Medical imaging
• Artificial intelligence in healthcare
• Biomedical engineering

Background:

• Holographic microwave imaging (HMI) is a promising technique for early breast cancer diagnosis.
• Automated classification of benign and malignant tumors in HMI is a significant challenge.
• Convolutional neural networks (CNNs) have shown strong performance in image classification and tumor detection.

Purpose of the Study:

• To investigate the feasibility of using CNNs for identifying and classifying HMI breast images.
• To evaluate the performance of a modified AlexNet architecture with transfer learning for HMI tumor classification.

Main Methods:

• A modified AlexNet architecture was employed, utilizing transfer learning.
• The network was trained and evaluated on a dataset of 966 HMI breast images.
• Performance was benchmarked against various pre-trained CNN models including ResNet, GoogLeNet, VGG19, DenseNet201, SqueezeNet, Inception v3, AlexNet, and Inception-ResNet-v2.

Main Results:

• The proposed modified AlexNet network achieved high classification accuracy.
• The network demonstrated effectiveness even with a small training dataset.
• Fast training times were observed for the proposed CNN approach.

Conclusions:

• CNNs, particularly the modified AlexNet with transfer learning, are feasible for automated classification of HMI breast images.
• This approach offers a potential solution for accurate and efficient breast cancer diagnosis using HMI.
• The study highlights the effectiveness of deep learning in analyzing complex medical imaging data.