Research on breast cancer pathological image classification method based on wavelet transform and YOLOv8

Area of Science:

• Oncology
• Medical Imaging
• Computer Science

Background:

• Breast cancer poses a significant global health challenge with high morbidity and mortality rates.
• Deep learning advancements are revolutionizing computer-aided diagnosis, particularly in pathological image analysis.
• Convolutional neural networks (CNNs) are increasingly replacing traditional methods for automatic feature extraction in medical imaging.

Purpose of the Study:

• To propose a novel deep learning-based method for classifying breast cancer pathological images.
• To enhance the accuracy of breast cancer diagnosis through improved image classification.
• To evaluate the effectiveness of combining wavelet transform with deep learning models for pathological image analysis.

Main Methods:

• Utilized image flipping for data augmentation to expand the dataset.
• Applied two-level wavelet decomposition and reconfiguration for image sharpening and enhancement.
• Employed the YOLOv8 network model for eight-class classification of breast cancer pathological images.
• Divided the processed dataset into training (80%) and testing (20% or 30%) sets.

Main Results:

• The proposed method, integrating wavelet transform with YOLOv8, demonstrated improved classification accuracy compared to using YOLOv8 on the original dataset.
• The enhancement in classification accuracy was observed across images with varying magnifications.
• The combination of wavelet decomposition and YOLOv8 proved effective in classifying breast cancer pathological images.

Conclusions:

• The novel method effectively enhances the classification of breast cancer pathological images.
• Combining two-level wavelet decomposition and reconfiguration with the YOLOv8 network model offers a promising approach for computer-aided breast cancer diagnosis.
• This technique shows potential for improving diagnostic accuracy in breast cancer detection.