Li Jianjun, Wang Kaiyue, Jiang Xiaozhe
School of Mechanical and Electrical Engineering, China Jiliang University, Hangzhou 310018, China.
Sensors (Basel). 2025 Jan 3;25(1):240. doi: 10.3390/s25010240.
Breast cancer (BC) is one of the most lethal cancers worldwide, and its early diagnosis is critical for improving patient survival rates. However, the extraction of key information from complex medical images and the attainment of high-precision classification present a significant challenge. In the field of signal processing, texture-rich images typically exhibit periodic patterns and structures, which are manifested as significant energy concentrations at specific frequencies in the frequency domain. Given the above considerations, this study is designed to explore the application of frequency domain analysis in BC histopathological classification. This study proposes the dual-branch adaptive frequency domain fusion network (AFFNet), designed to enable each branch to specialize in distinct frequency domain features of pathological images. Additionally, two different frequency domain approaches, namely Multi-Spectral Channel Attention (MSCA) and Fourier Filtering Enhancement Operator (FFEO), are employed to enhance the texture features of pathological images and minimize information loss. Moreover, the contributions of the two branches at different stages are dynamically adjusted by a frequency-domain-adaptive fusion strategy to accommodate the complexity and multi-scale features of pathological images. The experimental results, based on two public BC histopathological image datasets, corroborate the idea that AFFNet outperforms 10 state-of-the-art image classification methods, underscoring its effectiveness and superiority in this domain.
乳腺癌（BC）是全球最致命的癌症之一，其早期诊断对于提高患者生存率至关重要。然而，从复杂的医学图像中提取关键信息并实现高精度分类面临重大挑战。在信号处理领域，纹理丰富的图像通常呈现出周期性模式和结构，这在频域中表现为特定频率处的显著能量集中。基于上述考虑，本研究旨在探索频域分析在乳腺癌组织病理学分类中的应用。本研究提出了双分支自适应频域融合网络（AFFNet），旨在使每个分支专门处理病理图像的不同频域特征。此外，采用了两种不同的频域方法，即多光谱通道注意力（MSCA）和傅里叶滤波增强算子（FFEO），以增强病理图像的纹理特征并最小化信息损失。此外，通过频域自适应融合策略动态调整两个分支在不同阶段的贡献，以适应病理图像的复杂性和多尺度特征。基于两个公开的乳腺癌组织病理学图像数据集的实验结果证实了AFFNet优于10种先进的图像分类方法的观点，突出了其在该领域的有效性和优越性。
