Attention-Based Deep Learning Approach for Breast Cancer Histopathological Image Multi-Classification.

Breast cancer diagnosis from histopathology images is often time consuming and prone to human error, impacting treatment and prognosis. Deep learning diagnostic methods offer the potential for improved accuracy and efficiency in breast cancer detection and classification. However, they struggle with limited data and subtle variations within and between cancer types. Attention mechanisms provide feature refinement capabilities that have shown promise in overcoming such challenges. To this end, this paper proposes the Efficient Channel Spatial Attention Network (ECSAnet), an architecture built on EfficientNetV2 and augmented with a convolutional block attention module (CBAM) and additional fully connected layers. ECSAnet was fine-tuned using the BreakHis dataset, employing Reinhard stain normalization and image augmentation techniques to minimize overfitting and enhance generalizability. In testing, ECSAnet outperformed AlexNet, DenseNet121, EfficientNetV2-S, InceptionNetV3, ResNet50, and VGG16 in most settings, achieving accuracies of 94.2% at 40×, 92.96% at 100×, 88.41% at 200×, and 89.42% at 400× magnifications. The results highlight the effectiveness of CBAM in improving classification accuracy and the importance of stain normalization for generalizability.

The multiscale top-hat tensor enables specific enhancement of curvilinear structures in 2D and 3D images.

Quantification and modelling of curvilinear structures in 2D and 3D images is a common challenge in a wide range of biomedical applications. Image enhancement is a crucial pre-processing step for curvilinear structure quantification. Many of the existing state-of-the-art enhancement approaches still suffer from contrast variations and noise. In this paper, we propose to address such problems via the use of a multiscale image processing approach, called Multiscale Top-Hat Tensor (MTHT). MTHT produces a better quality enhancement of curvilinear structures in low contrast and noisy images compared with other approaches in a range of 2D and 3D biomedical images. The proposed approach combines multiscale morphological filtering with a local tensor representation of curvilinear structure. The MTHT approach is validated on 2D and 3D synthetic and real images, and is also compared to the state-of-the-art curvilinear structure enhancement approaches. The obtained results demonstrate that the proposed approach provides high-quality curvilinear structure enhancement, allowing high accuracy segmentation and quantification in a wide range of 2D and 3D image datasets.