Modality Specific CBAM-VGGNet Model for the Classification of Breast Histopathology Images via Transfer Learning

Abstract

Histopathology images are very distinctive, and one image may contain thousands of objects.Transferring features from natural images to histopathology images may not provide impressive outcomes. In this study, we have proposed a novel modality-specific CBAM-VGGNet model for classifying H and E stained breast histopathology images. Instead of using pre-trained models on ImageNet, we have trained VGG16 and VGG19 models on the same domain cancerous histopathology datasets which are then used as fixed feature extractors. We have added GAP layer and Convolutional block attention module (CBAM) after the first convolutional layer of convolutional blocks. CBAM is an effective module for neural networks to focus on relevant features. We have implemented the VGG16 and VGG19 in a novel way following the configuration of state-of-the-art models with our own concatenated layers. The addition of the GAP layer in VGGNet has reduced the number of parameters and hence requiring less computational power. Both models are ensembled using the averaging ensemble technique. Features are extracted from the final ensembled model then passed to the feed-forward neural network. A hybrid pre-processing technique is proposed that first use median filter and afterwards contrast limited adaptive histogram equalization (CLAHE). Median filter is utilized to remove the highly significant noise and is directly related to image quality. CLAHE improve the local contrast present in an image and boost the weak boundary edges in each image pixel. The proposed CBAM ensemble model has outperformed state-of-the-art models with an accuracy of 98.96% and 97.95% F1-score on 400X data of BreakHis dataset.

Fine tuning deep learning models for breast tumor classification

“…Some literature, such as the work by Joseph et al 16 , reports higher performance metrics; however, these results may be subject to overfitting, casting doubt on their generalizability. Similarly, other studies, like that of Ijaz et al 25 , have applied their models at only one magnification level, which may limit the applicability of their findings across varying conditions. These observations underscore the importance of cautious interpretation of comparative performance metrics and highlight the need for comprehensive testing across diverse conditions to ensure robust and reliable model performance.…”

“…Ijaz et al 25 introduced the CBAM-VGGNet model, a fusion of VGG16 and VGG19, specifically trained on cancerous histopathology datasets. The model's complexity was streamlined using the GAP layer and CBAM, ensuring a focus on vital features.…”

Histopathological Cancer Detection Using Intra-Domain Transfer Learning and Ensemble Learning

“…Remarkably, the model pre-trained on the breast histopathology image dataset outperformed the other model by a substantial 13.2% accuracy, despite the difference in organ sites between the source and target datasets. Similarly, Ijaz et al [31] introduced a transfer learning approach by utilizing modified versions of VGG16 and VGG19 pretrained on source lung, colon, and breast cancer datasets, which were then applied to a target breast cancer dataset for binary classification. This approach achieved the highest accuracy of 98.96% on the x400 magnification factor subset, outperforming other models by margins ranging from 2.65% to 15.81% accuracy.…”

SAMPLER: unsupervised representations for rapid analysis of whole slide tissue images

“… 13 , 14 , 15 Recent studies use attention-based models to construct slide-level representation by aggregating weighted tile-level features, 12 , 16 , 17 , 18 , 19 e.g., via multi-head attention (MHA), hierarchical attention, dual attention, or convolutional block attention modules. 12 , 16 , 20 , 21 , 22 , 23 , 24 Other important approaches have included multiscale attention and vision transformer models, which utilize the correlations across tiles to improve slide-level representations. 20 , 25 , 26 , 27 , 28 Fully CNN-based approaches have also been considered for attention-like functions, e.g., encoding of tiles by a CNN followed by an additional deep CNN, 29 possibly with multi-scale tiling.…”