FANet: A Feedback Attention Network for Improved Biomedical Image Segmentation (2103.17235v3)

Abstract: The increase of available large clinical and experimental datasets has contributed to a substantial amount of important contributions in the area of biomedical image analysis. Image segmentation, which is crucial for any quantitative analysis, has especially attracted attention. Recent hardware advancement has led to the success of deep learning approaches. However, although deep learning models are being trained on large datasets, existing methods do not use the information from different learning epochs effectively. In this work, we leverage the information of each training epoch to prune the prediction maps of the subsequent epochs. We propose a novel architecture called feedback attention network (FANet) that unifies the previous epoch mask with the feature map of the current training epoch. The previous epoch mask is then used to provide a hard attention to the learned feature maps at different convolutional layers. The network also allows to rectify the predictions in an iterative fashion during the test time. We show that our proposed \textit{feedback attention} model provides a substantial improvement on most segmentation metrics tested on seven publicly available biomedical imaging datasets demonstrating the effectiveness of FANet. The source code is available at \url{https://github.com/nikhilroxtomar/FANet}.

FANet: A Feedback Attention Network for Improved Biomedical Image Segmentation

The work by Tomar et al. presents a novel deep learning architecture termed Feedback Attention Network (FANet) designed to enhance biomedical image segmentation accuracy. Given the critical importance of image segmentation in quantitative biomedical analyses, particularly amidst the proliferation of large clinical datasets, FANet addresses the inefficiency inherent in current models which may fail to effectively leverage information accumulated over successive learning epochs.

Novel Architecture - FANet

FANet introduces a feedback attention mechanism that integrates the segmentation masks from previous learning epochs with the feature maps of the current training epoch. This approach is posited to rectify cumulative prediction errors and refines segmentation outputs iteratively, facilitating build-up and adjustment of model weights based on the input from previous outputs.

Key innovations in FANet include:

• Feedback Attention Learning: This component propagates sample-specific masks across epochs, potentially capturing intra- and inter-class variabilities more effectively than existing architectures.
• Iterative Refinement: Unlike conventional approaches that perform a one-step mask prediction, FANet allows for iterative updating of the input mask with predictions, bolstering segmentation accuracy.
• Memory Efficiency: The architecture employs a run-length encoding strategy, reducing the memory footprint during the propagation of feature maps and masks.
• Impressive Performance Metrics: When benchmarked against seven diverse biomedical datasets, FANet consistently outperformed state-of-the-art (SOTA) segmentation algorithms in most scenarios.

Experimental Validation and Performance

FANet's performance was rigorously validated on widely-used datasets like Kvasir-SEG, ISIC 2018, and DRIVE, among others, with empirical results underscoring its superiority over existing algorithms. For instance, FANet achieved a mean Intersection over Union (mIoU) of 0.8153 and a Dice coefficient of 0.8803 on Kvasir-SEG, which is a substantial improvement over benchmarks set by prior methods.

Throughout the empirical evaluations, FANet demonstrated enhanced precision, recall, and specificity across different datasets, confirming that iterative feedback significantly augments conventional segmentation techniques. Moreover, it achieved these results while maintaining competitive computational efficiency, underscoring its potential for integration into real-world biomedical imaging workflows.

Implications and Future Work

From a theoretical standpoint, FANet illustrates the benefits of incorporating feedback loops into deep learning models, especially for complex tasks like biomedical image segmentation where data variability is pronounced. In practical terms, FANet's results suggest potential improvements in automated diagnostic accuracy across medical imaging applications, such as tumor detection and organ segmentation, where precision is paramount.

Looking forward, the researchers suggest that future work could explore the application of FANet in other domains, such as remote sensing or satellite imagery, where similar challenges in image segmentation persist. Furthermore, scaling the approach with more complex attention mechanisms or integrating unsupervised learning paradigms could potentially unlock additional performance gains and broader applicability. The adaptability of the FANet framework indicates promising avenues for further research in the field of attention-augmented neural networks.

In summary, the FANet architecture presents a significant step towards robust and efficient biomedical image segmentation, offering a compelling case for the adoption of feedback-based learning strategies in deep learning models.