An Explainable Contrastive-based Dilated Convolutional Network with Transformer for Pediatric Pneumonia Detection

TL;DR

A novel EXplainable Contrastive-based Dilated Convolutional Network with Transformer with Transformer (XCCNet) for pediatric pneumonia detection is proposed, which harnesses the spatial power of dilated convolutions and the global insights from contrastive-based transformers for effective feature refinement.

Abstract

Pediatric pneumonia remains a significant global threat, posing a larger mortality risk than any other communicable disease. According to UNICEF, it is a leading cause of mortality in children under five and requires prompt diagnosis. Early diagnosis using chest radiographs is the prevalent standard, but limitations include low radiation levels in unprocessed images and data imbalance issues. This necessitates the development of efficient, computer-aided diagnosis techniques. To this end, we propose a novel EXplainable Contrastive-based Dilated Convolutional Network with Transformer (XCCNet) for pediatric pneumonia detection. XCCNet harnesses the spatial power of dilated convolutions and the global insights from contrastive-based transformers for effective feature refinement. A robust chest X-ray processing module tackles low-intensity radiographs, while adversarial-based data augmentation mitigates the skewed distribution of chest X-rays in the dataset. Furthermore, we actively integrate an explainability approach through feature visualization, directly aligning it with the attention region that pinpoints the presence of pneumonia or normality in radiographs. The efficacy of XCCNet is comprehensively assessed on four publicly available datasets. Extensive performance evaluation demonstrates the superiority of XCCNet compared to state-of-the-art methods.

Figures (3)

• Figure 1: Left: Visualizing Anatomical and Acquisition Variabilities in chest X-Rays labeled (a) to (f), Right: Highlighting Skewed Distribution of Chest X-Rays in the Pneumonia Dataset
• Figure 2: The workflow of our proposed approach for detecting pediatric pneumonia in chest X-rays. The process begins with the input raw CXR $x_{raw}$, which the Chest X-ray Preprocessing Module it into enhanced CXR enhanced CXR $x_{rs}$. The Spatial Feature Extractor Module captures fine-grained features from the enhanced CXR. These features are then fused with global features extracted by the Contrastive-based Transformer Feature Extractor Module within the Dual Hybrid Feature Fusion and Classification Module. Finally, this module classifies the fused features as normal CXR or pneumonic CXR.
• Figure 3: A detailed breakdown of the Explainable Contrastive-based Dilated Convolutional Network with Transformer (XCCNet) framework for pediatric pneumonia detection in chest X-rays (CXRs). The process begins with the Chest X-ray Preprocessing Module (CXP), which transforms the raw input into an enhanced CXR ($x_{raw} \rightarrow x_{rs}$) to enhance feature quality. The enhanced CXR $x_{rs}$ is then fed into both the Spatial Feature Extractor Module (SFx) and the Contrastive-based Transformer Feature Extractor Module (CoTFx) in parallel. SFx captures fine-grained spatial features $\mathcal{F}_{SFx}$, while CoTFx extracts global features $\mathcal{F}_{CoTFx}$. A combined loss function $\mathscr{L}_{combined}$ guides the network towards accurate classification. Finally, the features are fused $\mathcal{F}_{SFx} \oplus \mathcal{F}_{CoTFx} \rightarrow \mathcal{F}_{fused}$ and used to categorize the CXR as normal or pneumonia.