Application of deep learning techniques in non-contrast computed tomography pulmonary angiogram for pulmonary embolism diagnosis

TL;DR

This paper tackles the problem of pulmonary embolism (PE) diagnosis without the use of contrast media to avoid acute kidney injury and treatment delays. It proposes a 3D convolutional neural network that analyzes non-contrast chest CT scans, with HU-range preprocessing to boost discriminative features. Using data from 192 patients and $5$-fold cross-validation, the model achieves up to $85\%$ accuracy and AUROC of $0.84$ across several HU ranges, illustrating feasibility for non-contrast PE classification. The findings emphasize that HU-range selection critically impacts performance and suggest future directions in multimodal fusion and expanded HU-range exploration to enhance practical applicability.

Abstract

Pulmonary embolism is a life-threatening disease, early detection and treatment can significantly reduce mortality. In recent years, many studies have been using deep learning in the diagnosis of pulmonary embolism with contrast medium computed tomography pulmonary angiography, but the contrast medium is likely to cause acute kidney injury in patients with pulmonary embolism and chronic kidney disease, and the contrast medium takes time to work, patients with acute pulmonary embolism may miss the golden treatment time. This study aims to use deep learning techniques to automatically classify pulmonary embolism in CT images without contrast medium by using a 3D convolutional neural network model. The deep learning model used in this study had a significant impact on the pulmonary embolism classification of computed tomography images without contrast with 85\% accuracy and 0.84 AUC, which confirms the feasibility of the model in the diagnosis of pulmonary embolism.

Figures (10)

• Figure 1: Illustrates the difference between CTPA scans with and without contrast medium: the left image is without a contrast medium, while the right image is with a contrast medium. In the right image, the circled area clearly shows a darker region. This is because PE causes blockages in the pulmonary vessels, resulting in reduced blood circulation and lower oxygen levels within these vessels. The contrast medium enhances the visibility of the occlusion (darker area). In contrast, the left image, without a contrast medium, makes it more challenging to visually determine the presence of a PE.
• Figure 2: CNN model architecture by b24
• Figure 3: CNN model architecture by b15
• Figure 4: CNN model architecture by b02
• Figure 5: Workflow for lung parenchyma and lesion segmentation by b25