Adaptation to CT Reconstruction Kernels by Enforcing Cross-domain Feature Maps Consistency

TL;DR

This paper introduces F-Consistency, an unsupervised domain adaptation method that improves COVID-19 CT segmentation across different reconstruction kernels by enforcing feature map similarity, leading to better generalization and performance.

Contribution

The paper proposes a novel unsupervised adaptation approach, F-Consistency, that enhances model robustness to kernel-induced domain shifts in CT images by aligning feature representations.

Findings

F-Consistency outperforms baseline and other methods in Dice Score on unseen kernels.

It nearly doubles the similarity score between paired images compared to baseline.

The method improves generalization without requiring semantic content annotations.

Abstract

Deep learning methods provide significant assistance in analyzing coronavirus disease (COVID-19) in chest computed tomography (CT) images, including identification, severity assessment, and segmentation. Although the earlier developed methods address the lack of data and specific annotations, the current goal is to build a robust algorithm for clinical use, having a larger pool of available data. With the larger datasets, the domain shift problem arises, affecting the performance of methods on the unseen data. One of the critical sources of domain shift in CT images is the difference in reconstruction kernels used to generate images from the raw data (sinograms). In this paper, we show a decrease in the COVID-19 segmentation quality of the model trained on the smooth and tested on the sharp reconstruction kernels. Furthermore, we compare several domain adaptation approaches to tackle the problem, such as task-specific augmentation and unsupervised adversarial learning. Finally, we propose the unsupervised adaptation method, called F-Consistency, that outperforms the previous approaches. Our method exploits a set of unlabeled CT image pairs which differ only in reconstruction kernels within every pair. It enforces the similarity of the network hidden representations (feature maps) by minimizing mean squared error (MSE) between paired feature maps. We show our method achieving 0.64 Dice Score on the test dataset with unseen sharp kernels, compared to the 0.56 Dice Score of the baseline model. Moreover, F-Consistency scores 0.80 Dice Score between predictions on the paired images, which almost doubles the baseline score of 0.46 and surpasses the other methods. We also show F-Consistency to better generalize on the unseen kernels and without the specific semantic content, e.g., presence of the COVID-19 lesions.