Multi Source COVID-19 Detection via Kernel-Density-based Slice Sampling

TL;DR

This paper introduces a novel multi-source COVID-19 detection method using kernel-density-based slice sampling and a specialized feature learning framework, achieving high accuracy on chest CT scans from multiple medical centers.

Contribution

The paper proposes a new pipeline combining KDS and SSFL for multi-source CT scan classification, demonstrating improved performance over existing methods.

Findings

EfficientNet achieved an F1-score of 94.68%.

KDS-based pipeline effectively handles multi-source variability.

Dataset balance is crucial for multi-institutional evaluation.

Abstract

We present our solution for the Multi-Source COVID-19 Detection Challenge, which classifies chest CT scans from four distinct medical centers. To address multi-source variability, we employ the Spatial-Slice Feature Learning (SSFL) framework with Kernel-Density-based Slice Sampling (KDS). Our preprocessing pipeline combines lung region extraction, quality control, and adaptive slice sampling to select eight representative slices per scan. We compare EfficientNet and Swin Transformer architectures on the validation set. The EfficientNet model achieves an F1-score of 94.68%, compared to the Swin Transformer's 93.34%. The results demonstrate the effectiveness of our KDS-based pipeline on multi-source data and highlight the importance of dataset balance in multi-institutional medical imaging evaluation.