Advancing Lung Disease Diagnosis in 3D CT Scans

To enable more accurate diagnosis of lung disease in chest CT scans, we propose a straightforward yet effective model. Firstly, we analyze the characteristics of 3D CT scans and remove non-lung regions, which helps the model focus on lesion-related areas and reduces computational cost. We adopt ResNeSt-50 as a strong feature extractor, and use a weighted cross-entropy loss to mitigate class imbalance, especially for the minority squamous cell carcinoma category. Based on this, we further incorporate mixup and contrastive learning to improve the model's accuracy and robustness. Our model achieves a Macro F1 score of 84.29 on the validation set and 70.40 on the test set, and wins the first place in the Fair Disease Diagnosis Challenge.