Automatic lobe segmentation using attentive cross entropy and end-to-end fissure generation

TL;DR

This paper presents a novel end-to-end framework for lung lobe segmentation that emphasizes fissure regions using attentive loss and introduces a fissure generation method, achieving high accuracy on private and public datasets.

Contribution

The proposed framework uniquely combines attentive loss, fissure generation, and registration-based loss to improve lung lobe segmentation accuracy without additional network branches.

Findings

Achieved 97.83% dice score on private dataset

Achieved 94.75% dice score on public dataset

Enhanced fissure segmentation accuracy with new loss functions

Abstract

The automatic lung lobe segmentation algorithm is of great significance for the diagnosis and treatment of lung diseases, however, which has great challenges due to the incompleteness of pulmonary fissures in lung CT images and the large variability of pathological features. Therefore, we propose a new automatic lung lobe segmentation framework, in which we urge the model to pay attention to the area around the pulmonary fissure during the training process, which is realized by a task-specific loss function. In addition, we introduce an end-to-end pulmonary fissure generation method in the auxiliary pulmonary fissure segmentation task, without any additional network branch. Finally, we propose a registration-based loss function to alleviate the convergence difficulty of the Dice loss supervised pulmonary fissure segmentation task. We achieve 97.83% and 94.75% dice scores on our private dataset STLB and public LUNA16 dataset respectively.