Centre for Biomedical Engineering, Indian Institute of Technology Delhi, New Delhi, India.
All India Institute of Medical Sciences New Delhi, Medical Oncology, Dr. B.R.A. IRCH, New Delhi, India.
Int J Comput Assist Radiol Surg. 2024 Feb;19(2):261-272. doi: 10.1007/s11548-023-03010-0. Epub 2023 Aug 18.
The proposed work aims to develop an algorithm to precisely segment the lung parenchyma in thoracic CT scans. To achieve this goal, the proposed technique utilized a combination of deep learning and traditional image processing algorithms. The initial step utilized a trained convolutional neural network (CNN) to generate preliminary lung masks, followed by the proposed post-processing algorithm for lung boundary correction.
First, the proposed method trained an improved 2D U-Net CNN model with Inception-ResNet-v2 as its backbone. The model was trained on 32 CT scans from two different sources: one from the VESSEL12 grand challenge and the other from AIIMS Delhi. Further, the model's performance was evaluated on a test dataset of 16 CT scans with juxta-pleural nodules obtained from AIIMS Delhi and the LUNA16 challenge. The model's performance was assessed using evaluation metrics such as average volumetric dice coefficient (DSC), average IoU score (IoU), and average F1 score (F1). Finally, the proposed post-processing algorithm was implemented to eliminate false positives from the model's prediction and to include juxta-pleural nodules in the final lung masks.
The trained model reported a DSC of 0.9791 ± 0.008, IoU of 0.9624 ± 0.007, and F1 of 0.9792 ± 0.004 on the test dataset. Applying the post-processing algorithm to the predicted lung masks obtained a DSC of 0.9713 ± 0.007, IoU of 0.9486 ± 0.007, and F1 of 0.9701 ± 0.008. The post-processing algorithm successfully included juxta-pleural nodules in the final lung mask.
Using a CNN model, the proposed method for lung parenchyma segmentation produced precise segmentation results. Furthermore, the post-processing algorithm addressed false positives and negatives in the model's predictions. Overall, the proposed approach demonstrated promising results for lung parenchyma segmentation. The method has the potential to be valuable in the advancement of computer-aided diagnosis (CAD) systems for automatic nodule detection.
本研究旨在开发一种精确分割胸部 CT 扫描中肺实质的算法。为实现这一目标，该技术结合了深度学习和传统图像处理算法。该技术首先使用经过训练的卷积神经网络（CNN）生成初步的肺掩模，然后使用提出的肺边界修正后处理算法进行修正。
首先，该方法使用带有 Inception-ResNet-v2 作为骨干网络的改进 2D U-Net CNN 模型进行训练。该模型在两个不同来源的 32 个 CT 扫描上进行训练：一个来自 VESSEL12 挑战赛，另一个来自 AIIMS 德里。此外，该模型在来自 AIIMS 德里和 LUNA16 挑战赛的带有贴壁结节的 16 个 CT 扫描测试数据集上进行了评估。使用平均体积骰子系数（DSC）、平均交并比（IoU）和平均 F1 分数（F1）等评估指标评估模型的性能。最后，实施了提出的后处理算法，以消除模型预测中的假阳性，并将贴壁结节包括在最终的肺掩模中。
在测试数据集上，训练后的模型报告的 DSC 为 0.9791±0.008，IoU 为 0.9624±0.007，F1 为 0.9792±0.004。将后处理算法应用于预测的肺掩模，得到的 DSC 为 0.9713±0.007，IoU 为 0.9486±0.007，F1 为 0.9701±0.008。后处理算法成功地将贴壁结节包括在最终的肺掩模中。
该方法使用 CNN 模型对肺实质进行分割，得到了精确的分割结果。此外，后处理算法解决了模型预测中的假阳性和假阴性问题。总体而言，该方法在肺实质分割方面取得了有前景的结果。该方法有可能为自动结节检测的计算机辅助诊断（CAD）系统的发展提供价值。
