Review of Deep Learning Based Automatic Segmentation for Lung Cancer Radiotherapy

Abstract

Lung cancer is the leading cause of cancer-related mortality for males and females. Radiation therapy (RT) is one of the primary treatment modalities for lung cancer. While delivering the prescribed dose to tumor targets, it is essential to spare the tissues near the targets—the so-called organs-at-risk (OARs). An optimal RT planning benefits from the accurate segmentation of the gross tumor volume and surrounding OARs. Manual segmentation is a time-consuming and tedious task for radiation oncologists. Therefore, it is crucial to develop automatic image segmentation to relieve radiation oncologists of the tedious contouring work. Currently, the atlas-based automatic segmentation technique is commonly used in clinical routines. However, this technique depends heavily on the similarity between the atlas and the image segmented. With significant advances made in computer vision, deep learning as a part of artificial intelligence attracts increasing attention in medical image automatic segmentation. In this article, we reviewed deep learning based automatic segmentation techniques related to lung cancer and compared them with the atlas-based automatic segmentation technique. At present, the auto-segmentation of OARs with relatively large volume such as lung and heart etc. outperforms the organs with small volume such as esophagus. The average Dice similarity coefficient (DSC) of lung, heart and liver are over 0.9, and the best DSC of spinal cord reaches 0.9. However, the DSC of esophagus ranges between 0.71 and 0.87 with a ragged performance. In terms of the gross tumor volume, the average DSC is below 0.8. Although deep learning based automatic segmentation techniques indicate significant superiority in many aspects compared to manual segmentation, various issues still need to be solved. We discussed the potential issues in deep learning based automatic segmentation including low contrast, dataset size, consensus guidelines, and network design. Clinical limitations and future research directions of deep learning based automatic segmentation were discussed as well.

Deep Learning Model for Classifying Metastatic Epidural Spinal Cord Compression on MRI

“…DL is already being used in spine diseases to aid in the diagnosis of spinal stenosis on MRI spines, surgical planning, and prediction of outcomes in patients with spinal metastases ( 8 , 29 ). DL in spinal oncology imaging is limited with most researchers focusing on the detection of metastases ( 30 ), or automated spinal cord segmentation as an organ at risk for radiotherapy planning ( 31 ). Average Dice similarity coefficients for spinal cord segmentation are as high as 0.9 for automated lung cancer radiotherapy planning using DL on CT studies ( 32 , 33 ).…”

Deep Learning Model for Grading Metastatic Epidural Spinal Cord Compression on Staging CT

“…Deep learning is being used to aid the imaging diagnosis of many different conditions, including liver segmentation [31] and vertebral segmentation for evaluation of the spine [32]. Other deep learning applications in spinal conditions include degenerative lumbar spinal stenosis on an MRI [13], automated segmentation of the spinal cord for radiotherapy planning [33], and prediction of treatment outcomes and complications in spinal oncology [34]. Deep learning for automated detection of spinal canal compromise and spinal cord compression has been investigated on MRI.…”

Dose Prediction Using a Three-Dimensional Convolutional Neural Network for Nasopharyngeal Carcinoma With Tomotherapy

“…But the training strategy should be suitable for the specific prediction tasks. For example, the 2D U-Net can perform pretty well in the task of CT image segmentation ( 19 , 29 , 30 ). Slice by slice segmentation prediction is similar to the clinical logic flow.…”