Prediction of Radiation Pneumonitis With Dose Distribution: A Convolutional Neural Network (CNN) Based Model

Bin Liang, Yuan Tian, Xinyuan Chen, Hui Yan, Lingling Yan, Tao Zhang, Zongmei Zhou, Lvhua Wang, Jianrong DaiFrontiers in Oncology 2020 · DOI

TL;DR

This paper introduces a novel Convolutional Neural Network (CNN) approach, specifically a 3D CNN (C3D), for predicting radiation pneumonitis (RP) in lung cancer patients based on radiation dose distributions. The C3D model significantly outperforms traditional logistic regression models, demonstrating the power of deep learning in capturing complex spatial features within dose data for improved toxicity prediction.

What Problem Does This Paper Address?

Radiation pneumonitis (RP) is a significant side effect of thoracic radiotherapy, limiting the dose that can be safely delivered to tumors. While traditional methods use dosimetric factors (like mean lung dose or dose-volume histograms) and even more advanced "dosiomics" (texture features from dose distributions) to predict RP, these approaches often fail to capture the full spatial information within the 3D dose distribution. This gap limits the accuracy of RP prediction, potentially leading to under- or over-treatment. The authors aim to build a more accurate prediction model by leveraging the full 3D spatial information of the dose distribution, hypothesizing that high-order, subtle features are crucial for predicting RP.

Methods

The core of this study is the application of a 3D Convolutional Neural Network (C3D) for RP prediction. The C3D network, originally designed for action recognition in videos, was adapted for this task. The 3D dose distribution of the lungs served as the input. Key methodological aspects include:

  • Transfer Learning: The C3D network was pre-trained on the large UCF101 video dataset. This allows the model to leverage learned spatial feature extraction capabilities, mitigating the challenge of limited patient data for RP prediction.
  • 3D Convolutions: Unlike 2D approaches, the C3D network performs convolutions in three dimensions, enabling it to directly process the volumetric dose distribution and capture spatial relationships.
  • Guided Grad-CAM: To understand why the model makes certain predictions, the authors employed Guided Gradient-weighted Class Activation Mapping (Grad-CAM). This technique highlights the regions within the dose distribution that are most influential for predicting RP, providing visual interpretability.
  • Model Comparison: The performance of the C3D model was rigorously compared against three multivariate logistic regression (LR) models that used:
  • Dosimetric factors (e.g., V5, V10, MLD).
  • Normal Tissue Complication Probability (NTCP) factors (e.g., EUD, Lyman, PS models).
  • Dosiomics features (texture features extracted from the dose distribution).

Datasets

The study utilized a dataset of 70 non-small cell lung cancer (NSCLC) patients treated with volumetric modulated arc therapy (VMAT) at their institution between 2013 and 2016. RP grade was evaluated using the Common Terminology Criteria for Adverse Events (CTCAE v3.0). The primary endpoint for prediction was grade ≥ 2 RP.

The C3D network was pre-trained on the UCF101 video dataset, which contains 13,320 video clips across 101 action categories. This pre-training is crucial for transfer learning.

Key Insights and Ideas

  • 3D Spatial Information is Key: The study strongly suggests that capturing the full 3D spatial distribution of radiation dose is critical for accurate RP prediction, going beyond traditional 2D or feature-extracted methods.
  • Deep Learning Outperforms Classical Methods: The C3D CNN model demonstrated superior predictive performance compared to established logistic regression models using dosimetric, NTCP, or dosiomics features. This highlights the power of deep learning in uncovering complex patterns.
  • Transfer Learning is Effective: Pre-training a CNN on a related task (action recognition) and then fine-tuning it for a different domain (dose distribution) is a viable strategy to overcome data scarcity in medical applications.
  • Interpretability Matters: Guided Grad-CAM provides valuable insights into which regions of the dose distribution are most predictive of RP, offering a window into the "black box" of the CNN.
  • Hierarchical Feature Learning: Deeper layers of the CNN learn increasingly complex and abstract features, which are difficult to define mathematically but are highly effective for prediction. Shallow layers capture simpler, more global patterns.

Key Results

  • AUC Performance: The C3D prediction model achieved an Area Under the Curve (AUC) of 0.842. This significantly outperformed the LR models:
  • Dosimetric factors: AUC = 0.676
    • NTCP factors: AUC = 0.744
    • Dosiomics factors: AUC = 0.782
  • Training Strategies: Fine-tuning both the convolution layers and the fully connected (FC) layers simultaneously ("Both") yielded the best prediction performance (AUC = 0.842), outperforming training from scratch or training only the FC layers.
  • Discriminative Regions: Guided Grad-CAM indicated that the low-dose region of the contralateral lung and the high-dose region of the ipsilateral lung were strongly correlated with RP incidence (grade ≥ 2) and non-RP cases (grade < 2), respectively.
  • Feature Complexity: Shallow CNN filters learned simple, monotonous features, while deeper filters learned complex, non-globally consistent patterns.

Limitations and Future Work

  • Dataset Size: The study acknowledges that the dataset of 70 patients is relatively small, which makes training a complex CNN from scratch challenging and increases the risk of overfitting. While transfer learning mitigates this, larger datasets would further enhance robustness.
  • Interpretability Limitations: While Grad-CAM provides visual cues, the exact mathematical definition and clinical meaning of the high-order features learned by deeper CNN layers remain unclear.
  • Clinical Factors: The study primarily focuses on dose distribution. The authors note that other clinical factors (age, chemotherapy) are also associated with RP incidence, suggesting that future models could benefit from integrating these alongside dose information.
  • Future Directions: The authors suggest further investigation into how the learned features can be utilized for treatment plan design and the integration of other clinical factors into the prediction model.

Summary updated

Abstract

Radiation pneumonitis (RP) is one of the major side effects of thoracic radiotherapy. The aim of this study is to build a dose distribution based prediction model, and investigate the correlation of RP incidence and high-order features of dose distribution. A convolution 3D (C3D) neural network was used to construct the prediction model. The C3D network was pre-trained for action recognition. The dose distribution was used as input of the prediction model. With the C3D network, the convolution operation was performed in 3D space. The guided gradient-weighted class activation map (grad-CAM) was utilized to locate the regions of dose distribution which were strongly correlated with grade≥2 and grade<2 RP cases, respectively. The features learned by the convolution filters were generated with gradient ascend to understand the deep network. The performance of the C3D prediction model was evaluated by comparing with three multivariate logistic regression (LR) prediction models, which used the dosimetric, normal tissue complication probability (NTCP) or dosiomics factors as input, respectively. All the prediction models were validated using 70 non-small cell lung cancer (NSCLC) patients treated with volumetric modulated arc therapy (VMAT). The area under curve (AUC) of C3D prediction model was 0.842. While the AUC of the three LR models were 0.676, 0.744 and 0.782, respectively. The guided grad-CAM indicated that the low-dose region of contralateral lung and high-dose region of ipsilateral lung were strongly correlated with the grade≥2 and grade<2 RP cases, respectively. The features learned by shallow filters were simple and globally consistent, and of monotonous color. The features of deeper filters displayed more complicated pattern, which was hard or impossible to give strict mathematical definition. In conclusion, we built a C3D model for thoracic radiotherapy toxicity prediction. The results demonstrate its performance is superior over the classical LR models. In addition, CNN also offers a new perspective to further understand RP incidence.
BibTeX
@inproceedings{liang2020prediction,
  author = {Bin Liang and Yuan Tian and Xinyuan Chen and Hui Yan and Lingling Yan and Tao Zhang and Zongmei Zhou and Lvhua Wang and Jianrong Dai},
  title = {{Prediction of Radiation Pneumonitis With Dose Distribution: A Convolutional Neural Network (CNN) Based Model}},
  booktitle = {Frontiers in Oncology},
  year = {2020},
  doi = {10.3389/fonc.2019.01500},
}
Figure