Background: Single photon emission computed tomography (SPECT) images provides functional information in both diagnosis and radiation therapy. This study aimed to evaluate a unified framework of different approaches and find the optimal one in integrating SPECT lung perfusion imaging into radiotherapy to estimate radiation pneumonitis.
Methods: Twenty-five patients with thoracic tumors were included in this study. All patients had SPECT perfusion imaging before radiotherapeutic treatment. The SPECT images were registered to the planning computed tomography (CT) via a rigid-body transformation. Then a unified framework was presented to integrate functional information with anatomical information to generate functional dose-volume parameters. The framework contained different mapping approaches using uniform, thresholding, linear or non-linear functions. To compare the different approaches in the unified framework, the ability of predicting lung toxicity outcome was evaluated via the receiver operating characteristic (ROC) curve analysis.
Results: Functional dose-volume metrics defined using the linear function achieves the highest value of area under the curve (AUC), compared to those defined with the other three types of mapping functions. With the linear mapping function, significant factors for predicting radiation pneumonia (p < 0.05) includes the functional mean lung dose (fMLD) with the threshold of 50-100% and lung volume receiving more than 30 Gy dose (fV_30) with the threshold of 60-100%, among which fMLD showed the highest prediction accuracy (AUC=0.772, threshold=50-100%).
Conclusions: We proposed a framework of functional dose-volume metrics to predict the outcome of radiation pneumonitis. The metrics using the linear function outperform the others.