Design and Selection of Machine Learning Methods Using Radiomics and Dosiomics for Normal Tissue Complication Probability Modeling of Xerostomia

Abstract

PurposeThe purpose of this study is to investigate whether machine learning with dosiomic, radiomic, and demographic features allows for xerostomia risk assessment more precise than normal tissue complication probability (NTCP) models based on the mean radiation dose to parotid glands.Material and methodsA cohort of 153 head-and-neck cancer patients was used to model xerostomia at 0–6 months (early), 6–15 months (late), 15–24 months (long-term), and at any time (a longitudinal model) after radiotherapy. Predictive power of the features was evaluated by the area under the receiver operating characteristic curve (AUC) of univariate logistic regression models. The multivariate NTCP models were tuned and tested with single and nested cross-validation, respectively. We compared predictive performance of seven classification algorithms, six feature selection methods, and ten data cleaning/class balancing techniques using the Friedman test and the Nemenyi post hoc analysis.ResultsNTCP models based on the parotid mean dose failed to predict xerostomia (AUCs < 0.60). The most informative predictors were found for late and long-term xerostomia. Late xerostomia correlated with the contralateral dose gradient in the anterior–posterior (AUC = 0.72) and the right–left (AUC = 0.68) direction, whereas long-term xerostomia was associated with parotid volumes (AUCs > 0.85), dose gradients in the right–left (AUCs > 0.78), and the anterior–posterior (AUCs > 0.72) direction. Multivariate models of long-term xerostomia were typically based on the parotid volume, the parotid eccentricity, and the dose–volume histogram (DVH) spread with the generalization AUCs ranging from 0.74 to 0.88. On average, support vector machines and extra-trees were the top performing classifiers, whereas the algorithms based on logistic regression were the best choice for feature selection. We found no advantage in using data cleaning or class balancing methods.ConclusionWe demonstrated that incorporation of organ- and dose-shape descriptors is beneficial for xerostomia prediction in highly conformal radiotherapy treatments. Due to strong reliance on patient-specific, dose-independent factors, our results underscore the need for development of personalized data-driven risk profiles for NTCP models of xerostomia. The facilitated machine learning pipeline is described in detail and can serve as a valuable reference for future work in radiomic and dosiomic NTCP modeling.

Radiomics for radiation oncologists: are we ready to go?

“…xerostomia. [63][64][65][66][67] More generally, highly predictive OAR or tumour volume changes during RT would be particularly helpful in the RT workflow, as it would trigger, e.g. replanning, which could be better organised or anticipated, or an adjustment to be made to the prescribed dose, with, e.g.…”

“…72 Improvements in machine learning algorithms 73,74 -through the huge amount of data they can process-have made it possible to investigate the hypothesis that OAR radio-induced toxicity is associated with the OAR structural organisation as captured by image analysis. The use of radiomics to identify specific image signatures in healthy tissue and the correlation with radiationinduced toxicity, is a promising emerging area of research, 5,75,76 which is mostly studied in terms of the risk of xerostomia or radiation pneumonitis centred on the parotid glands 61,66,77,78 and lung analysis, [79][80][81][82] respectively. Recently, "dosiomic" analysis has also emerged, as an extension of texture analysis but based on the dose distribution which is planned.…”

“…Recently, "dosiomic" analysis has also emerged, as an extension of texture analysis but based on the dose distribution which is planned. 66,83,84 Dosiomic parameters could be interestingly combined to other variables, e.g. clinical factors or radiomic parameters, for the construction of more accurate toxicity prediction models.…”

Artificial intelligence in radiation oncology treatment planning: a brief overview

“…Multiple manuscripts have been published using radiomics to predict radiation response, in some cases with prediction power outperforming standard clinical variables (77)(78)(79)(80)(81)(82), though not in all (83). Radiomicsbased statistical approaches can predict various radiation normal tissue complication probabilities including radiation pneumonitis, xerostomia, and rectal wall toxicity (84)(85)(86)(87)(88)(89). Radiomics data, coupled with genomic data and increasingly computable clinical record data, may escort radiation oncology into a new epoch of truly personalized radiation plans based on patient-specific knowledge.…”

Electron Density and Biologically Effective Dose (BED) Radiomics-Based Machine Learning Models to Predict Late Radiation-Induced Subcutaneous Fibrosis

“…Quantitative analysis of medical images could provide information about intensity, shape, size or volume, and texture of tumor or organs at risk that is distinct or complementary to that provided by other data sources (5). Recently, the combination of quantitative analysis of radiological images with Machine Learning (ML) methods, also known as "radiomics, " has been applied also to predict side effects of RT such as lung-injury following Stereotactic Body RT (SBRT) for lung cancer (6), gastrointestinal and genitourinary toxicities (7) and xerostomia (8).…”