Combining handcrafted features with latent variables in machine learning for prediction of radiation‐induced lung damage

Abstract

Purpose There has been burgeoning interest in applying machine learning methods for predicting radiotherapy outcomes. However, the imbalanced ratio of a large number of variables to a limited sample size in radiation oncology constitutes a major challenge. Therefore, dimensionality reduction methods can be a key to success. The study investigates and contrasts the application of traditional machine learning methods and deep learning approaches for outcome modeling in radiotherapy. In particular, new joint architectures based on variational autoencoder (VAE) for dimensionality reduction are presented and their application is demonstrated for the prediction of lung radiation pneumonitis (RP) from a large‐scale heterogeneous dataset. Methods A large‐scale heterogeneous dataset containing a pool of 230 variables including clinical factors (e.g., dose, KPS, stage) and biomarkers (e.g., single nucleotide polymorphisms (SNPs), cytokines, and micro‐RNAs) in a population of 106 nonsmall cell lung cancer (NSCLC) patients who received radiotherapy was used for modeling RP. Twenty‐two patients had grade 2 or higher RP. Four methods were investigated, including feature selection (case A) and feature extraction (case B) with traditional machine learning methods, a VAE‐MLP joint architecture (case C) with deep learning and lastly, the combination of feature selection and joint architecture (case D). For feature selection, Random forest (RF), Support Vector Machine (SVM), and multilayer perceptron (MLP) were implemented to select relevant features. Specifically, each method was run for multiple times to rank features within several cross‐validated (CV) resampled sets. A collection of ranking lists were then aggregated by top 5% and Kemeny graph methods to identify the final ranking for prediction. A synthetic minority oversampling technique was applied to correct for class imbalance during this process. For deep learning, a VAE‐MLP joint architecture where a VAE aimed for dimensionality reduction and an MLP aimed for classification was developed. In this architecture, reconstruction loss and prediction loss were combined into a single loss function to realize simultaneous training and weights were assigned to different classes to mitigate class imbalance. To evaluate the prediction performance and conduct comparisons, the area under receiver operating characteristic curves (AUCs) were performed for nested CVs for both handcrafted feature selections and the deep learning approach. The significance of differences in AUCs was assessed using the DeLong test of U‐statistics. Results An MLP‐based method using weight pruning (WP) feature selection yielded the best performance among the different hand‐crafted feature selection methods (case A), reaching an AUC of 0.804 (95% CI: 0.761–0.823) with 29 top features. A VAE‐MLP joint architecture (case C) achieved a comparable but slightly lower AUC of 0.781 (95% CI: 0.737–0.808) with the size of latent dimension being 2. The combination of handcrafted features (case A) and latent representat...

Radiomic prediction of radiation pneumonitis on pretreatment planning computed tomography images prior to lung cancer stereotactic body radiation therapy

“…The previous studies often used DVH parameters such as V20 and MLD as risk factors for RP prediction [8][9][10][11]14,31 . Various clinical factors and biomarkers such as cytokines, single nucleotide polymorphisms (SNPs), and microRNA have also been used for RP prediction 14,31 . In the field of radiomics, Cunliffe et al proposed that dose-dependent texture changes between pre-and post-RT CT images could classify patients with and without grade ≥ 2 RP.…”

Introduction to machine and deep learning for medical physicists

“…In another words, it is the equivalent of PCA analysis but for DL applications. It can be widely applied in radiation oncology considering the prevalence of high‐dimension data due to the limitation of patient sample sizes 14 . Similar to a VAE, a GAN is also a generative model 61 that can learn the multivariate distribution and describe how the data are generated.…”

“…Applications of machine learning (ML) and deep learning (DL), as a branch of intelligence (AI) in medical physics have witnessed rapid growth over the past few years. These techniques have been studied as effective tools for a wide range of applications in medicine and oncology, including computer‐aided detection and diagnosis, 1,2 image segmentation 3,4 knowledge‐based planning, 5–7 quality assurance, 8,9 radiomics feature extraction, 10,11 and outcomes modeling 12–16 . Numerous studies, as summarized in Fig.…”

Balancing accuracy and interpretability of machine learning approaches for radiation treatment outcomes modeling

“… 55 Recently, the combination of traditional ML methods and DL variational autoencoders (VAE) techniques was developed to deal with limited datasets for radiation-induced lung damage prediction as shown in Figure 2 . 56 It was demonstrated that a multilayer perceptron (MLP) method using weight pruning (WP) feature selection achieved the best performance among different hand-crafted feature selection methods, and the combination of handcrafted features and latent representation (Case D: latent Z + WP + MLP) yielded significant prediction performance improvement compared with handcrafted features only (Case A: WP + MLP), VAE-MLP disjoint (Case B) and VAE-MLP joint architectures (Case C).…”

“… The evaluation of combination of handcraft features and latent variables for radiation-induced lung damage prediction, 56 where, “RF”, random forest; “SVM”, support vector machine; “MLP”, multi-layer perceptron; “FQI”, feature quality index; “FSPP”, feature-based sensitivity of posterior probability; “WP”, weight pruning; “VAE”, variational autoencoders. …”