Generative models improve radiomics performance in different tasks and different datasets: An experimental study. (June 2022)

• Record Type: Journal Article
• Authors: Chen, Junhua; Bermejo, Inigo; Dekker, Andre; Wee, Leonard
• Is Part Of: Physica medica. Volume 98(2022)
• Page Start: 11
• Page End: 17
• Publication Date: 2022-06
• Subjects: Radiomics -- Generative models -- Image denoising -- Comparative study; Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2022.04.008
• Languages: English
• ISSNs: 1120-1797

Abstract

Highlights: Generative models can improve radiomics performance in different tasks when radiomics extracted from low dose CTs. Simulation paired low-high dose CTs trained generative models can be used to denoise low dose CT without re-training. Generative models can improve AUC by 0.05 of radiomics in survival predication and lung cancer diagnosis. Denoising using generative models seems to be a necessary pre-processing step for radiomic features from low dose CTs. Abstract: Purpose: Radiomics is an active area of research focusing on high throughput feature extraction from medical images with a wide array of applications in clinical practice, such as clinical decision support in oncology. However, noise in low dose computed tomography (CT) scans can impair the accurate extraction of radiomic features. In this article, we investigate the possibility of using deep learning generative models to improve the performance of radiomics from low dose CTs. Methods: We used two datasets of low dose CT scans – NSCLC Radiogenomics and LIDC-IDRI – as test datasets for two tasks – pre-treatment survival prediction and lung cancer diagnosis. We used encoder-decoder networks and conditional generative adversarial networks (CGANs) trained in a previous study as generative models to transform low dose CT images into full dose CT images. Radiomic features extracted from the original and improved CT scans were used to build two classifiers – a support vector machine (SVM) and a deep attention based multiple instance learning model – for survival prediction and lung cancer diagnosis respectively. Finally, we compared the performance of the models derived from the original and improved CT scans. Results: Denoising with the encoder-decoder network and the CGAN improved the area under the curve (AUC) of survival prediction from 0.52 to 0.57 (p-value < 0.01). On the other hand, the encoder-decoder network and the CGAN improved the AUC of lung cancer diagnosis from 0.84 to 0.88 and 0.89 respectively (p-value < 0.01). Finally, there are no statistically significant improvements in AUC using encoder-decoder networks and CGAN (p-value = 0.34) when networks trained at 75 and 100 epochs. Conclusion: Generative models can improve the performance of low dose CT-based radiomics in different tasks. Hence, denoising using generative models seems to be a necessary pre-processing step for calculating radiomic features from low dose CTs.