Radiomics study using image features of dose distribution and intra-treatment CBCT

2020 Fiscal Year Final Research Report

• Project/Area Number: 17K15799
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Medical Physics and Radiological Technology
• Research Institution: The University of Tokyo
• Principal Investigator: Nawa Kanabu 東京大学, 医学部附属病院, 助教 (00456914)
• Project Period (FY): 2017-04-01 – 2021-03-31
• Keywords: レディオミクス解析 / 画質改善 / 機械学習 / 深層学習

Outline of Final Research Achievements

We studied the radiomics analysis to establish the new pattern recognition system of accurately predicting the response and prognosis of treatment from image features in medical images such as planning CT, dose distribution, and intra-treatment CBCT. The image quality improvement of medical images is an essential technology for extracting features with high accuracy. As a preprocessing for feature extraction, we constructed a method to improve the image quality of CBCT using deep learning. We investigated the effect of tumor delineation for the values of extracted features. We also examined the dependence on the dose calculation algorithm and the calculation grid size for the features extracted from the dose distribution.

Free Research Field

医学物理

Academic Significance and Societal Importance of the Research Achievements

レディオミクス解析における画質改善の重要性を明らかにし、深層学習を用いた医用画像の画質改善の手法を構築した。また線量分布に対するレディオミクス解析について線量計算のパラメターに対する詳細な検討を行った。