ESTRO 2024 - Abstract Book

S5043

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

Conclusion:

Conclusion: The deep learning model based on pre-treatment multi-sequence MRI-radiomics and planed dose-omics of re-irradiation therapy can be used to predict necrotic prognosis as a potential indicator for planning the re irradiation therapy of individuals with recurrent nasopharyngeal carcinoma.

Keywords: Recurrent nasopharyngeal carcinoma;Necrosis;MRI

1422

Digital Poster

Development of homodogy prediction method of radiation pneumonitis using lung CT ventilation image

Xiao Mei Hiromi Foo, Noriyuki Kadoya, Ryota Tozuka, Taichi Hoshino, WingYi Lee, Shohei Tanaka, Yoshiyuki Katsuta, Kazuhiro Arai, Keiichi Jingu

Tohoku University Graduate School of Medicine, Radiation Oncology, Miyagi, Japan

Purpose/Objective:

Lung cancer is a leading cause of global cancer mortality and over 50% of lung cancer patients receive radiotherapy as part of the treatment plan. One of the side effects of radiotherapy is radiation pneumonitis (RP). RP affects between 10% and 40% of people who undergo radiation therapy for lung cancer [1]. Predicting RP with high accuracy is of paramount importance as one can optimize treatment plan, such that it provides tumor tissues with sufficient doses while sparing the normal tissues from excessive radiation effects based on expected risk. Thus, topological data analysis (TDA), and in particular, dose weighted homology (homodogy), is adopted in this study to develop novel approach to feature extraction through image analysis which makes use of quantification of global structures in data and dose parameters. In this study, we investigated the effectiveness of homodogy-based method developed in predicting RP in lungs using CT ventilation images by comparing the outcome to conventional radiomics method.