ESTRO 2024 - Abstract Book

S1155

Clinical - Haematology

ESTRO 2024

Keywords: Indolent, Extranodal, Lymphoma

References:

Teckie S, Qi S, Lovie S et al. Long-term outcomes and patterns of relapse of early-stage extranodal marginal zone lymphoma treated with radiation therapy with curative intent. Int J Radiat Oncol Biol Phys 2015; 92: 130 – 137.

Teckie S, Qi S, Chelius M et al. Long-term outcomes of 487 patients with early-stage extra-nodal marginal zone lymphoma. Annals of Oncology 2017; 28: 1064-1069.

1507

Poster Discussion

Assessment of lymph node area coverage with TMI and implementation of TMLI using deep learning

Hyeon seok Choi 1,2 , Hyun-Cheol Kang 1,2 , Eui Kyu Chie 1,2 , Kyung Hwan Shin 1,2 , Ji Hyun Chang 1,2 , Bum-Sup Jang 1,2

1 Seoul National University Hospital, Department of Radiation Oncology, Seoul, Korea, Republic of. 2 Seoul National University College of Medicine, Department of Radiation Oncology, Seoul, Korea, Republic of

Purpose/Objective:

Total marrow irradiation (TMI) and total marrow and lymphoid irradiation (TMLI) are administered before hematopoietic stem cell transplantation. Compared with total body irradiation, these radiation therapies have the advantages of minimizing side effects and increasing the prescribed dose by reducing the radiation dose to organs at risk. However, delineating target lesions across the entire body according to TMI and TMLI plans is labor-intensive and time-consuming. In addition, although the delineation of target lesions between TMI and TMLI differs, the clinical distinction is not clear, and the lymph node (LN) coverage area during TMI remains uncertain. Accordingly, this study calculates the LN area coverage according to the TMI plan. Further, a deep learning-based model for delineating whole-body regional LN areas is trained and evaluated.

Material/Methods:

Whole-body regional LN areas (cervical, axillary, mediastinal, para-aortic, common iliac, external iliac, internal iliac, obturator, presacral, and inguinal LN areas) were manually contoured and confirmed by three radiation oncologists in patients treated according to a TMI plan. The dose coverage of the delineated LN areas in the TMI plan was estimated using V100% (the percentage of volume receiving 100% of the prescribed dose), V95%, and V90%. To train the deep learning model for automatic segmentation, additional whole-body computed tomography data were obtained from other patients to delineate the whole-body LN areas. The patients and data were divided into training/validation and test groups. Moreover, models were developed using the “nnU - NET” framework. The trained models were evaluated using Dice similarity coefficient (DSC), precision, recall, and Hausdorff distance 95 (HD95).