Less is More: Efficient PET/CT Segmentation and Multimodal Prediction of Recurrence-Free Survival and HPV Status in Head and Neck Cancer

Lishan Cai, Xinglong Liang, Tianyu Zhang, Jiaju Huang, Tao Tan, Yunchao Yin

Keywords: Medical Image Analysis，HECKTOR challenge，Deep Learning，Segmentation，Recurrence-Free Survival，HPV Status

TL;DR: In this study, Team MEDAI presents our solutions for the three HECKTOR 2025 tasks: segmentation, recurrence-free survival prediction, and HPV status classification.

Abstract:

ccurate delineation of primary head and neck tumors and
metastatic lymph nodes on PET/CT is critical for radiotherapy planning
and prognostic assessment. Building on this clinical need, the HECK-
TOR 2025 challenge uses a large multi-centric dataset to provide a com-
prehensive benchmark for multimodal methods that integrate imaging
and clinical information across three key tasks: segmentation of the pri-
mary tumor and involved lymph nodes, recurrence-free survival predic-
tion, and HPV status classification. In this study, we (Team MEDAI)
present our solutions for all three challenge tasks. For automated tu-
mor segmentation, we employed an ensemble of ten lightweight STU-Net
(small) models, achieving efficient and precise delineation of both pri-
mary tumors and metastatic lymph nodes. For recurrence-free survival
prediction and HPV status classification, we developed a multimodal
framework that integrates volumetric PET/CT imaging, lesion masks
derived from the segmentation models, and structured clinical variables.
Code is available at https://github.com/Liiiii2101/HECKTOR2025-MEDAI.
Team: MEDAI.

Submission Number: 6