S428
Clinical - Gynaecological
ESTRO 2026
metastasis received an additional boost of approximately 56 Gy of irradiation. We assessed the cumulative recurrence rate and the recurrence pattern. Results: This study included a total of 39 patients. The median age of the patients was 82 years (range: 70–96 years). According to the 2018 FIGO staging system, the patients were classified as follows: stage II (n = 5), stage III (n = 31), and stage IV (n = 3). The reasons for radiotherapy alone were as follows: older age (n = 34), renal dysfunction (n = 3), dementia (n = 1), and cerebrovascular disease (n = 1). The median prescribed doses for WP, CS, and brachytherapy were 30 Gy (range: 30–50), 20 Gy (range: 0–20), and 24 Gy (range: 7–30), respectively. Four patients received a median additional lymph node boost irradiation dose of 6 Gy (range: 6–10 Gy). The recurrence pattern was as follows: locoregional (n = 4; local [n = 3] and pelvic lymph node metastasis [n = 1]) and distant metastasis (n = 4; para-aortic lymph node metastasis [n = 2] and lung [n = 2]). The three-year cumulative locoregional recurrence rate was 13% (95%
1 Radiation Oncology, Ba ş ak ş ehir Çam and Sakura City Hospital, Istanbul, Turkey. 2 Mechatronics Engineering, Yıldız Technical University, Istanbul, Turkey. 3 Computer Engineering, Yıldız Technical University, Istanbul, Turkey. 4 Electronics and Communication Engineering, Yıldız Technical University, Istanbul, Turkey. 5 Radiology, Ba ş ak ş ehir Çam and Sakura City Hospital, Istanbul, Turkey Purpose/Objective: This study aimed to develop and evaluate an artificial intelligence–based model integrating clinical and radiomic features from pretreatment MRI to estimate early treatment response in patients with locally advanced cervical cancer undergoing definitive chemoradiotherapy. The goal was to provide a noninvasive tool for early identification of responders and non- responders before brachytherapy, potentially guiding individualized treatment adaptation. Material/Methods: This retrospective study included 68 patients treated with definitive concurrent chemoradiotherapy between January 2020 and June 2025. Pretreatment and pre- brachytherapy MRI scans were analyzed to assess early response, defined radiologically according to RECIST 1.1 criteria as complete (CR) or non-complete response (non-CR). Tumors were manually segmented on T2- weighted MRI sequences by a radiation oncologist and a radiologist using 3D Slicer software. Inter-observer agreement was evaluated using the intraclass correlation coefficient (ICC); segmentations with ICC ≥ 0.7 were combined, and others were re- segmented to reach consensus. All images were resampled to isotropic voxel size and intensity normalized before feature extraction. Radiomic features were obtained using the PyRadiomics platform compliant with IBSI standards. A total of 107 radiomic and six clinical variables were analyzed. Feature selection involved variance and correlation filtering (r > 0.9) followed by
CI: 10–25%). Conclusion:
These findings suggest that definitive RT alone provides acceptable local control in carefully selected elderly patients. Regarding recurrence patterns, local recurrence was the most common type of locoregional failure. Keywords: Elderly patients, radiotherapy alone
Digital Poster Highlight 1629 AI-Based Prediction of
Chemoradiotherapy Response Using Clinical and Radiomic Data in Locally Advanced Cervical Cancer Mehmet Halici 1 , İ rem Sayin 2 , Taha Unal 3 , Serkan Salturk 4 , Kimia Cepni 1 , Merve Sam Ozdemir 5 , Huseyin Uvet 2 , Aytul Hande Yardimci 5 , Huriye Senay Kiziltan 1
Made with FlippingBook - Share PDF online