S2485
Physics - Radiomics, functional and biological imaging, and outcome prediction
ESTRO 2026
Conclusion: A compartmental model for lymphocyte dynamics prediction in cervical cancer was optimized with dosimetric and ALC information of 38 LARC patients. Lymphocyte radiosensitivities found were consistent with literature [5]. The proposed model showed good ALC prediction accuracy, improving PTV volume-based predictions. The compartmental model, furthermore, facilitates exploring RIL mitigation strategies. References: [1] Damen PJJ et al. Int J Radiat Oncol Biol Phys 2021;111:936–48. [2] Liu H et al. Radiother Oncol 2019;131:52–9. [3] Kuipers SC et al. Phys Med Biol 2025;70:215005.[4] Beekman C et al. Phys Med Biol 2023;68:225007.[5] McCullum L et al. Int J Radiat Oncol 2023;116:1226–33. Keywords: lymphopenia, dose-response modeling, rectal cancer Poster Discussion 4417 Externally validating radiation pneumonitis radiomics signatures in chemoradio- immunotherapy for locally advanced Non Small Cell Lung Cancer Leyla Ebrahimpour 1 , Aditya Apte 1 , Narek Shaverdian 2 , Joseph O Deasy 1 , Maria Thor 1 1 Medical Physics, Memorial Sloan Kettering Cancer Center, NYC, USA. 2 Radiation Oncology, Memorial Sloan Kettering Cancer Center, NYC, USA
and ALC were predicted using the optimized per- patient models. Results: Thirty-eight LARC patients were selected. Median (range) ALC collection was 5 (2-7) samples. Mean(SD) radiosensitivity=0.75±0.41Gy-1. Median (IQR) RMSE of ALC predictions was 0.08(0.05)x109counts/L. Figure 1 shows a patient with median RMSE. RMSE of nadir predictions with PTV volume was higher (0.21x109counts/L)than that with the compartmental model (0.08x109counts/L). Modest but significant differences were found between nadir predictions with planned doses vs synthetic doses (Figure 2).
Purpose/Objective: Radiation pneumonitis (RP) remains the key
treatment-obstructing side effect in locally-advanced non-small cell lung cancer (LA-NSCLC) after combined chemo-radiotherapy (cCRT) and immunotherapy (IO). While routinely acquired pretreatment imaging could provide baseline RP risk indicators, such tools are not readily available. This work identified published pretreatment radiomics RP signatures after RT and externally validated them for RP and early-onset RP (RPEarly) after cCRT and IO for LA-NSCLC. Material/Methods: A literature search surveyed all previously published pretreatment, CT-based radiomics RP signatures. Radiomics signature eligibility criteria were: AUC >0.70, signature derived after cCRT for LA-NSCLC, and interpretable radiomics/modeling enabling external validation. The external validation cohort included 195 LA-NSCLC patients treated with cCRT, to 60Gy in 2 Gy- fractions, and follow-on IO in 2017-2022. Of these, 25% and 13% (N=48, and 26) developed grade ≥ 2 RP and RPEarly ( ≤ three months after completed cCRT). Each study’s anatomical region was defined in the external cohort, the planning-CT radiomics were extracted using the IBSI-anchored pyCERR, and predictive
Made with FlippingBook - Share PDF online