Abstract Introduction Pelvic radiotherapy exposes female erectile and vaginal tissues and their supplying neurovasculature to radiation, contributing to long-term sexual dysfunction in gynecologic cancer survivors. Current assessments are subjective and lack objective biomarkers to quantify tissue injury. Quantitative ultrasound radiomics may enable noninvasive, reproducible characterization of radiation-induced microstructural changes in female erectile tissues. Objective To assess the feasibility of transvaginal ultrasound-based Gray Level Co-occurrence Matrix (GLCM) radiomics for quantifying radiation-related changes in the corpus cavernosum (CC) and corpus spongiosum (CS) of gynecologic cancer patients, with proximal–distal vaginal wall comparisons used for biological validation of the method. Methods In this prospective pilot study, four patients with cervical or uterine cancer underwent transvaginal ultrasound before and 1–7 months after pelvic radiotherapy. Regions of interest were manually segmented in the CC, CS, and in 3.0 mm proximal and distal anterior vaginal wall segments (delineated at the bladder neck, with 2cm washout) using 3D Slicer. Eleven GLCM features were extracted per structure (Autocorrelation, Cluster Tendency, Contrast, Correlation, Difference Entropy, Difference Variance, Inverse Difference Moment Normalized (Idmn), Inverse Difference Normalized (Idn), Inverse Variance, Joint Energy, and Joint Entropy). Baseline and follow-up values were descriptively compared, focusing on direction and consistency of change rather than statistical significance. Results Feature extraction from the CC and CS was technically feasible in all participants, demonstrating stable segmentation and radiomic quantification across scans. Preliminary results showed directionally consistent changes in erectile tissue texture metrics, including decreased autocorrelation and increased cluster tendency, suggestive of increased heterogeneity post-radiation. In the proximal vaginal wall-expected to receive higher radiation dose-GLCM changes were concordant with early fibrotic remodeling (e.g., increased contrast and reduced homogeneity), while distal wall features remained stable. This proximal–distal difference supports the biological plausibility of the observed CC/CS signal changes. Conclusions Transvaginal ultrasound GLCM radiomics of the corpus cavernosum and corpus spongiosum is feasible and yields biologically consistent post-radiation textural alterations. Proximal–distal vaginal comparisons confirm the analytic method’s sensitivity to radiation-related tissue effects, supporting further study of ultrasound radiomics as a quantitative biomarker for sexual tissue toxicity after pelvic radiotherapy. Disclosure No
Gupta et al. (2026) studied this question.