Exhibit Hall | Forum 9
Purpose: To investigate the changes in MRI iterative decomposition of water and fat with echo asymmetrical and least squares estimation (IDEAL IQ) and T2 fat suppression (T2fs) radiomics features (RFs) before and after concurrent chemo-radiotherapy (CCRT) for cervical cancer, quantitatively characterize pelvic bone marrow (PBM) injury and analyze the correlation between the changes in RFs and peripheral blood cell counts.
Methods: A total of 54 patients with cervical cancer who received CCRT were retrospectively analyzed. The patients underwent MRI IDEAL IQ and T2fs scanning at pre- and post-CCRT. PBM RFs were extracted from each region of interest at 5–10Gy, 10–15Gy, 15–20Gy, 20–30Gy, 30–40Gy, 40–50Gy, and >50Gy doses gradients, and the changes in peripheral blood cells counts during radiotherapy were evaluated. The RFs changes at different dose gradients between pre- and post-CCRT were compared. Pearson’s correlation analysis was performed to analyze the correlation between peripheral blood cell counts and changes in RFs at different dose gradients pre- and post-CCRT.
Results: The white blood cells (WBC), neutrophils (ANC), lymphocytes (ALC), platelets (PLT), and hemoglobin (HGB) counts during treatment decreased significantly. The first-order features showed a significant dose-response relationship, especially in FatFrac IDEAL IQ and WATER IDEAL IQ images (maximum slope=10.09, −7.93, respectively). Two features that firstorder-Range in FAT IDEAL IQ and glcm-DifferenceEntropy in T2fs showed a significant correlation between the changes in ANC counts under the 5-10Gy low dose gradient (r=0.744, 0.638, respectively).
Conclusion: IDEAL IQ radiomics of PBM showed a significant dose-response relationship and significantly correlated with the changes in peripheral blood cell counts in the low-dose irradiation region. The study showed that analyzing the changes in IDEAL IQ RFs provides an objective reference for bone marrow sparing in CCRT for cervical cancer.
Not Applicable / None Entered.
Not Applicable / None Entered.