Radiomics by Quantitative Diffusion-weighted MRI for Predicting Response in Patients with Extremity Soft-tissue Undifferentiated Pleomorphic Sarcoma

• DOI: 10.29328/journal.jro.1001066
• Publication Date: 2024-07-18T09:32:03Z
• Authors: Valenzuela RF; Duran-Sierra E; Canjirathinkal M; Amini B; Torres KE; Benjamin RS; Ma J; Wang WL; Hwang KP; Stafford RJ; Wu C; Zarzour AM; Bishop AJ; Lo S; Madewell JE; Kumar R; Murphy Jr WA; Costelloe CM
• Keywords: 03 medical and health sciences; 0302 clinical medicine

Purpose: This study aimed to determine the relevance of first- and high-order radiomic features derived from Diffusion-Weighted Imaging (DWI) Apparent Diffusion Coefficient (ADC) maps for predicting treatment response in patients with Undifferentiated Pleomorphic Sarcoma (UPS). Methods: retrospective included 33 extremity UPS pre-surgical DWI/ADC surgical resection. Manual volumetric tumor segmentation was performed on acquired at Baseline (BL), Post-Chemotherapy (PC), Post-Radiation Therapy (PRT). The percentage pathology-assessed effect (PATE) specimens categorized into responders (R; PATE ≥ 90%; 16 patients), partial-responders (PR; 89% - 31% PATE; 10 non-responders (NR; ≤ 30%; 7 patients). 107 were extracted BL, PC, PRT ADC maps. Statistical analyses compared R vs. PR/NR. Results: Pseudo-progression PC universal stability observed PR/NR based RECIST, WHO, assessments. At PRT, displayed a 35% increase mean (p = 0.0034), 136% decrease skewness 0.0001), 363% 90th percentile proportion 0.0009). Comparing statistically significant differences glrlm_highgraylevelrunemphasis 0.0081), glrlm_shortrunhighgraylevelemphasis 0.0138), gldm_highgraylevelemphasis glcm_sumaverage 0.0164), glcm_jointaverage glcm_autocorrelation 0.0193). firstorder_meanabsolutedeviation 0.0078), firstorder_interquartilerange 0.0109), firstorder_variance firstorder_robustmeanabsolutedeviation 0.0151) provided differences. Conclusion: Observing high post-therapeutic mean, low skewness, respect baseline is predictive successfully treated presenting > 90% PATE. Highly higher-order results include glrlm-highgraylevelrunemphasis (BL) first-order-mean absolute deviation (PC).