Program Information
Evaluation of Radiomic Features in MRI of Acoustic Neuromas
G Narayanasamy1*, G Zhang2 , G Campbell3 , E Siegel4, E Moros2 , X Zhang1 , S Morrill1 , J Penagaricano1 , (1) Department of Radiation Oncology, University of Arkansas for Medical Sciences, Little Rock, AR, (2) H. Lee Moffitt Cancer Center, Tampa, FL, (3) College of Medicine, University of Arkansas for Medical Sciences, Little Rock, AR, (4) Department of Biostatistics, University of Arkansas for Medical Sciences, Little Rock, AR
Presentations
WE-RAM3-GePD-J(B)-1 (Wednesday, August 2, 2017) 10:30 AM - 11:00 AM Room: Joint Imaging-Therapy ePoster Lounge - B
Purpose: The aim of this study was to evaluate prognostic properties of radiomic features extracted from MR images with the progression of acoustic neuromas (AN) based on volume changes and audiology hearing examinations.
Methods: The study was based on MR images acquired prior and post Gamma Knife (GK) treatment of thirty two patients who underwent stereotactic radiosurgery (SRS) for AN between 2010 and 2015. All patients received the same dose of 12 Gy to 50 % isodose line. Radiomic features extracted using a software coded in Matlab (ver R2016a) include features based on Intensity, Fractals, Laplacian of Gaussian and textural co-occurrence, Run-length, Gray-level size zone, and Neighborhood Gray-tone difference matrices. Audiology report findings were graded on a Gardner-Robertson (GR) hearing scale (1–good, 2–serviceable, 3–non-serviceable, 4–poor, and 5–deaf) based on pure-tone average and word recognition score. Audiology reports acquired prior and post GK were used to validate against Radiomic features independently and establish prognostic value. The responders were also classified into three groups based on the change in lesion volume.
Results: The mean±SD time gap between 2 MRI examinations before and after GK were 268±260 days. With less than 10% lesion volume change, 13/32 lesions were considered stable; 15/32 were poor responders with at least 10% increase in lesion volume, and 4/32 were complete responders with at least 10% volume shrinkage. On a multivariate analysis, three variables, (complexity, convexity and sphericity) displayed week correlation with changes in lesion volume (R2=0.22). Based on 14 pairs of audiology reports, univariate correlation analysis was insufficiently powered to establish a relationship between GR hearing score and Radiomic features.
Conclusion: This initial Radiomic study on AN finds no correlation between the Radiomic features, lesion volume and audiological findings; however, it was insufficiently powered to establish relationship of Radiomics with GR hearing score.
Contact Email: