Changes in radiomic and radiologic features in meningiomas after radiation therapy

Autor: Sang Won Jo, Eun Soo Kim, Dae Young Yoon, Mi Jung Kwon
Jazyk: angličtina
Rok vydání: 2023
Předmět:
Zdroj: BMC Medical Imaging, Vol 23, Iss 1, Pp 1-14 (2023)
Druh dokumentu: article
ISSN: 1471-2342
DOI: 10.1186/s12880-023-01116-0
Popis: Abstract Objectives This study evaluated the radiologic and radiomic features extracted from magnetic resonance imaging (MRI) in meningioma after radiation therapy and investigated the impact of radiation therapy in treating meningioma based on routine brain MRI. Methods Observation (n = 100) and radiation therapy (n = 62) patients with meningioma who underwent MRI were randomly divided (7:3 ratio) into training (n = 118) and validation (n = 44) groups. Radiologic findings were analyzed. Radiomic features (filter types: original, square, logarithm, exponential, wavelet; feature types: first order, texture, shape) were extracted from the MRI. The most significant radiomic features were selected and applied to quantify the imaging phenotype using random forest machine learning algorithms. Area under the curve (AUC), sensitivity, and specificity for predicting both the training and validation sets were computed with multiple-hypothesis correction. Results The radiologic difference in the maximum area and diameter of meningiomas between two groups was statistically significant. The tumor decreased in the treatment group. A total of 241 series and 1691 radiomic features were extracted from the training set. In univariate analysis, 24 radiomic features were significantly different (P
Databáze: Directory of Open Access Journals