Texture analysis of magnetic resonance images to predict histologic grade of meningiomas in dogs
2017
Banzato, Tommaso | Bernardini, Marco | Cherubini, Giunio B. | Zotti, Alessandro
OBJECTIVE To predict histologic grade of meningiomas in dogs via texture analysis (TA) of MRI scans of the brain and spinal cord. SAMPLE 58 sets of MRI scans of the brain and spinal cord of dogs with histologically diagnosed meningioma. PROCEDURES MRI sequences were divided into a training set and a test set, and results of histologic assessment were obtained. Tumors were histologically grouped as benign (stage I) or atypical-anaplastic (stage II or III). Texture analysis was performed by use of specialized software on T2-weighted (T2W) and pre- and postcontrast T1-weighted (T1W) images. A set of 30 texture features that provided the highest discriminating power between the 2 histologic classes in the training set was automatically selected by the TA software. Linear discriminant analysis was performed, and the most discriminant factor (MDF) was calculated. The previously selected texture features were then used for linear discriminant analysis of the test set data, and the MDF was calculated. RESULTS For the training set, TA of precontrast T1W images provided the best diagnostic accuracy; a cutoff MDF of < 0.0057 resulted in a sensitivity of 97.4% and specificity of 95.0% for discriminating benign from atypical-anaplastic meningiomas. Use of postcontrast T1W and T2W images yielded poorer diagnostic performances. Application of the MDF cutoff calculated with the training set to the MDF calculated with the test set provided a correct classification rate of 96.8% for precontrast T1W images, 92.0% for postcontrast T1W images, and 78.9% for T2W images. CONCLUSIONS AND CLINICAL RELEVANCE Findings supported the potential clinical usefulness of TA of MRI scans for the grading of meningiomas in dogs.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Cette notice bibliographique a été fournie par National Agricultural Library
Découvrez la collection de ce fournisseur de données dans AGRIS
