Jose Maria Mateos-Prez1, Juan Adan Guzmn de Villoria2, Irina Vidal-Migalln1, Manuel Desco1,3
1Medicina y Ciruga Experimental, Hospital
General Universitario Gregorio Maran, Madrid, Spain; 2Radiodiagnostico,
Hospital General Universitario Gregorio Maran, Madrid, Spain; 3Centro
de investigacin en red en salud mental (CIBERSAM), Madrid, Spain
This study compares two approaches for automatic classification of brain tumors
according to their grade: one using image-based variables (necrosis, gadolinium
uptake, neovascularization, haemorrhage, calcifications and edema) and another
one including quantitative data from functional studies (perfusion, difusion
and spectroscopy). 134 patients with intraaxial brain tumors of known grade
were recruited over 4 years in a Radiology Department. Two different
classifiers were made, one for each set of variables. No significant
differences were found between the classification accuracy. We conclude that
the data provided by functional studies does not necessarily increase
diagnostic accuracy in a real routine clinical setting.