Automatic Correction ofthe3DOrientation oftheBrainImagery

Classification ofhumanbrain pathologies canbe guidedbytheestimation ofthedeparture of3D internal structures fromthenormalbilateral symmetry. However symmetry based analysis can'tbeprecisely carried outwhenthe 3D brain orientation ismisaligned, acommonoccurrence in clinical practice. Inthis paper, a technique toautomatically identify thesymmetry planeandcorrect the3Dorientation of volumetric brain images inacosteffective wayisdeveloped. Thealgorithm seeks thebest sampling strategies torealign 3D volumetric representation ofthebrain within scanner coordinate system. Theinertia matrix iscomputed onthesampled brain, andtheprinciple axesarederivedfrom theeigenvectors ofthe inertia matrix. Thetechnique isdemonstrated onMR andCT brainimagesandthedetected symmetry planethatis orthogonal totheprinciple vectors isprovided. A spatial affine transform isapplied torotate the3Dbrain images andalign themwithin thecoordinate system ofthescanner. Thecorrected brain volume isre-sliced suchthat eachplanar image represents thebrain atthesameaxial level.