\ l t S I K . \ ( ' l \ * l ) .u'tivc surlace tnethod tcehmque is . ipplu-d lo analvse dil ' lerent medieul inniges and to make ,i -IM i l M i . i i n u cauhac mndcl. Gradient Veetor ilow M Λ l ) is uscil to generale external image force. The . i l i ' . o i i t l u n o l ( i \ T is optiini/cd by using Gauss-Seidel a iu l StkvcssiNc Overrelaxation methods (SOR) to solve tlk· p . n i i a l d i l ' l c rcnt ia l eijuation [1]. The different image louvs sho\\ t h a t ( i V F method has udvantage in piopaLMtmi; cdge Information of object boundarics into cxiended areas near the target contours. Two kinds of p ieprocess ing procedures, i.e., Sobcl and CannyIKiu -he -Monga methods, are used to cnhance the iMadicm i n t o r n i a t i o n in noisy images. The proposcd meihod is applied to cl inical images, e.g., MRI eardiac unagcs of r igh t ventriclc, endocardial and epicardial \ \ . i l K o l ' l e f t ventriele. 1t delivers accuratc scgmentation rcs i i l t s m ohjeet surface extraction. The segmentation ivsii l ts are visualized through shape models, which are dcsuibcJ by inangle meshes. The proposed algorithm is implemented m Unix environment using C++.