PURPOSE: To develop computer-assisted image processing to identify the central sulcus from the MRI data sets in patients with brain tumors. MATERIALS AND METHODS: All MR images were acquired with a 3.0 Tesla scanner by three-dimensional (3D) spoiled gradient recalled acquisition using the steady state (SPGR) gradient-echo sequence and short inversion time inversion-recovery fast spin-echo sequence. The sulci were extracted automatically from reconstructed two-dimensional images of the cortical surface of 30 patients with brain tumors, and the extracted sulci were scored according to matching of the accepted anatomical features of the central sulcus. The candidates with the three highest scores were then superimposed on the 3D images. The correct position of the central sulcus was agreed by two senior neurosurgeons and one neuroradiologist. RESULTS: One of the three candidates corresponded to the central sulcus identified by the manual segmentation method for all 60 affected and unaffected hemispheres in the 30 patients. The candidate with the highest score coincided with the central sulcus in 26 of the 28 unaffected hemispheres and in 28 of the 32 affected hemispheres. CONCLUSION: The proposed method of computer-assisted image processing can provide accurate guidance to identifying the central sulcus in patients with brain tumors. 2008 Wiley-Liss, Inc
PURPOSE: To develop computer-assisted image processing to identify the central sulcus from the MRI data sets in patients with brain tumors. MATERIALS AND METHODS: All MR images were acquired with a 3.0 Tesla scanner by three-dimensional (3D) spoiled gradient recalled acquisition using the steady state (SPGR) gradient-echo sequence and short inversion time inversion-recovery fast spin-echo sequence. The sulci were extracted automatically from reconstructed two-dimensional images of the cortical surface of 30 patients with brain tumors, and the extracted sulci were scored according to matching of the accepted anatomical features of the central sulcus. The candidates with the three highest scores were then superimposed on the 3D images. The correct position of the central sulcus was agreed by two senior neurosurgeons and one neuroradiologist. RESULTS: One of the three candidates corresponded to the central sulcus identified by the manual segmentation method for all 60 affected and unaffected hemispheres in the 30 patients. The candidate with the highest score coincided with the central sulcus in 26 of the 28 unaffected hemispheres and in 28 of the 32 affected hemispheres. CONCLUSION: The proposed method of computer-assisted image processing can provide accurate guidance to identifying the central sulcus in patients with brain tumors. 2008 Wiley-Liss, Inc