Details

Autor(en) / Beteiligte

• Vinitski, Simon
• Iwanaga, Tad
• Gonzalez, Carlos
• Andrews, David
• Knobler, Robert
• Mack, John
• Del Bimbo, Alberto

Titel

Fast tissue segmentation based on a 4D feature map: Preliminary results

Ist Teil von

• Image Analysis and Processing, 2005, p.445-452

Ort / Verlag

Berlin, Heidelberg: Springer Berlin Heidelberg

Erscheinungsjahr

2005

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• The primary aim of this work was to develop a fast and accurate method for tissue segmentation based on a 4D feature map to be used in stereotactic neurosurgery and the evaluation of multiple sclerosis, MS. Secondly, we wanted to validate our method with biological tissue studied in-vivo obtained by biopsy. Tissue segmentation based on both 3D and 4D feature maps were derived from high resolution MR images and was performed in five normal individuals, six patients with MS plaques in the brain, and six patients with malignant brain tumors from which four had undergone stereotactic biopsy. Three inputs: proton density, T2, and T1-weighted MR images were routinely utilized. As a fourth input, magnetization transfer was used in some patients, and T1-weighted post contrast MRI in others. To speed up computation, our k-Nearest Neighbor segmentation algorithm was optimized by: 1) discarding redundant seed points, 2) discarding points within one half of a standard deviation from the cluster center that were non-overlapping with other tissue classes, and 3) discarding outlying seed points located beyond five standard deviations from the cluster center of each tissue class. After segmentation, a stack of color-coded segmented images was created. Our new technique, utilizing all four MRI inputs provided better segmentation than that based on only three inputs. The tissues were smoother due to the reduction of statistical noise, and the delineation of the tissues was increased. Details that were previously blurred or invisible now became apparent. For example, in normal persons detailed depiction of deep gray matter nuclei was obtained. In malignant tumors, up to five abnormal tissues were identified: 1) solid tumor core, 2) cystic tumor, 3) white matter edema, 4) gray matter edema, and 5) tissue necrosis. Subsequent stereotactic biopsy and histological analysis confirmed the results of the tissue segmentations. In MS patients, delineation of MS plaque became much sharper. In conclusion, the proposed 4D methodology warrants further development and clinical evaluation.