Haney S, Thompson PM, Alger JR, Cloughesy TF, Toga AW
Laboratory of Neuro Imaging, Dept. Neurology, Division of Brain Mapping,
UCLA School of Medicine, Los Angeles CA 90095, USA,
UCLA Dept. of Radiological Sciences, and
Neuro-Oncology Program, UCLA School of Medicine
Tissue classification approaches (described below) can create maps of different tissue types
in the brain.
Together with other imaging approaches designed to assess the chemical and genetic content of brain lesions, tissue maps can be used to measure growth rates, determine response to therapy, and track subtle changes in the dynamics of tumor growth.
Summary. As part of a comprehensive longitudinal study of patients with high-grade gliomas, a variety of novel pattern recognition and 3-dimensional image analysis algorithms were applied to serial MRI scans. Rates of change were tracked in tumor, necrosis, and gadolinium-enhancing tissue as well as peritumoral edema, adjacent white matter, and cystic compartments, to identify optimal algorithms and scanning protocols.
3D T2-weighted and gadolinium-enhanced T1-weighted (256x256 resolution; 3 mm spacing) SPGR MRI volumes were acquired over a 3.5 year period (6 week to 6 month intervals) from 12 patients with glioblastoma multiforme (GBM; age: 4-54 yrs. at first scan). Scans were automatically aligned into Talairach stereotaxic space with 6-parameter rigid transformations. Population-based tissue maps, containing probabilistic information on tissue locations in stereotaxic space, were automatically aligned with the scan data, adjusted for herniation effects with non-linear registration, and used to determine a Gaussian mixture distribution reflecting the intensities of specific tissue classes at each time-point in the scan series. A nearest neighbor algorithm was used to differentiate tissue types, and its accuracy confirmed by tagging 160 tags per anatomical region (170 when a cystic compartment was present). Tissue maps were manually adjusted to better delineate class boundaries and the results of automated and manual segmentations were compared.
Results. T1-based segmentations provided excellent tissue differentiation and growth rate evaluation for tumor, necrosis, cysts and enhancement. Volumetric changes in tumor tissue ranged from -73% to +2600% corresponding to a halving time of 130 days and a doubling time of 13.8 days. Vasogenic edema, however, was optimally detected by classifying T2-weighted scans. Absolute edema volumes, vital in monitoring blood-brain barrier integrity and extravasation of plasma water, were systematically 16.5 +/- 8.8% lower in T1-based maps (average volume: 70.5 +/- 55.9 cm3) than in T2-based maps (average volume: 82.5 +/- 63.3 cm3; p < 0.001, paired t-test). Nonetheless, volumetric changes detected by each type of map were highly correlated (r2=0.98).
Conclusions. Automated tissue mapping provides significant power in tracking the regional dynamics of multiple tissue classes in glioma patients. The resulting tissue classification and growth rate detection systems also provide a 3D structural framework to correlate spectroscopic, diffusion imaging, and histopathologic indices in interventional studies. They may also be useful for quantitative evaluation of therapeutic response in individual patients and clinically-defined groups.
Grant Support: (to P.T. and A.W.T.): NIMH/NIDA (P20 MH/DA52176), P41 NCRR (RR13642); (A.W.T.): NLM (LM/MH05639), NSF (BIR 93-22434), NCRR (RR05956) and NINCDS/NIMH (NS38753).
|RESUME| E-MAIL ME| PERSONAL HOMEPAGE| PROJECTS|