P2‐108: USING COMPUTED TOMOGRAPHY TO ASSESS BRAIN VOLUMETRICS IN AGING

Abstract

Although magnetic resonance imaging (MRI) remains the gold standard for the noninvasive evaluation of white matter (WM), gray matter (GM) and cerebrospinal fluid (CSF) volumes in the aging brain, CT continues to be used widely for brain imaging, particularly when MRI is unavailable or contraindicated. In developing countries, CT is often the only imaging modality available for evaluation of brain atrophy in patients with mild cognitive impairment (MCI) or Alzheimer's disease (AD). This has renewed interest in the development of approaches to use head CT to estimate brain volumetrics. A brain segmentation approach was developed to delineate WM, GM and CSF from head CT using probabilistic, atlas-based classification. Feasibility and utility were evaluated by comparing MRI-only to CT-only segmentations in 10 older adults [mean (μ) ± standard deviation (σ) of age = 65 ± 7 yrs; 5 females] from whom both MRI and CT scans were acquired within an eight-week period. Segmentation similarity was quantified using the Dice coefficient (DC), a robust measure of inter-modality tissue classification agreement. Comparison of MRI vs. CT segmentations yielded normally-distributed DCs [μ ± σ across participants: 85.5% ± 4.6% (WM), 86.7% ± 5.6% (GM) and 91.3% ± 2.8% (CSF)], indicating satisfactory ability to calculate brain volumetrics from the CT scans of the participants, relative to MRI measurements. For this sample, bootstrapping suggests that the tissue classification method is sufficiently sensitive to estimate WM, GM and CSF volumes within ∼5%, ∼4% and ∼3% of their MRI-based values, respectively. Compared to MRI, volumes computed from CT displayed no evidence of systematic over- or under-estimation [t (9) = 0.89, p > 0.80]. Our contribution broadens the ability to integrate CT imaging findings with other research on brain aging in health and disease, and complements other methodologies for the study of brain volumetrics in neurodegenerative diseases, including AD.