Early detection of Alzheimer's disease using MRI hippocampal texture

Abstract

Cognitive impairment in patients with Alzheimer's disease (AD) is associated with reduction in hippocampal volume in magnetic resonance imaging (MRI). However, it is unknown whether hippocampal texture changes in persons with mild cognitive impairment (MCI) that does not have a change in hippocampal volume. We tested the hypothesis that hippocampal texture has association to early cognitive loss beyond that of volumetric changes. The texture marker was trained and evaluated using T1-weighted MRI scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database, and subsequently applied to score independent data sets from the Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBL) and the Metropolit 1953 Danish Male Birth Cohort (Metropolit). Hippocampal texture was superior to volume reduction as predictor of MCI-to-AD conversion in ADNI (area under the receiver operating characteristic curve [AUC] 0.74 vs. 0.67; DeLong test, p = 0.005), and provided even better prognostic results in AIBL (AUC 0.83). Hippocampal texture, but not volume, correlated with Addenbrooke's cognitive examination score (Pearson correlation, r = -0.25, p < 0.001) in the Metropolit cohort. The hippocampal texture marker correlated with hippocampal glucose metabolism as indicated by fluorodeoxyglucose-positron emission tomography (Pearson correlation, r = -0.57, p < 0.001). Texture statistics remained significant after adjustment for volume in all cases, and the combination of texture and volume did not improve diagnostic or prognostic AUCs significantly. Our study highlights the presence of hippocampal texture abnormalities in MCI, and the possibility that texture may serve as a prognostic neuroimaging biomarker of early cognitive impairment.

Keywords: biomarker; classification; early diagnosis; hippocampus; image analysis; machine learning; magnetic resonance imaging; mild cognitive impairment.

Figure 1

Diagnostic and prognostic ADNI results. (A) Box plots of the hippocampal texture scores for the diagnostic groups. The central line marks the median; the lower and upper edges of the box mark the 25th percentile (q1) and the 75th percentile (q3), respectively; the notch marks the 95% confidence interval of the median as ±1.57 (q3 − q1)/n ^½, where n is the number of observations; the whiskers mark the most extreme inlier data points; and the circles mark outliers defined as >q3 + 1.5 (q3 − q1) or <q1 − 1.5 (q3 − q1). (B) Box plots of the hippocampal texture scores for the prognostic groups. The upper and lower dashed horizontal lines mark the median hippocampal texture score of the AD and CTRL group, respectively. (C) ROC curves for AD diagnosis and AD prognosis. The AUCs are (p‐values according to a DeLong, DeLong, and Clarke–Pearson's test in parentheses): CTRL vs AD 0.912 (p < 0.001), CTRL vs MCI 0.764 (p < 0.001), MCI‐NC12 vs MCI‐C12 0.740 (p < 0.001), MCI‐NC24 vs MCI‐C24 0.742 (p < 0.001).

Figure 2

Illustration of hippocampi with a negative texture score (indicating CTRL) versus hippocampi with a positive texture score (indicating AD) for different absolute hippocampus volumes. (A) AD with small volume and positive texture score. (B) AD with medium volume and positive texture score. (C) AD with large volume and positive texture score. (D) CTRL with small volume and negative texture score. (E) CTRL with medium volume and negative texture score. (F) CTRL with large volume and negative texture score.

Figure 3

Scatter plots of MRI biomarkers vs MMSE score in the ADNI cohort. Crosses correspond to CTRL, dots correspond to MCI, and asterisks correspond to AD. (A) Hippocampal texture. (B) Hippocampal volume. Note that uniform random noise in the range of −0.5 to 0.5 was added to each MMSE score for better visualization.

Figure 4

Scatter plots of MRI biomarkers vs ACE score in the Metropolit cohort. Crosses correspond to CTRL and asterisks correspond to CL. (A) Hippocampal texture. (B) Hippocampal volume. Note that uniform random noise in the range −0.5 to 0.5 was added to each ACE score for better visualization.

Figure 5

Scatter plots of MRI biomarkers vs metabolic rate of glucose in the hippocampus in FDG‐PET. Crosses correspond to CTRL, dots correspond to MCI, and asterisks correspond to AD. (A) Hippocampal texture. (B) Hippocampal volume.

Figure 6

Schematic view of the proposed texture working hypothesis in AD. Top row: NFTs inside the neurons and Aβ plaques between neurons spread throughout the brain, causing neuronal death. Middle row: changes in the statistical properties of the image intensities due to the accumulated effect of NFTs and/or Aβ plaques may be reflected as certain textural patterns prior to atrophy. Bottom row: atrophy manifests as the shrinkage and possible morphological change of brain structures.