CT-based volumetric measures obtained through deep learning: Association with biomarkers of neurodegeneration

Abstract

Introduction: Cranial computed tomography (CT) is an affordable and widely available imaging modality that is used to assess structural abnormalities, but not to quantify neurodegeneration. Previously we developed a deep-learning-based model that produced accurate and robust cranial CT tissue classification.

Materials and methods: We analyzed 917 CT and 744 magnetic resonance (MR) scans from the Gothenburg H70 Birth Cohort, and 204 CT and 241 MR scans from participants of the Memory Clinic Cohort, Singapore. We tested associations between six CT-based volumetric measures (CTVMs) and existing clinical diagnoses, fluid and imaging biomarkers, and measures of cognition.

Results: CTVMs differentiated cognitively healthy individuals from dementia and prodromal dementia patients with high accuracy levels comparable to MR-based measures. CTVMs were significantly associated with measures of cognition and biochemical markers of neurodegeneration.

Discussion: These findings suggest the potential future use of CT-based volumetric measures as an informative first-line examination tool for neurodegenerative disease diagnostics after further validation.

Highlights: Computed tomography (CT)-based volumetric measures can distinguish between patients with neurodegenerative disease and healthy controls, as well as between patients with prodromal dementia and controls. CT-based volumetric measures associate well with relevant cognitive, biochemical, and neuroimaging markers of neurodegenerative diseases. Model performance, in terms of brain tissue classification, was consistent across two cohorts of diverse nature. Intermodality agreement between our automated CT-based and established magnetic resonance (MR)-based image segmentations was stronger than the agreement between visual CT and MR imaging assessment.

Keywords: CSF biomarkers; CT; brain segmentation; cognition; deep learning; dementia; plasma biomarkers.

FIGURE 1

Flowchart showing the details of (A) the Gothenburg H70 Birth Cohort and (B) Memory Clinic Cohort, Singapore. Aβ, amyloid beta; CSF, cerebrospinal fluid; CT, computed tomography; MRI, magnetic resonance imaging; MTA, medial temporal atrophy; NfL, neurofilament light; p‐tau, phosphorylated tau (Color print).

FIGURE 2

Imaging‐based volumetric measures. The CT‐based GM, WM, CSF, ICV and VCSF maps were automatically segmented using deep‐learning techniques. The MR images were automatically segmented to GM, WM, and CSF maps using SPM12, ICV map using Pincram, and VCSF map using MAPER. Six volumetric measures were determined from these segmentation maps. CSF/VCSF adjusted BV/GM volumetric measures were obtained using regression‐based adjustment of the respective brain tissue volumes. CSF, cerebrospinal fluid; CT, computed tomography; ICV, intracranial volume; GM, gray matter; MRI, magnetic resonance imaging; WM, white matter; VCSF, ventricular cerebrospinal fluid (Color print).

FIGURE 3

ROC curves for (A) dementia versus CN and (B) prodromal dementia versus CN using CT‐ and MR‐based volumetric measures in the Memory Clinic Cohort, Singapore (n = 204). All volumes were adjusted for intracranial volume. The ROC curves of variables with AUC above 0.6 are shown. AUC, area under the curve, BV, brain volume; CSF, cerebrospinal fluid; CN, cognitively normal; GM, gray matter; ROC, receiver‐operating characteristic (Color print).

FIGURE 4

Distribution of CT‐ and MR‐based volumetric measures across diagnostic groups in the Memory Clinic Cohort, Singapore (n = 204). ***p < 0.001, **p < 0.01, *p < 0.05; uncorrected p‐values derived from Kruskal–Wallis test. All volumes were adjusted for intracranial volume. BV, brain volume; CN, cognitively normal; CSF, cerebrospinal fluid; GM, gray matter (Color print).

FIGURE 5

CT‐ and MR‐based volumes with other neurodegenerative disease biomarkers. Correlation of CT‐ and MR‐based GM volumes with (A) MMSE, (B) CDR‐SB, (C) plasma NfL, and (D) plasma p‐tau181 in the Memory Clinic Cohort, Singapore (n = 204). The correlation values (ρ) were obtained from partial rank correlation analysis between CT‐based volumetric measures and other biomarkers controlled for intracranial volume, age, gender and education. CDR‐SB, Clinical Dementia Rating Sum of Boxes; GM, gray matter; MMSE, Mini‐Mental State Examination; MR, magnetic resonance; NfL, neurofilament light; p‐tau, phosphorylated tau (Color print).