Application of the anatomical fiducials framework to a clinical dataset of patients with Parkinson's disease

Abstract

Establishing spatial correspondence between subject and template images is necessary in neuroimaging research and clinical applications such as brain mapping and stereotactic neurosurgery. Our anatomical fiducial (AFID) framework has recently been validated to serve as a quantitative measure of image registration based on salient anatomical features. In this study, we sought to apply the AFIDs protocol to the clinic, focusing on structural magnetic resonance images obtained from patients with Parkinson's disease (PD). We confirmed AFIDs could be placed to millimetric accuracy in the PD dataset with results comparable to those in normal control subjects. We evaluated subject-to-template registration using this framework by aligning the clinical scans to standard template space using a robust open preprocessing workflow. We found that registration errors measured using AFIDs were higher than previously reported, suggesting the need for optimization of image processing pipelines for clinical grade datasets. Finally, we examined the utility of using point-to-point distances between AFIDs as a morphometric biomarker of PD, finding evidence of reduced distances between AFIDs that circumscribe regions known to be affected in PD including the substantia nigra. Overall, we provide evidence that AFIDs can be successfully applied in a clinical setting and utilized to provide localized and quantitative measures of registration error. AFIDs provide clinicians and researchers with a common, open framework for quality control and validation of spatial correspondence and the location of anatomical structures, facilitating aggregation of imaging datasets and comparisons between various neurological conditions.

Keywords: Accuracy; Biomarker; Deep brain stimulation; Fiducials; Parkinson’s disease; Registration.

Fig. 1

Schematic of workflow to obtain localization errors (above), and registration errors (below). In summary, 5 raters placed 32 anatomical fiducials (AFIDs) on each clinical image (blue). The mean location was calculated for each AFID (green), and the Euclidean distance from each rater’s placement was calculated (termed the localization error). Each rater independently placed AFIDs on the MNI images, and the mean location was calculated (purple). Rater placed AFIDs were transformed to MNI space. The Euclidean distance between each rater’s transformed AFID to the mean location of that AFID placed in MNI space was calculated and termed real-world registration error. Each mean AFID placement on the clinical images was transformed to MNI space, its Euclidean distance to that AFID placed in MNI space was calculated and termed consensus registration error

Fig. 2

Mean anatomical fiducial localization error (AFLE) for each anatomical fiducial (AFID) and subject. Bottom colormap represents mean AFLEs across all raters for each AFID and subject, illustrating the distribution of AFLEs across all subjects and AFIDs. Top bar graph represents the mean AFLEs for each AFID across all 39 subjects + standard deviation. AFIDs 1, 2 had the lowest AFLEs, while AFIDs 25 and 26 had the greatest AFLEs

Fig. 3

Mean real-world anatomical registration error (AFRE) for each anatomical fiducial (AFID) and subject. Bottom colormap represents mean non-linear AFREs across all raters for each AFID and subject, illustrating the distribution of non-linear AFREs across all subjects and AFIDs. Top bar graph represents the mean non-linear AFRE for each AFID across all 39 subjects + standard deviation. AFIDs 1, 2, 11, 12, 13, 31 and 32 had decreased AFREs across most subjects. AFIDs 29 and 30 had large AFREs across most subjects

Fig. 4

Summary of mean pairwise distances between each anatomical fiducial (AFID) with significant differences. Bottom right shows heatmap representing the difference between mean pairwise distances between each AFID for OASIS-1 subjects and Parkinson’s disease (PD) patients. Positive differences represent a greater pairwise distance in the OASIS-1 subjects relative to PD patients. Significant differences illustrated in top left of figure, designated by a black box. Significance is determined by Wilcoxon rank-sum tests with Bonferroni correction, with a significance threshold of 0.05/496. 40 pairwise distances reached thresholds of statistical significance between PD vs controls (see Online Resource 4 for details)