Developing an automated skeletal phenotyping pipeline to leverage biobank-level medical imaging databases

Abstract

Objectives: Collecting skeletal measurements from medical imaging databases remains a tedious task, limiting the research utility of biobank-level data. Here we present an automated phenotyping pipeline for obtaining skeletal measurements from DXA scans and compare its performance to manually collected measurements. Materials and Methods: A pipeline that extends and modifies the Advanced Normalization Tools (ANTs) framework was developed on 341 whole-body DXA scans of UK Biobank South Asian participants. A set of 10 measurements throughout the skeleton was automatically obtained via this process, and the performance of the method was tested on 20 additional DXA images by calculating percent error and concordance correlation coefficients (CCC) for manual and automated measurements. Stature was then regressed on the automated femoral and tibia lengths and compared to published stature regressions to further assess the reliability of the automated measurements. Results: Based on percent error and CCC, the performance of the automated measurements falls into three categories: poor (sacral and acetabular breadths), variable (trunk length, upper thoracic breadth, and innominate height), and high (maximum pelvic aperture breadth, bi-iliac breadth, femoral maximum length, and tibia length). Stature regression plots indicate that the automated measurements reflect realistic body proportions and appear consistent with published data reflecting these relationships in South Asian populations. Discussion: Based on the performance of this pipeline, a subset of measurements can be reliably extracted from DXA scans, greatly expanding the utility of biobank-level data for biological anthropologists and medical researchers.

Developing an automated skeletal phenotyping pipeline to leverage biobank‐level medical imaging databases

“…Table S4 . Cataldo‐Ramirez et al (2022) Table 3 replicated. Percent error between distances derived from manually‐placed landmarks and propagated landmarks of the DXA images “accepted” by the post‐hoc modification procedures.…”

“…Highly mobile regions (e.g., joints of the upper limb) are more difficult to uniformly define in 2D space, and this difficulty is compounded when standard anatomical positioning is not enforced. Despite prior efforts to standardize landmark placement at the proximal humerus, elbow, and wrist, the upper limb could not be incorporated into the current assessment as global body positioning was too variable for any of the image transformations to overcome during image registration (Cataldo‐Ramirez et al, 2022; SI). Additionally, many images did not include the elbow or forearm as these parts of the body were physically located outside of the scanning area.…”

“…Before adding to the pipeline, the registration framework was optimized for use with 2D whole body DXA scans by experimenting with parameter modifications and visualizing the resulting landmark propagations. When visual inspections no longer yielded clear improvements in skeletal landmark propagation, we formally tested the performance of ANTs on automating the extraction of postcranial skeletal measurements on a small subset of ethnically diverse DXA images from the UKB (Cataldo‐Ramirez et al, 2022; Tables S1 and S2). The results of this preliminary investigation showed that using ANTs registration and landmark propagation alone yielded high variance in percent error across all skeletal measurements assessed, and that not all atlases perform equally well for any given moving image (see Supplemental Information: for further details).…”

“…The registration and warping step consists of a rigid transformation to align the moving to the atlas image, an affine transformation to match the moving to the atlas image by scaling and shearing, and a non-linear transformation to match localized regions of the moving to the atlas image. Before adding to the pipeline, the registration framework was optimized for use with 2D whole body DXA scans by experimenting with parameter modifications and visualizing the resulting landmark propagations.When visual inspections no longer yielded clear improvements in skeletal landmark propagation, we formally tested the performance of ANTs on automating the extraction of postcranial skeletal measurements on a small subset of ethnically diverse DXA images from the UKB(Cataldo-Ramirez et al, 2022; Tables…”

“… Table S1. Self‐reported ethnicities of participants included in Cataldo‐Ramirez et al (2022) analyses. Table S2 .…”