Boney

This is the Boney toolbox, an extension to SPM12 and its CAT12 toolbox, supporting the extraction of measurements related to the bone and head structure. It is developed by Polona Kalc and Robert Dahnke and is a free but copyright software, distributed under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

If you find any bugs, please report them to [email protected] or [email protected].

Introduction

The increasing interest in the bone-brain crosstalk suggests the implication of bone metabolism in mood, cognition, energy homeostasis, etc. (Khrimian et al. 2017; Nakamura, Imaoka, and Takeda 2021 ; Obri et al. 2018; Rousseaud et al. 2016). Furthermore, low bone mineral density (BMD) and osteoporosis have been associated with an increased risk of Alzheimer's disease (Kostev, Hadji, and Jacob 2018; Zhang et al. 2022; Xiao et al. 2023). However, the BMD measures are typically not available in open-source brain imaging databases, such as IXI, ADNI, AIBL, OASIS, etc. We therefore decided to extract a proxy measure for head BMD from the skull.

Quick method

To extract bone parameters from MRI data, we use a refined SPM tissue segmentation procedure and derive different cranial intensity and thickness measures that can be used to approximate the head BMD measure. An estimation of the head's skin-fold (subcutaneous fat thickness) is also available and can be used as an additional information to the typically available BMI measure.

Quick start

Install Matlab/Octave, SPM12 and its CAT12 toolbox. Download the zip-file and unpack it into your SPM12 toolbox directory.

Run SPM and open the toolbox:

spm fmri; spm_boney;

Open the bone processing batch:

and select the structural images that should be processed (currently supported are only T1-weighted images without fat suppression). Adapt relevant parameters (see paragraph Parameters and SPM batch help) and start the processing. Voilà!

After the processing has finished, you can use the XML2CSV batch to extract the estimated (bone) values for further analyses.

Parameters

Furthermore, you can specify which output files to write, e.g., the short bone-report (as a JPG), or the processed NIFTI volumes or GIFTI surfaces that are also shown in the bone-report.

Results

Boney writes a report file and a mat file that include all processed values for each individual. The most important measures are prompted on the command line while processing. The parameters are grouped by processing aspects listed in table #. The T* parameters are derived from SPM/CAT preprocessing and provide important information about the given MRI input. We estimate thickness (th) or intensity parameters, which can be volume- or surface-based (v/s) in the masked occipital region of the head (H), bone cortex (Bcor), bone marrow (Bmar), or full structure (see table). Intensity-based measures were normalized for CSF intensity but depend strongly on the protocol.

Table: Command-line and report abbreviations:

nr	parameter	content
1	Tw	MRI image weighting (T1w, T2w, PDw, MTw, IRw)
2	Tbg	Backgound intensity (e.g. low in MPRage, high in many MP2Rage, MT, R1 protocols
3	Tfat	Use of fat suppression protocol that reduces the bone (marrow) intensity
4	Tres	RMS resolution quality measure, i.e. weighted-average resolution in mm (lower=better)
5	Tcnr	Contrast-to-noise ratio to roughly quantify image quality and the presence of motion artifacts (lower=better)
6	v/sBmar	Volume/surface-based bone marrow measure
7	v/sBcor	Volume/surface-based bone cortex measure (~bone mineral density)
8	v/sBth	Volume/surface-based bone thickness (diploë and cortex)
9	v/sHth	Volume/surface-based head thickness (fat and muscles)
10	v/sHmed	Volume/surface-based head intensity (median of all voxel-values, protocol dependent!)
11	Tbcor	SPM-based bone cortex intensity - the minimum of the typically 3 Gaussians of the unified segmentation (basic measure only for debugging/tests/comparison)
12	Tbmar	SPM-based bone cortex intensity - the maximum of the typically 3 Gaussians of the unified segmentation (basic measure only for debugging/tests/comparison)
13	Tbdns	SPM-based bone density - volume-ratio between SPM minimum and median Gaussian (expert, only for comparison)
14	Tmed(c)	Volume-based (classic) median bone intensity of the corrected bone (expert, only for comparison)

In addition, a csv-table of the most relevant measures (including brain tissue volumes) is created at the end of the processing. The csv-export batch can be used to create independent table of previously processed data by selecting the XML-reports of interest.

Table: Major XML report fields:

struct	group	function	content
set8t	SPM	boney_segment_get_segmat	The main field "seg8t" includes single values from the SPM preprocessing used by SPM12, CAT12, or CTseg.
tis	SPM	boney_segment_get_segmat	The main "tis" structure inlcudes SPM-based measures (seg8*,WMth) image resolution (res_*) and major image class values (WM,GM,CSF,bone[cortex
tismri	vol	boney_segment_evalSPMseg

Table: Detailed XML report fields:

struct	group	parameter	content
set8t	SPM	Affine	Affine transformation matrix from individual to (MNI) template space.
set8t	SPM	lkp	Gaussians for each TPM class.
set8t	SPM	wp	-
set8t	SPM	mg	Weighting within each TPM class defined by lkp.
set8t	SPM	mn	Gaussian peak value of each TPM class define by lkp.
set8t	SPM	vr	Variance of each Gaussian peak value of each TPM class define by lkp.
set8t	SPM	ll	Final total log-likelyhood of the SPM preprocessing.
tis	SPM	seg8o	SPM seg8 main tissue intensity (mn of max mg in lkp).
tis	SPM	seg8ov	SPM seg8 main tissue variance (vr of max mg in lkp).
tis	SPM	seg8n	SPM main tissue intensity (mn of max mg in lkp) normalized by the WM.
tis	SPM	seg8nv	SPM main tissue variance (vr of max mg in lkp) normalized by the WM.
tis	SPM	seg8con	Minimum brain tissue contrast in SPM seg8t.
tis	SPM	seg8conn	Minimum brain tissue contrast in SPM seg8t normalized by the WM.
tis	SPM	seg8CNR	Noise estimate as minimum of the WM and CSF variance in percent (similar to BWP).
tis	SPM	WMth	SPM WM tissue intensity.
tis	SPM	res_vx_vol	Image voxel resolution in mm.
tis	SPM	res_RES	RMS voxel resolution.
tis	SPM	weighting(n)	MRI image weighting (T1w, T2w, PDw, MTw, IRw).
tis	SPM	highBG	RMS voxel resolution.
tis	SPM	headFatType	Protocol intensity type of the head based on SPM seg8t values (0-low[<CSV], 1-mid[<WM], 2-[>WM]).
tis	SPM	boneIntType	Protocol intensity type of the bone based on SPM seg8t values (0-low[<CSF], 1-mid[<WM], 2-[>WM]).
tis	SPM	WM	Averaged SPM WM intensity.
tis	SPM	GM	Averaged SPM GM intensity.
tis	SPM	CSF	Averaged SPM CSF intensity.
tis	SPM	bonecortex	Cortical bone intensity, i.e. "min( seg8.mn( seg8t.lkp==4) )".
tis	SPM	bonemarrow	Cortical bone intensity, i.e. "min( seg8.mn( seg8t.lkp==4) )".
tis	SPM	bonestruct	Ratio between cortical and spongy bone.
tis	SPM	bone	Average bone intensity, i.e. "mean( seg8.mn( seg8t.lkp==4 ) )".
tis	SPM	head	Average head intensity, i.e. "mean( seg8.mn( seg8t.lkp==5 ) )".
tis	SPM	background	Averaged SPM background intensity.
tismri	vol	TIV	Total intracranial volume (GM + WM + CSF).
tismri	vol	vol	Volume of the SPM tissues classes in mm (probability >.5).
tismri	vol	volr	Relative volume of the SPM tissues classes (probability >.5) normalized by TIV.
tismri	vol	TIV	Total intracranial volume (GM + WM + CSF).
tismri	vol	vol	Volume of the SPM tissues classes in mm (probability >.5).
tismri	vol	volr	Relative volume of the SPM tissues classes (probability >.5) normalized by TIV.
tismri	vol	volfat	Volume of fat tissue in the masked upper head (simple threshold to separate the head tissues, in mm).
tismri	vol	volmus	Volume of muscle-like tissue in the masked upper head (simple threshold to separate the head tissues, in mm).
tismri	vol	volmusr	Relative volume of muscle-like tissue in the masked upper head (simple threshold to separate the head tissues).
tismri	vol	clsQC	Relation of voxel with high vs. low density within 30 mm distance.
tismri	vol	int	Intensity based evaluated tissue classes: (1) GM: median (=tismri.Tth(1)); (2) WM: median (=tismir.Tth(2)); (3) CSF: median (=tismri.Th(3)); (4) bone: kmeans with 3 classes for bone (1) and bone marrow (3); (5) head: kmeans with 3 classes head (1), muscle (2), and fat (3); (6) bg: median (=tismri.Th(6)).

Figure: Bonereport with (1) Table 1 with the tissue intensities and absolute/relative volumes, (2) Table 2 with specific bone and head values (see abbreviations for details, (3) a histogram with the major values, (4) two volumes (i) the orignal image with bone and fat overlay and (ii) the normalized bone intensities with GM, WM and head (HD) segment for orientation, and (5) two surface sets (i) with bone thickness (and atlas boundaries) and (ii) the bone marrow intensity.
Abbreviations: Table 1: GM = grey matter, WM = white matter, CSF = cerebrospinal fluid, BG = background, TPM = tissue probability map, Norm. Med. Int. = normalized median intensity. Table 2: Tw = image weighting, Tres = root mean square resolution rating, Tbg = background type, Tfat = heat fat suppression type, Tbone = bone fat intensity, sBcor = surface-based bone-cortex intensity, sBmar = surface-based bone-marrow intensity, sBth = surface-based bone thickness, sHth = surface-based head thickness.

Validation

*for the updated validation see the preprint

The figure shows the results for the selected bone measures estimated on a UKB subsample created by the evaluation scripts. The most relevant regional bone measures are (i) the occipital surface-based bone cortex estimate sROI_bonecortex3, (ii) the occipital volume-based bone marrow estimate vROI_bonemarrow3, and the bone mineral density estimate (BMD) that show a high correlation to the UKB BMD measures.

In addition, the estimated head fat measure is supported by high correlations with visceral adipose tissue (VAT), abdominal subcutaneous adipose tissue (ASAT), body fat percentage, BMI, and waist measurements of the UKB.

References

You can find out more about the bone and adiposity-related measures in our preprint.

Supplement

Main functions

Concept and overview with main functions and data-flow chart.

Table of Boney functions with important subfunctions and variables vars (see table #) stored in the boney XML output.

order	(sub)function / directories	content
1	boney_segment	Main processing function of the Bone segmentation batch that uses the following subroutines to extract bone and head specific measures.
1.1	boney_segment_preprocessing	Call of SPM or CAT to prepare the MRI images and tissue segments.
1.2	boney_segment_filenames	Setup of filenames depending on the selected input files.
1.3	boney_segment_prepare_print	Prepare command-line output.
1.4	boney_segment_get_segmat	Get SPM and CAT presprocessing structures (seg8t and tis values)
1.5.A1	boney_segment_loadMRI	Load the images and limit the resolution.
1.5.A2	boney_segment_simpleBone	Original prototype processing function with "classic" bone measures.
1.5.A3	boney_segment_evalSPMseg	Estimate refined tissue peaks on the given images (tismri values).
1.5.A4	boney_segment_refineSPM	Refined SPM/CAT bone and head classes (high intensity bone tissue is often misslabed as head).
1.5.A5	boney_segment_extractbone	Estimate the bone/head intensity and thickness maps and parameters (vROI values and boney_*.nii volumes files).
1.5.A6	boney_segment_create_bone_surface	Create median bone surface and map meassures (sROI values and boney_*.gii surface files).
1.5.B	boney_segment_fst	Fast processing without refinements, just using SPM/CAT parameters.
1.6	.	Prepare and write XML output (boney_.xml*).
1.7	boney_segment_print_figure	A lot of code to print a result figure with images and surfaces (boney_report_*.jpg)
1.8	boney_segment_cmdline	Print results on the command-line.
1.9	boney_segment_cleanup	Remove temporary data.
2	boney_xml2csv	cat_io_xml2csv
3	tbx_cfg_boney, spm_boney	SPM toolbox and batch configuration files.
D1	docs [dir]	Publication connected with Boney.
D2	evaluation [dir]	Directory with functions for evaluation (in progress).
D2.1	boney_run_boneseg	.
D2.2	boney_eval_main	.
D2.2.1	boney_eval_readCSV	.
D2.2.2	boney_eval_prepareMeasures	.
D2.2.3	boney_eval_printCrossCorrelation	.
D2.2.4	boney_eval_printDetails	.
D2.3	boney_getBoneMeasures	Old function.
D3	images [dir]	Images and result figures used in this document.