MIA-clustering: a novel method for segmentation of paleontological material

Dunmore, Christopher J. and Wollny, Gert and Skinner, Matthew M. (2018) MIA-clustering: a novel method for segmentation of paleontological material. PeerJ, . ISSN 2167-8359. (doi:https://doi.org/10.7717/peerj.4374) (Full text available)

This is the latest version of this item.

PDF - Publisher pdf

Creative Commons Licence
This work is licensed under a Creative Commons Attribution 4.0 International License.
Download (3MB) Preview
[img]
Preview
Official URL
http://dx.doi.org/10.7717/peerj.4374

Abstract

Paleontological research increasingly uses high-resolution micro-computed tomography (mCT) to study the inner architecture of modern and fossil bone material to answer important questions regarding vertebrate evolution. This nondestructive method allows for the measurement of otherwise inaccessible morphology. Digital measurement is predicated on the accurate segmentation of modern or fossilized bone from other structures imaged in mCT scans, as errors in segmentation can result in inaccurate calculations of structural parameters. Several approaches to image segmentation have been proposed with varying degrees of automation, ranging from completely manual segmentation, to the selection of input parameters required for computational algorithms. Many of these segmentation algorithms provide speed and reproducibility at the cost of flexibility that manual segmentation provides. In particular, the segmentation of modern and fossil bone in the presence of materials such as desiccated soft tissue, soil matrix or precipitated crystalline material can be difficult. Here we present a free opensource segmentation algorithm application capable of segmenting modern and fossil bone, which also reduces subjective user decisions to a minimum. We compare the effectiveness of this algorithm with another leading method by using both to measure the parameters of a known dimension reference object, as well as to segment an example problematic fossil scan. The results demonstrate that the medical image analysis-clustering method produces accurate segmentations and offers more flexibility than those of equivalent precision. Its free availability, flexibility to deal with non-bone inclusions and limited need for user input give it broad applicability in anthropological, anatomical, and paleontological contexts.

Item Type: Article
Uncontrolled keywords: Digital image processing, Micro-CT, Machine-learning, Fossil, Trabecular bone
Divisions: Faculties > Social Sciences > School of Anthropology and Conservation
Faculties > Social Sciences > School of Anthropology and Conservation > Biological Anthropology
Depositing User: Matthew Skinner
Date Deposited: 23 Feb 2018 16:35 UTC
Last Modified: 28 Jan 2019 08:38 UTC
Resource URI: https://kar.kent.ac.uk/id/eprint/66128 (The current URI for this page, for reference purposes)
Dunmore, Christopher J.: https://orcid.org/0000-0002-8634-9777

Available versions of this item

  • Depositors only (login required):

Downloads

Downloads per month over past year