Automated segmentation of micro-CT images of bone formation in calcium phosphate scaffolds

Theoretical error of sectional method for estimation of shape memory polyurethane foam mass loss

INTRODUCTION

Measuring in vivo degradation for polymeric scaffolds is critical for analysis of biocompatibility. Traditionally, histology has been used to estimate mass loss in scaffolds, allowing for simultaneous evaluation of mass loss and the biologic response to the implant. Oxidatively degradable shape memory polyurethane (SMP) foams have been implemented in two vascular occlusion devices: peripheral embolization device (PED) and neurovascular embolization device (NED). This work explores the errors introduced when using histological sections to evaluate mass loss.

METHODS

Models of the SMP foams were created to mimic the device geometry and the tetrakaidekahedral structure of the foam pore. These models were degraded in Blender for a wide range of possible degradation amounts and the mass loss was estimated using m sections.

RESULTS

As the number of sections (m) used to estimate mass loss for a volume increased the sampling error decreased and beyond m = 5, the decrease in error was insignificant. NED population and sampling errors were higher than for PED scenarios. When m ≥ 5, the averaged sampling error was below 1.5% for NED and 1% for PED scenarios.

DISCUSSION/CONCLUSION

This study establishes a baseline sampling error for estimating randomly degraded porous scaffolds using a sectional method. Device geometry and the stage of mass loss influence the sampling error. Future studies will use non-random degradation to further investigate in vivo mass loss scenarios.

Semi-automatic micro-CT segmentation of the midfoot using calibrated thresholds

PURPOSE

In the field of skeletal research, accurate and reliable segmentation methods are necessary for quantitative micro-CT analysis to assess bone quality. We propose a method of semi-automatic image segmentation of the midfoot, using the cuneiform bones as a model, based on thresholds set by phantom calibration that allows reproducible results in low cortical thickness bones.

METHODS

Manual and semi-automatic segmentation methods were compared in micro-CT scans of the medial and intermediate cuneiforms of 24 cadaveric specimens. The manual method used intensity thresholds, hole filling, and manual cleanup. The semi-automatic method utilized calibrated bone and soft tissue thresholds Boolean subtraction to cleanly identify edges before hole filling. Intra- and inter-rater reliability was tested for the semi-automatic method in all specimens. Mask volume and average bone mineral density (BMD) were measured for all masks, and the three-dimensional models were compared to the initial semi-automatic segmentation using an unsigned distance part comparison analysis. Segmentation methods were compared with paired t-tests with significance level 0.05, and reliability was analyzed by calculating intra-class correlation coefficients.

RESULTS

There were statistically significant differences in mask volume and BMD between the manual and semi-automatic segmentation methods in both bones. The intra- and inter-reliability was excellent for mask volume and bone density in both bones. Part comparisons showed a higher maximum distance between surfaces for the manual segmentation than the repeat semi-automatic segmentations.

CONCLUSION

We developed a semi-automatic micro-CT segmentation method based on calibrated thresholds. This method was designed specifically for use in bones with high rates of curvature and low cortical bone density, such as the cuneiforms, where traditional threshold-based segmentation is more challenging. Our method shows improvement over manual segmentation and was highly reliable, making it appropriate for use in quantitative micro-CT analysis.

Insight into 3D micro-CT data: exploring segmentation algorithms through performance metrics

Three-dimensional (3D) micro-tomography (µ-CT) has proven to be an important imaging modality in industry and scientific domains. Understanding the properties of material structure and behavior has produced many scientific advances. An important component of the 3D µ-CT pipeline is image partitioning (or image segmentation), a step that is used to separate various phases or components in an image. Image partitioning schemes require specific rules for different scientific fields, but a common strategy consists of devising metrics to quantify performance and accuracy. The present article proposes a set of protocols to systematically analyze and compare the results of unsupervised classification methods used for segmentation of synchrotron-based data. The proposed dataflow for Materials Segmentation and Metrics (MSM) provides 3D micro-tomography image segmentation algorithms, such as statistical region merging (SRM), k-means algorithm and parallel Markov random field (PMRF), while offering different metrics to evaluate segmentation quality, confidence and conformity with standards. Both experimental and synthetic data are assessed, illustrating quantitative results through the MSM dashboard, which can return sample information such as media porosity and permeability. The main contributions of this work are: (i) to deliver tools to improve material design and quality control; (ii) to provide datasets for benchmarking and reproducibility; (iii) to yield good practices in the absence of standards or ground-truth for ceramic composite analysis.

Multimodal analysis of in vivo resorbable CaP bone substitutes by combining histology, SEM, and microcomputed tomography data

This study introduced and demonstrated a new method to investigate the repair process of bone defects using micro- and macroporous beta-tricalcium phosphate (β-TCP) substitutes. Specifically, the new method combined and aligned histology, SEM, and preimplantation microcomputed tomography (mCT) data to accurately characterize tissue phases found in biopsies, and thus better understand the bone repair process. The results included (a) the exact fraction of ceramic remnants (CR); (b) the fraction of ceramic resorbed and substituted by bone (CSB); and (c) the fraction of ceramic resorbed and not substituted by bone (CNSB). The new method allowed in particular the detection and quantification of mineralized tissues within the 1-10 µm micropores of the ceramic ("micro-bone"). The utility of the new method was demonstrated by applying it on biopsies of two β-tricalcium phosphate bone substitute groups with two differing macropore sizes implanted in an ovine model for 6 weeks. The total bone deposition and ceramic resorption of the two substitute groups, having macropore sizes of 510 and 1220 μm, were 25.1 ± 8.1% and 67.5 ± 3.2%, and 24.4 ± 4.1% and 61.4 ± 6.5% for the group having the larger pore size. © 2017 Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater, 106B: 1567-1577, 2018.

A novel method for segmenting and aligning the pre- and post-implantation scaffolds of resorbable calcium-phosphate bone substitutes

Micro-computed tomography (microCT) is commonly used to characterize the three-dimensional structure of bone graft scaffolds before and after implantation in order to assess changes occurring during implantation. The accurate processing of the microCT datasets of explanted β-tricalcium phosphate (β-TCP) scaffolds poses significant challenges because of (a) the overlap in the grey values distribution of ceramic remnants, bone, and soft tissue, and of (b) the resorption of the bone substitute during the implantation. To address those challenges, this article introduces and rigorously validates a new processing technique to accurately distinguish these three phases found in the explanted β-TCP scaffolds. Specifically, the microCT datasets obtained before and after implantation of β-TCP scaffolds were aligned in 3D, and the characteristic grey value distributions of the three phases were extracted, thus allowing for (i) the accurate differentiation between these three phases (ceramic remnants, bone, soft tissue), and additionally for (ii) the localization of the defect site in the post-implantation microCT dataset. Using the similarity matrix, a 94±1% agreement was found between algorithmic results and the visual assessment of 556,800 pixels. Moreover, the comparison of the segmentation results of the same microCT and histology section further confirmed the validity of the present segmentation algorithm. This new technique could lead to a more common use of microCT in analyzing the complex 3D processes and to a better understanding of the biological processes occurring after the implantation of ceramic bone graft substitutes.

STATEMENT OF SIGNIFICANCE

Calcium-phosphate scaffolds are being increasingly used to repair critical bone defects. Methods for the accurate characterization of the repair process are still lacking. The present study introduced and validated a novel image-processing technique, using micro-computed tomography (mCT) datasets, to investigate material phases present in biopsies. Specifically, the new method combined mCT datasets from the scaffold before and after implantation to access the characteristic data of the ceramic for more accurate analysis of bone biopsies, and as such to better understand the interactions of the scaffold design and the bone repair process.

Micropore-induced capillarity enhances bone distribution in vivo in biphasic calcium phosphate scaffolds

The increasing demand for bone repair solutions calls for the development of efficacious bone scaffolds. Biphasic calcium phosphate (BCP) scaffolds with both macropores and micropores (MP) have improved healing compared to those with macropores and no micropores (NMP), but the role of micropores is unclear. Here, we evaluate capillarity induced by micropores as a mechanism that can affect bone growth in vivo. Three groups of cylindrical scaffolds were implanted in pig mandibles for three weeks: MP were implanted either dry (MP-Dry), or after submersion in phosphate buffered saline, which fills pores with fluid and therefore suppresses micropore-induced capillarity (MP-Wet); NMP were implanted dry. The amount and distribution of bone in the scaffolds were quantified using micro-computed tomography. MP-Dry had a more homogeneous bone distribution than MP-Wet, although the average bone volume fraction, BVF‾, was not significantly different for these two groups (0.45±0.03 and 0.37±0.03, respectively). There was no significant difference in the radial bone distribution of NMP and MP-Wet, but the BVF‾, of NMP was significantly lower among the three groups (0.25±0.02). These results suggest that micropore-induced capillarity enhances bone regeneration by improving the homogeneity of bone distribution in BCP scaffolds. The explicit design and use of capillarity in bone scaffolds may lead to more effective treatments of large and complex bone defects.

STATEMENT OF SIGNIFICANCE

The increasing demand for bone repair calls for more efficacious bone scaffolds and calcium phosphate-based materials are considered suitable for this application. Macropores (>100μm) are necessary for bone ingrowth and vascularization. However, studies have shown that microporosity (<20μm) also enhances growth, but there is no consensus on the controlling mechanisms. In previous in vitro work, we suggested that micropore-induced capillarity had the potential to enhance bone growth in vivo. This work illustrates the positive effects of capillarity on bone regeneration in vivo; it demonstrates that micropore-induced capillarity significantly enhances the bone distribution in the scaffold. The results will impact the design of scaffolds to better exploit capillarity and improve treatments for large and load-bearing bone defects.

Geometric-attributes-based segmentation of cortical bone slides using optimized neural networks

In cortical bone, solid (lamellar and interstitial) matrix occupies space left over by porous microfeatures such as Haversian canals, lacunae, and canaliculi-containing clusters. In this work, pulse-coupled neural networks (PCNN) were used to automatically distinguish the microfeatures present in histology slides of cortical bone. The networks' parameters were optimized using particle swarm optimization (PSO). When forming the fitness functions for the PSO, we considered the microfeatures' geometric attributes-namely, their size (based on measures of elliptical perimeter or area), shape (based on measures of compactness or the ratio of minor axis length to major axis length), and a two-way combination of these two geometric attributes. This hybrid PCNN-PSO method was further enhanced for pulse evaluation by combination with yet another method, adaptive threshold (AT), where the PCNN algorithm is repeated until the best threshold is found corresponding to the maximum variance between two segmented regions. Together, this framework of using PCNN-PSO-AT constitutes, we believe, a novel framework in biomedical imaging. Using this framework and extracting microfeatures from only one training image, we successfully extracted microfeatures from other test images. The high fidelity of all resultant segments was established using quantitative metrics such as precision, specificity, and Dice indices.

Development of three-dimensional tissue engineered bone-oral mucosal composite models

Tissue engineering of bone and oral mucosa have been extensively studied independently. The aim of this study was to develop and investigate a novel combination of bone and oral mucosa in a single 3D in vitro composite tissue mimicking the natural structure of alveolar bone with an overlying oral mucosa. Rat osteosarcoma (ROS) cells were seeded into a hydroxyapatite/tri-calcium phosphate scaffold and bone constructs were cultured in a spinner bioreactor for 3 months. An engineered oral mucosa was fabricated by air/liquid interface culture of immortalized OKF6/TERET-2 oral keratinocytes on collagen gel-embedded fibroblasts. EOM was incorporated into the engineered bone using a tissue adhesive and further cultured prior to qualitative and quantitative assessments. Presto Blue assay revealed that ROS cells remained vital throughout the experiment. The histological and scanning electron microscope examinations showed that the cells proliferated and densely populated the scaffold construct. Micro computed tomography (micro-CT) scanning revealed an increase in closed porosity and a decrease in open and total porosity at the end of the culture period. Histological examination of bone-oral mucosa model showed a relatively differentiated parakeratinized epithelium, evenly distributed fibroblasts in the connective tissue layer and widely spread ROS cells within the bone scaffold. The feasibility of fabricating a novel bone-oral mucosa model using cell lines is demonstrated. Generating human 'normal' cell-based models with further characterization is required to optimize the model for in vitro and in vivo applications.

Automated Segmentation and Object Classification of CT Images: Application to In Vivo Molecular Imaging of Avian Embryos

Background. Although chick embryogenesis has been studied extensively, there has been growing interest in the investigation of skeletogenesis. In addition to improved poultry health and minimized economic loss, a greater understanding of skeletal abnormalities can also have implications for human medicine. True in vivo studies require noninvasive imaging techniques such as high-resolution microCT. However, the manual analysis of acquired images is both time consuming and subjective. Methods. We have developed a system for automated image segmentation that entails object-based image analysis followed by the classification of the extracted image objects. For image segmentation, a rule set was developed using Definiens image analysis software. The classification engine was implemented using the WEKA machine learning tool. Results. Our system reduces analysis time and observer bias while maintaining high accuracy. Applying the system to the quantification of long bone growth has allowed us to present the first true in ovo data for bone length growth recorded in the same chick embryos. Conclusions. The procedures developed represent an innovative approach for the automated segmentation, classification, quantification, and visualization of microCT images. MicroCT offers the possibility of performing longitudinal studies and thereby provides unique insights into the morpho- and embryogenesis of live chick embryos.

Segmentation of histology slides of cortical bone using pulse coupled neural networks optimized by particle-swarm optimization

The aim of this study is to automatically discern the micro-features in histology slides of cortical bone using pulse coupled neural networks (PCNN). To the best knowledge of the authors, utilizing PCNN in such an application has not been reported in the literature and, as such, constitutes a novel application. The network parameters are optimized using particle swarm optimization (PSO) where the PSO fitness function was introduced as the entropy and energy of the bone micro-constituents extracted from a training image. Another novel contribution is combining the above with the method of adaptive threshold (T) where the PCNN algorithm is repeated until the best threshold T is found corresponding to the maximum variance between two segmented regions. To illustrate the quality of resulting segmentation according to this methodology, a comparison of the entropy/energy obtained of each pulse is reported. Suitable quality metrics (precision rate, sensitivity, specificity, accuracy, and dice) were used to benchmark the resulting segments against those found by a more traditional method namely K-means. The quality of the segments revealed by this methodology was found to be of much superior quality. Another testament to the quality of this methodology was that the images resulting from testing pulses were found to be of similarly good quality to those of the training images.