An open source image processing method to quantitatively assess tissue growth after non-invasive magnetic resonance imaging in human bone marrow stromal cell seeded 3D polymeric scaffolds

Abstract

Monitoring extracellular matrix (ECM) components is one of the key methods used to determine tissue quality in three-dimensional (3D) scaffolds for regenerative medicine and clinical purposes. This is even more important when multipotent human bone marrow stromal cells (hMSCs) are used, as it could offer a method to understand in real time the dynamics of stromal cell differentiation and eventually steer it into the desired lineage. Magnetic Resonance Imaging (MRI) is a promising tool to overcome the challenge of a limited transparency in opaque 3D scaffolds. Technical limitations of MRI involve non-uniform background intensity leading to fluctuating background signals and therewith complicating quantifications on the retrieved images. We present a post-imaging processing sequence that is able to correct for this non-uniform background intensity. To test the processing sequence we investigated the use of MRI for in vitro monitoring of tissue growth in three-dimensional poly(ethylene oxide terephthalate)-poly(butylene terephthalate) (PEOT/PBT) scaffolds. Results showed that MRI, without the need to use contrast agents, is a promising non-invasive tool to quantitatively monitor ECM production and cell distribution during in vitro culture in 3D porous tissue engineered constructs.

Figure 1. MRI scans of a scaffold after 14 days of culture with hMSCs (donor 1) taken with slice thicknesses of 80 µm.

Slices from one scan from the top (A), middle (B) and bottom (C) of the scaffold areas are represented, respectively, as indicated in the marques in (D) the side-view of the scaffold in which cells are stained with methylene blue. (E-F) From the bottom-view of the scaffold, it can be seen that methylene blue stained hMSCs and ECM formed with the appearance of a dense string-like tissue (indicated with arrows) distributed non-homogeneously throughout the scaffold (indicated with asterisks). (G) This was also observed during culture by bright field microscopy. Scale bars represent 2 mm.

Figure 2. Image processing steps on MRI-derived image stacks in the absence (top, images A-D) and presence (bottom, images E-H) of hMSCs (donor 1, 5 weeks of culture).

A and E represent the unprocessed MR images; B and F the images after background subtraction; C and G show the results of segmentation based on a single intensity threshold within a circular ROI, comprising scaffold and surrounding water within the MRI glass tubes; D and H are the obtained 3D mesh models, representing the tissue localization, amounts and density. Scale bars represent 2 mm.

Figure 3. MRI scans presented by a single slice from each of the following volumes; top, middle and bottom of the scaffold after 5 weeks of culture (donor 2) obtained without the use of any contrast agents (A, G and M) and after the addition of Endorem T₂ contrast agent (D, J and P).

Tissue is identified and appears white after image processing (B, H, N, E, K and Q). Scaffold material is extracted from unprocessed images by image processing sequences (C, I, O, F, L and R). By a mathematical combination of the images representing tissue-material and the images displaying scaffold material, a 3D model could be retrieved (S and T).

Figure 4. Processed images can be analysed to quantify tissue formation with high spatial resolution throughout the 3D construct.

(A) Z-stacks were retrieved by summarizing multiple slices into one projection as a 4 stage look-up table. The colors represent the amount of tissue in x-y location over the full thickness of the projection as indicated by the legend. (B) By determination of the amount of tissue over a subset of slices, an indication of tissue presence in different areas of two scaffolds after 5 weeks of culture can be given (n = 15 and 18 slices per presented volume for donor 1 and donor 2, respectively). The tissue amount is presented as the percentage of the available pore volume which is occupied by tissue. The error bars represent the standard deviation in tissue amount over the different slices per scaffold volume (***: p<0.001). (C) Bright field microscopy observation through the full scaffold height in the z-direction showed a homogeneous distribution of cells and tissue between the pores in the x-y plane. By varying the focal plane upon observation an impression of pore filling per scaffold part (bottom, middle, top) can be retrieved. (D) By SEM analysis on a top-to-bottom cross-section of the scaffold, tissue is observed throughout the scaffold. Yet, from the bottom-view was observed that large, more dense, cell sheets were found on the bottom of the scaffold. (F) This finding was confirmed by histological analysis of a top-to-bottom cross-section of a construct. (G, H) From a section of the bottom layer of the scaffold it can be observed that the tissue forms circular patterns in the pores of the scaffolds. Arrows indicate cells and tissue, asterisks indicate scaffold material. Scale bars represent (C, D, E and F) 500 µm, (G) 1 mm and (H) 100 µm.