With MIT, Stevens Develops New AI-Assisted Techniques to Improve Printing of Stem Cells, Biomaterials
The interdisciplinary work has also produced important new insights that could enable programming unique functions into newly grown stem cell cultures
A team of Stevens researchers, in collaboration with MIT’s Center for Bits and Atoms, have developed a new method of fabricating high fidelity, three-dimensional biomaterial substrates leveraging machine learning techniques to enhance 3D printing technology.
The research could pave the way for the additive manufacture of highly homogenous bioscaffolds, upon which more uniform, robust cell cultures possessing highly specific characteristics could eventually be grown.
The Stevens team included primary investigator and mechanical engineer Robert Chang; biomedical engineer Hongjun Wang; chemical engineer and current Vice Provost for Research, Innovation and Entrepreneurship Dilhan Kalyon; and postdoctoral researcher Chao Jia, working with MIT’s Filippos Tourlomousis, Thrasyvoulus Karydis and Andreas Mershin. The investigative work was primarily carried out at Stevens, in association with Tourlomousis’ doctoral thesis research.
New insights into substrate geometry, cell differentiation
The new method uses melt electrowriting to weave patterns and build materials from tiny polymer strands as small as 10 microns wide — smaller than current standards by at least a factor of ten.
Machine learning tools were used to classify and analyze large numbers of images of cell samples being grown on various types of printed microstructures in order to discover relationships between different fiber arrangements and the cells’ growth patterns.
The best scaffolds produced by the team were highly homogenous when compared with non-woven structures, the researchers found — and the smaller scale achieved during fabrication approximates the scale of cells themselves, enabling investigation of the effects of the geometries of the fabricated structures on the morphology, growth and activity of cells at a scale of examination not previously attempted.
The work also produced a remarkable new finding: that the type and rate of differentiation of cells seeded on lattice-like substrates appears to depend upon the geometry of the substrates themselves. The insight suggests that the differentiation rate and characteristics of cells being grown can be controlled by tuning the specific geometry of the substrates upon which those cells proliferate; this could prove important when developing stem cell-based therapies, note the researchers.
Findings on fibroblasts as model cell systems were published in Nature Microsystems and Nanoengineering, while continuing work on mesenchymal stem cells will appear in proprietary patent application materials currently in the submission process.