Page 264 - IJB-10-4

International Journal of Bioprinting Cell viability in printing structured inks

Figure 13. Suggested workflow for material design in structured ink-based 3D printing with emphasis on consideration for cell viability.

inks based on the principle of subjecting cells to consistent vector, encompassing variables such as velocity at the
fluid forces for pre-experimental assessments of cell inlets, cartridge specifications, and nozzle specifications.
viability in advance. If deemed acceptable, the ensuing Employing two nonlinear kernel functions to translate
steps involve the preparation and 3D printing experiments the input space into a feature space, the resulting first
of structured inks. Should any perceived unreasonableness output vector incorporated critical parameters, including
arise, adjustments to geometric material parameters will be maximum and average wall shear stress, maximum and
necessary. If challenges persist, exploration into modifying average shear stress at material phase interfaces, and
material properties will be undertaken. This workflow maximum and average pressure within each material
expedites structured ink design and, to some extent, phase. The related mapping function is as follows:
mitigates cell death resulting from design irrationalities.
Kz z(, )+
K xx,)+∑
)
fx z(, ) =∑ n i=1 (α i −α i ∗ )( i m j=1 (β j − β ∗ j (VI) b
j
Interest in bioprinting extends beyond engineering and
n
fx z(, ) =∑
(
K xx,)+∑
materials scientists to include clinicians. Nevertheless, the −α i α i ∗ )( i m j=1 (β j − β j ∗ ) Kz z(, )+ b
i=1
j
limited engineering expertise among clinical practitioners where K(x , x) and K(z , z) represent the respective
poses a challenge for their involvement this field, Mercer kernel function between x and x, and between z
50
j
i
including but not limited to structured ink-based printing. and z, respectively; α represents the Lagrange multipliers j
i
A research group introduced a mapping database theory associated with x , and β represents the Lagrange
51
i
using support vector machines, transforming the input multipliers associated with z ; b represents a constant
j
i
space of computed tomography (CT) images into a high- ∗ ∗ j
dimensional feature space incorporating relevant structural term; and α and β represent the respective optimal
j
i
parameters. This approach addressed challenges arising solutions of maximum objective function of the formula
from constrained engineering expertise. Consequently, in as follows:
a quest to understand the fluid forces acting on cells and n 1 n n
determine the corresponding equivalent homogeneous L()α =∑ i=1 α −∑ i=1 ∑ j=1 αα j y yK xx(, ) (VII)
i
i
i
j
i
j
2
inks for structured inks under specific conditions, we
implemented a methodology integrating fluid forces and 1
equivalent analysis for structured inks through support L()β =∑ m i=1 β −∑ m i=1 ∑ m j=1 ββ j y yK zz(, ) (VIII)
j
i
i
i
j
i
vector machines, as depicted in Figure S20 (Supplementary 2
File). The primary objective was to establish robust
mapping relationships leveraging machine learning and where y represents the class label of the training
i
objective functions through data training. sample x . i
The initial input space comprised the input vector Following this, the elements of the general input vector,
of structured inks, spanning pattern types, geometric the first output vector, and the cell positions collectively
parameters, viscosity, and density. Simultaneously, constituted the components of the next input vector.
the secondary input space involved the general input Additionally, through the application of another three
Volume 10 Issue 4 (2024) 256 doi: 10.36922/ijb.2362