The impact factor is 70+, and the hydrogel is in the top issue: How does 3D printing play a role in biomedicine?

The environment and the global energy crisis have put forward demands for clean energy. In this case, hydrogen becomes the most promising energy carrier, and electrocatalytic hydrogen evolution reaction is an effective hydrogen production method. However, the cost of Pt-based electrocatalysts is too expensive, and it is very necessary to develop low-cost and efficient electrocatalysts.

The impact factor is 70+, and the hydrogel is in the top issue: How does 3D printing play a role in biomedicine?

Original Glenn Fantastic Objects Yesterday

Author: Zhijie Zhu
Corresponding author: Hyun Soo Park, Michael C. McAlpine
Correspondence unit: University of Minnesota

1. Research highlights

(1) An overview of the functional inks for 3D printing, focusing on electronic materials and hydrogels.
(2) The latest development of artificial intelligence-driven 3D printing, including open loop, closed loop and prediction systems;
(3) The surgical robot platform was discussed.

2. Research background

3D printing technology has been developed since the early 1980s, ranging from the manufacture of fused filaments with complex plastic structures to next-generation technologies such as direct ink writing and inkjet printing. The latter method can interweave a series of functional materials other than hard plastics, such as conductors, semiconductors, tissue-like soft materials and biological materials (Figure 1a). With its inherent customizability and rapid prototyping, 3D printing offers the potential to manufacture personalized artificial or bionic organs and smart wearable devices.

Despite the above potential advantages, 3D printed multifunctional wearable devices and implantable devices are still in the early stages of development. The challenge lies in two points: (1) Formulating inks that show functionality after printing and are compatible with other inks for the manufacture of complex multi-functional structures; (2) The traditional printing platform is applied to the The target surface is "blind" because of the nature of ex-situ printing: the prescribed design is usually made on a flat and horizontal substrate, and then transferred to the target through an adhesive. One way to avoid these shortcomings is to use artificial intelligence-driven minimally invasive 3D printing technology to manufacture directly on the target surface.

The artificial intelligence defined here includes not only machine learning, but also a broader category of automatic systems that can convey "understanding" or "thinking" (Figure 1b), such as computer vision systems, which can reliably detect, track, and identify targets Print surface, check printing quality and diagnose printing status. In addition, artificial intelligence-driven printing can learn from past experience to predict possible future states and thus react quickly to fast-moving and deforming targets.

Figure 1 Overview of 3D printing functional materials and devices supported by artificial intelligence-assisted manufacturing technology.

Three, the core content

Recently, the team of Hyun Soo Park and Michael C. McAlpine of the University of Minnesota in the United States published an article on Nature Materials Review with the topic "3D-printed multifunctional materials enabled by artificial-intelligence-assisted fabrication technologies", discussing the use of in-situ 3D printing Electronic and biological inks, artificial intelligence enhanced 3D printing methods with open loop, closed loop and predictive control. In the future, surgical robots and artificial intelligence can also integrate 3D printing methods. The fusion of artificial intelligence, 3D printing, functional materials, and personalized biomedical equipment will bring a compelling future to smart manufacturing.

Four, summary points

1. Key research on artificial intelligence-assisted 3D printing
In addition to the traditional 3D printing system without artificial intelligence, the author focuses on the three levels of artificial intelligence in the printing process under the background of the feedback control system to compensate for the complexity and uncertainty of the target shape and movement: open-loop artificial intelligence, Closed loop artificial intelligence and predictive artificial intelligence. Table 1 summarizes the results of various levels of artificial intelligence. Note that the row of 3D printing with predictive AI is still empty, revealing a broad field of research.
Table 1 Summary of key research on artificial intelligence-assisted 3D printing

2. Functional inks for 3D printing
3D printing materials need to be compatible with corresponding 3D printing methods, which are generally classified as light-based or deposition-based. Light-based 3D printing methods include selective laser sintering, stereolithography, and digital light projection. These methods usually have strict limitations on the photoresponsibility of inks (their photocrosslinking or sintering capabilities) and the format of the printing platform. More multifunctional light-based printing platforms with conformal printing capabilities and replaceable inks have been developed to achieve in-situ printing of multiple materials. Deposition-based printing methods, such as direct ink writing and inkjet printing, are based on the on-demand deposition of droplets or filaments of materials to form 3D structures in free space. They can use multiple extrusion heads for multi-material printing. The extrusion heads are connected to contain various properties (such as conductive, semi-conductive and biological inks) and physical forms (such as filaments, gels, liquids, suspensions, and Viscous solution) ink container. Unlike traditional light-based printing operations that require material barrels, deposition-based printing has inherent compatibility with in-situ manufacturing inside living bodies.

2.1 Electronic materials
Materials used for in-situ 3D printing of electronic devices should meet certain requirements: display the electrical properties required to achieve functions such as sensing, photoelectric, and driving; have mechanical properties that match the target biological surface to obtain mechanical compliance ; Demonstrate appropriate rheological properties to achieve predictable behavior and spatial control of material deposition; maintain a non-irritating interface with the target biological surface during and after manufacturing

2.2 Hydrogel
3D printing materials that can come into contact with soft tissues and organs are essential for patient-specific functional scaffolds, biocompatible adhesives, and some electronic wearable devices and implants. Among them, hydrogel is the 3D printed biological interface Ideal candidate material. Similar to biological tissues, hydrogels are viscoelastic polymer networks permeable to water and can be used in tissue engineering and bioelectronics applications. They meet the main design goals of in-situ 3D printing of biological interfaces: if they are based on biopolymers such as collagen, cellulose and gelatin, or based on biopolymer-silica hybrid materials, they can provide suitable cells for cell survival and proliferation. Support; they are similar to natural extracellular matrix, with tissue-like softness and viscoelasticity, which can be adjusted by changing the porosity of the polymer network and the viscosity of the liquid phase; their interface properties can be adjusted so that they can adhere On the wet tissue, it can adhere to various solid surfaces commonly found in electronic devices; by adding photoinitiators, their crosslinking behavior can be optimized for light-based printing methods; and by adding rheology modifiers To optimize their rheological properties for use in deposition-based printing methods.

Figure 2 Functional ink for in-situ 3D printing.

3. Open loop artificial intelligence for 3D printing
Open-loop 3D printing relies on obtaining geometric information of the target surface before the manufacturing process (offline), and artificial intelligence uses this geometric information to determine tool path design and material distribution. Table 2 summarizes some 3D sensing methods used to capture the target surface geometry to assist 3D printing. The 3D reconstruction of the target surface geometry is usually performed offline using complex algorithms with sensors, such as the sensors of computed tomography, laser scanning and structured light scanners. Then, the geometry can be calibrated relative to the printing platform, and the trajectory of the conformal tool path can be generated for direct printing on the target surface. In the case where the target surface is not used as a support substrate for conformal printing, for example, for damaged nerve pathways that will be inserted into a regenerative implant, artificial intelligence can use the previously scanned library or scan data of incomplete anatomical structures to reconstruct it for Patient-specific models of implants manufactured by 3D printing.

Table 2 Common 3D sensing methods that assist 3D printing

More interestingly, considering the latest developments in deformable materials, the study of model-based 4D printed programmable structures can achieve precise and predictable shape control of implants to obtain, for example, stimulus-responsive implants that can be used for manufacturing. In minimally invasive implantation, the implant is compact in size and expands to fill the target space after being deployed to the target location, or deforms as the body heals.

3.1 Conformal 3D printing on non-planar surfaces
In order to achieve direct printing on static non-planar surfaces, artificial intelligence must obtain the geometric information of the target before the printing process. We define this kind of offline sensing capability artificial intelligence-driven 3D printing as open-loop 3D printing. Conformal printing on a non-planar, pre-existing surface can not only use deposition-based 3D printing, but also light-based printing methods, such as stereolithography.

Figure 3 Conformal 3D printing on a non-planar surface.

3.2 Shape design based on target geometry

Wearable medical implants can be directly 3D printed on the human body to achieve clinical diagnosis or help wound recovery. The shape programmability of active materials or passive materials under guided deformation can further expand the functions of 3D printing equipment. Shape design or shape deformation refers to the design of intrinsic (e.g. thermal expansion and expansion) and extrinsic (e.g. mesoporous structure and non-uniform material distribution) material properties in order to respond to specific stimuli (e.g. temperature, ion concentration and/or mechanical load). Change) to achieve the specified shape change.

In two- or three-dimensional scales, the ability to mix hybrid woven materials makes 3D printing an ideal manufacturing method for shape programming. It can be integrated into in-situ 3D printed medical wearable devices to compensate for body movements (such as joint movement and Tissue deformation) to enhance the robustness and durability of the device.

By acquiring scanned images of the deformable target surface, predicting the uncertainty in the printing environment and the nonlinear characteristics of the functional ink, artificial intelligence facilitates the design and printing of wearable devices that conform to body movements. In 3D printing, the use of artificial intelligence for shape programming is considered an open-loop process because artificial intelligence is used in the design phase, not during the printing process.

Figure 4 Programming based on target geometry.

4. 3D printing closed-loop artificial intelligence

Closed-loop artificial intelligence printing refers to real-time (online) 3D printing that adapts to changes in the operating environment (such as the movement and deformation of the target surface, printing defects, ink flow control, and nozzle functions). Based on detection, tracking and recognition algorithms, the target geometry and movement, the surface topography of the printing layer and the extrusion state of the print head can be updated during the printing process. In this part, the author first discusses the two main functions of closed-loop artificial intelligence: improving print quality through online correction and in-situ printing on moving targets through online tracking. Subsequently, robot perception was discussed as a core perception technology, which can realize closed-loop 3D printing.

4.1 Improve print quality

Due to disturbances in the printing environment (such as temperature changes and vibration), as well as uncertainties in material properties (such as ink viscosity and material morphology) and mechanical behavior (such as movement errors in the drive mechanism), closed-loop control is required to be in situ During 3D printing, precise spatial processing of material distribution is achieved. Various sensors such as cameras and strain gauges have been integrated into the 3D printing platform to observe the status of the feeding system and printing structure. The sensor data is sent to computing tools such as computer vision and machine learning algorithms to identify printing defects and provide feedback to the material handling and motion control systems to correct printing errors. Closed-loop artificial intelligence improves print quality and opens up new opportunities in quality-controlled industrial manufacturing, including the manufacture of high-precision parts.

4.2 3D printing on moving targets

Biological systems are dynamic in nature. Therefore, for in-situ 3D printing, the target biological surface usually cannot be completely fixed. For example, the skin and soft organs in the living body undergo rigid deformation and non-rigid deformation, and these deformations are caused by breathing, heartbeat, and surgery (such as gas blowing). In-situ printing on these dynamically changing geometric figures requires online updates based on sensor data to adjust the tool path.

Figure 5 3D printing with closed-loop correction and 3D printing on moving targets.

4.3 Robot perception for in-situ 3D printing

In order to safely perform 3D printing on the fragile organs and tissues of the human body, a mixture of 3D printers and "robotic surgeons" can be used. This hybrid should have the ability to perceive, be able to recognize the geometric shapes and mechanical properties of organs and tissues, as well as disturbances and uncertainties in dynamic systems. Perceived geometric properties (such as surface deformation and movement) are directly related to the spatial control of the printing process. Compared with medical imaging methods used for real-time 3D reconstruction, such as magnetic resonance imaging and CT, vision-based sensing methods do not require huge data acquisition equipment. Therefore, they can be integrated into minimally invasive surgical robots using small image sensors ( Such as endoscopic camera) to help perform in situ surgery on or inside the human body.

In the article, the author distinguishes between low-level and high-level visual perception systems and discusses them separately. Low-level perception is similar to human visual perception. It uses imaging systems and image processing algorithms to detect visual features in the captured image, such as color and texture, for 3D reconstruction. High-level perception is similar to the perception of the brain and nervous system; in 3D printing, it understands the semantics of the 3D scene reconstructed by low-level perception for subsequent tool path planning.

Figure 6 Robot perception for in-situ 3D printing.

5. 3D printed predictive artificial intelligence

The ability to perform 3D printing on living organs or organisms is fundamentally limited by the inherent delays of sensing, control and calculation. The inability of a 3D printing robot to respond to the deformation of the tissue surface may cause the robot device to collide or penetrate the tissue, thereby affecting the printing quality, damaging the tissue, and causing safety issues. This requires predictive artificial intelligence, which can not only understand the current state, but also predict the future state based on past experience, predict the future deformation of surrounding tissues, and plan for future commands that effectively reduce or eliminate printing errors. Future 3D printing robots can be equipped with predictive artificial intelligence trained by sophisticated machine learning algorithms, which will make printing more adaptable, interactive and accurate.

6. Surgical robot for in-situ 3D printing

In the medical field, in-situ 3D printing can be used to directly transport biological materials with required electrical, chemical and biological functions to the human body through precise spatial control. This can help modern medical treatments in many ways, such as replacing sutures with surgical glue, implanting electrode arrays for neural interfaces, and printing biological scaffolds with engineered cells to repair or replace damaged tissues. Complex hand-held printing tools have been developed for coaxial printing of two biological links into a shell-core structure (Figure 7a) and printing of multilayer thin-film skin grafts (Figure 7b). These printing tools are currently used to print a limited type of ink in a fixed 3D structure, and are currently limited by the ability of different parts of the human anatomy to contain multi-purpose materials.
In addition, in different printing scenarios, the print quality associated with manual operations is inconsistent. On the contrary, the machine control currently used in the 3D printing industry