Hydrogels and MRI: meet three biomedical engineers reshaping health care

Biomedical engineering encompasses a broad range of research with clinical applications, from creating surgical implants to using physics to improve medical imaging. Here we focus on three researchers whose fundamental work is paving the way for such health-care solutions.

XINYU DONG: Cold comfort

Hydrogels are among a biomedical engineer’s most favoured materials, thanks to their water-rich, highly porous and elastic nature being able to closely mimic human tissue. But they face some key limitations, says Xinyu Dong: “They’re often quite fragile and their high water content makes them sensitive to dehydration and freezing when surrounding conditions change.”

Dong, a materials scientist at the National University of Singapore (NUS), creates next-generation, high-performance hydrogels using freeze casting, a technique normally used to make ceramic scaffolds that are used as 3D templates to guide bone repair. Also known as ice templating, freeze casting involves mixing a material with water to form a slurry, placing it in a mould, and then using a temperature gradient to grow ice crystals in a particular direction. This forces the slurry’s particles into the spaces between the crystals so that, after evaporating the ice, “what’s left behind is an organized network of pores” in the desired material, explains Dong.

Nature Index 2026 Biomedical engineering

Dong began experimenting with freeze casting in 2020 during her PhD at NUS, creating porous ceramic scaffolds that mimic the structure of bone: strong, yet able to deform to a degree without breaking¹. She later applied this to aerogels — ultra-lightweight materials composed mostly of air — and hydrogels. “These materials are very different,” says Dong. “But the basic idea is the same: by controlling their structure at different scales, we can change how they behave and give them useful properties.”

In 2024, Dong proposed a twist on freeze casting²: instead of evaporating the ice after freezing, the mould is removed and soaked in a chilled organic solvent. As the ice gradually dissolves into the solvent and the organic liquid takes its place in the channels throughout the newly formed gel, the polymer walls are pulled closer together, increasing the linkages between them. The solvent also incorporates salt ions into the matrix, which makes the material electrically conductive. The result? Tougher, stronger and more stretchable skin-like organohydrogels — ideal for use in wearable electronics and soft robotics.

Dong and her collaborators described one such application this year in Nature Communications³, using an extension of this technique. “We developed a soft hydrogel-based electronic skin that formed close and stable contact with the body,” she says. When integrated into a wristband, the gel could be used to monitor skin temperature, pulse and sweat concentration.

As an undergraduate, Dong studied materials science at Soochow University in China, before switching to mechanical engineering for her PhD. This blend, she says, helps give her important insights into how a material’s properties “can be shaped through manufacturing and structural design” and “used to solve real engineering problems”.

Dong credits her father with inspiring her problem-solving skills from a young age. He often travelled for work and would bring back “strange and curious” toys, which she loved to prise apart and put together again, she says.

Dong now hopes to channel this ingenuity towards creating personalized medical technologies that can repair tissues, treat disease and monitor health. “The real opportunity is to create programmable hydrogel materials whose physical, mechanical and biological properties can be tailored to a particular tissue, patient and clinical need,” she says. — Sandy Ong

JAMES EILLS: Strong signals

The body’s conversion of molecules such as pyruvate, a naturally occurring metabolite, can be visualized with equipment including magnetic resonance imaging (MRI) scanners. This is because each metabolite’s chemical state emits a unique magnetic signal, providing a real-time readout of cellular metabolism. However, these signals are inherently weak because only a tiny fraction of atomic nuclei in such molecules align with the magnetic field emitted by even the most powerful scanners. For James Eills, head of the Hyperpolarization Methods Lab at the Jülich Research Centre in Germany, innovative hyperpolarization techniques are the most promising answer to enhance magnetic signals, one of the biggest challenges in MRI and nuclear magnetic resonance (NMR).

Hyperpolarization forces the nuclear spins of atoms to point in the same direction, increasing the fraction of spins that are aligned with the magnetic field and making undetectable signals visible. “Using hyperpolarized pyruvate allows clinicians to monitor diseases better than conventional imaging,” says Eills.

Currently, in biomedical MRI imaging, before injecting a contrast reagent solution, clinicians hyperpolarize the molecules with a method called dissolution dynamic nuclear polarization. Although powerful, this method requires expensive machinery and needs long preparation times. For this reason, Eills believes that it is important to find better ways to hyperpolarize the molecules used as contrast reagents.

He has worked on this challenge since his undergraduate days at the University of Southampton, UK, in the early 2010s, pursuing it through master’s and PhD research at the University of California, Berkeley, and Southampton, and continuing during a postdoc at the Helmholtz Institute in Mainz, Germany, in 2018. He developed methods to try to improve the transfer of spin polarization from parahydrogen, a form of hydrogen gas in which the nuclear spins are aligned in opposite directions, to clinically relevant metabolites such as pyruvate and fumarate. Parahydrogen-induced polarization (PHIP) increases the magnetic resonance signal by several orders of magnitude while avoiding the complexity and cost of conventional dissolution dynamic methods. This work⁴, carried out through collaborations with researchers in the United States, United Kingdom, Italy and Germany, earned him and his colleagues the 2021 Erwin Schrödinger Prize to recognize outstanding collaborative research.

Nevertheless, there is a drawback in the PHIP method that Eills and his collaborators developed: it applies only to molecules that are amenable to the addition of hydrogen — a process called hydrogenation. “The challenge we face in my lab is to find new ways to induce PHIP,” he says. Such innovative ways to hyperpolarize molecules would widen the range of options available for biomedical imaging.

It is a quest that Eills has been pursuing since his second postdoc in 2021 at the Institute for Bioengineering of Catalonia in Barcelona, Spain, where he tackled another important bottleneck in the field. When using strong radio-frequency pulses to manipulate nuclear spins, the strength of the pulse can partially destroy the delicate quantum state of the nuclei. As a result, the tiny magnetic signals emitted by the manipulated nuclei can be lost. Eills and his collaborators attempted to solve this by introducing an atomic sensor to detect microscopic changes in local magnetic fields using laser light, rather than radio waves⁵.

Eills traces his passion for physical chemistry back to attending a 2011 lecture as a college student by the renowned British physical chemist, Peter Atkins. “What caught my attention was Professor Atkins describing the idea of negative temperature in a thermodynamic system,” remembers Eills. “Years later we create negative nuclear spin temperatures all the time using pulsed magnetic fields. What seemed impossible back then is now a simple trick to help us in our experiments.”

Every advance that Eills has brought to the field has helped to address a different obstacle in metabolic MRI applications. Although PHIP-derived imaging agents still need improvements in purity for clinical use, in a recent preprint⁶ Eills’s team showed their potential on human biopsy samples to measure drug response in real time. — Anna Napolitano