AI Accelerates Radiotherapy, JWST Discovers Methyl Cation CH3, and Green Computing Spotlight

Overview

The Naked Scientists episode highlights three threads at the intersection of AI, astronomy and sustainable computing: AI-assisted radiotherapy planning that cuts waiting times for cancer patients, a James Webb Space Telescope infrared detection of a methyl cation in a young star system, and Lumi the Finnish green supercomputer which demonstrates how renewable energy and waste heat can reduce computing's carbon footprint. The program also covers an Imperial College EMPA heat resistant drone prototype designed to map dangerous fire environments and aid firefighters.

• AI in radiotherapy speeds up healthy-tissue mapping so radiotherapy can begin sooner, improving patient outcomes.
• James Webb observations reveal CH3+ in a circumstellar disk, informing carbon-based chemistry in planet formation.
• Green computing Lumi uses 100% renewable power and waste heat for local heating, illustrating the carbon footprint of large scale computing and how to mitigate it.
• Heat resistant drone demonstrates shared autonomy and aerogel insulation to operate in high temperature environments for rescue operations.

AI in Radiotherapy: Accelerating Cancer Treatment

The podcast begins with a detailed look at an artificial intelligence technology being developed at Cambridge University that speeds up the planning of radiotherapy. In deep-seated cancers the X-ray beams must pass through healthy tissue, which constrains how much dose can be delivered and complicates planning. The team trained a machine learning model on data from around 150 patients, with experts marking healthy tissues so the AI could learn their geometry and boundaries. The AI then returns scans with healthy tissue already marked, giving oncologists a head start. In practice the oncologist can proceed about two and a half times faster, which translates to shorter waiting times for treatment. In Addenbrookes Hospital the goal is to shorten the time from plan to treatment to 14 days for fast-growing tumors and even faster in pursuit of five days, where possible.

The system remains a safety tool rather than a tumor marker. Tumor segmentation remains the human task, while the AI handles the normal tissue markup. The oncologist reviews and approves AI outputs before clinical use. The real-world impact is significant: shaving even a single day off waiting time increases the probability of tumor control for the fastest-growing cancers by roughly 2% per day. This is not only a clinical improvement but also helps reduce the psychological burden of waiting during a cancer diagnosis and treatment process.

James Webb Space Telescope: CH3 Discovery and Cosmic Chemistry

The program then shifts to space science with David Whitehouse explaining that international teams, using the James Webb Space Telescope, detected a methyl cation CH3+ in a circumstellar disk around a very young red dwarf star system called D203506, about 1350 light-years away. CH3+ is a carbon-bearing molecule considered a foundational building block for more complex hydrocarbons that seed planet formation. Webb’s infrared sensitivity is well suited to spotting CH3+ in regions where a radio telescope would miss the signal. This discovery helps confirm the pathways by which simple carbon compounds can assemble into increasingly complex molecules, potentially setting the stage for the chemical precursors of life to form on nascent planets. The finding was shared with Nature, highlighting how infrared astronomy can illuminate chemical evolution in the early stages of planetary systems.

Whitehouse notes that the CH3 root molecule is critical because carbon forms long chains and compounds that can lead to organic chemistry essential to life. The discovery thus informs theoretical models about how carbon-based chemistry evolves in circumstellar disks and how some of these molecules might survive as disks evolve into planetary systems. The James Webb findings are part of a broader set of JWST results that week, including studies of black holes in young galaxies and hints of the spiderweb-like large-scale structure that buckles the early universe into clusters and filaments.

Cosmic Context: Black Holes and the Cosmic Web

Beyond CH3, the podcast discusses two other JWST findings reported in the same week. First, Webb has found evidence of massive black holes at galactic centers less than a billion years after the Big Bang, with masses ranging into the hundreds of millions to billions of solar masses. This rapid early growth challenges existing models of black hole formation and growth in the young universe. Second, Webb captured a striking image of a string of young galaxies that provides observational support for the existence of the cosmic web that structures galaxy clusters. Taken together, these discoveries showcase Webb’s ability to probe conditions in the early universe with unprecedented detail and resolution, advancing our understanding of both cosmology and galaxy evolution.

Green Algorithms and Lumi: Sustainable Computing

The episode then turns to the carbon footprint of computing, a large and often invisible source of emissions. The Green Algorithms initiative, led by Cambridge researchers and Loic Lanlong, aims to quantify and reduce the environmental impact of computational science. The discussion travels to Lumi, Finland’s supercomputer, described by Kimo Koski as one of the most sustainable in the world because it runs on 100% hydroelectric power and captures heat to warm nearby housing. Lumi’s location in Kajaani and its collaboration across nine European countries are highlighted as a model for equitable, sustainable high-performance computing. The conversation underscores the complexities of balancing renewable energy availability with grid needs and the risk of transferring energy demand from one region to another. The Lumi project demonstrates how data centers can be integrated with regional energy and heating infrastructure, reducing the net carbon footprint of research computing while enabling advanced simulations and data analysis that drive science forward.

Heat Resistant Drone for Firefighting: Shared Autonomy in Hazardous Environments

Another forward-looking project covered in the program is a heat resistant drone developed through a collaboration between Imperial College London and the EMPA Materials Science Institute in Switzerland. The drone uses a specially developed aerogel composite to endure high temperatures, with a design that keeps the motors protected inside a sphere-like body. The drone is intended to operate autonomously or with a human operator, in what the researchers call shared autonomy. In a rescue scenario, the drone could be deployed into a burning building to map internal structures, detect heat sources, and locate survivors, delivering crucial data to firefighters while reducing their exposure to dangerous conditions. The system combines materials science, control theory and autonomous robotics to deliver real-time information from hazardous environments.

The podcast closes with a lightning-phenomenon Q&A featuring Met Office lightning scientist Graham Martin, who explains how lightning strikes are located using broadband radio signals and how optical sensors in space complement ground-based networks. The program ends by inviting questions for a future Q&A episode and previews next week’s focus on ocean protection research under the microscope.