UCLA scientists have built an AI-powered “test track” for cancer drugs, using thousands of tiny lab-grown tumor organoids that can be watched in real time as they live, grow and react to treatment. The platform blends 3D bioprinting, label-free imaging and machine-learning analysis to measure how individual organoids respond to different drugs, with an eye toward speeding and sharpening therapy screening for patients with tough-to-treat cancers.
How the platform works
At the heart of the system is an extrusion bioprinting protocol that places tumor cells into high-throughput multiwell formats, essentially laying out hundreds to thousands of mini tumors in a neat grid. The team then uses high-speed, label-free quantitative phase imaging to continuously track organoid biomass and growth dynamics without adding any dyes that might interfere with the cells.
From there, automated image reconstruction, deep-learning segmentation and machine-learning tracking take over, turning raw images into quantitative data. The workflow lets researchers measure drug responses at single-organoid resolution across thousands of samples at once. As detailed in Nature Protocols, the protocol is laid out to be reproducible so other labs can plug it into their own systems.
Researchers say it reveals rare resistant cells
Being able to follow each organoid individually is key, according to co-senior author Dr. Michael Teitell. “This allows us to measure drug responses across thousands of individual organoids,” he said, adding that single-organoid tracking can uncover rare drug-resistant tumor populations that bulk assays would simply average away. His comments were published in a press release from UCLA Health that accompanied the protocol.
Built on earlier proof-of-concept work
The new protocol is not a one-off trick but a refinement of a 2023 proof-of-concept pipeline that first paired bioprinted organoids with high-speed live-cell interferometry to gauge drug sensitivity in hundreds to thousands of 3D cultures. That earlier study, published in Nature Communications, demonstrated label-free, time-resolved mass measurements and provided much of the analytic backbone that this protocol now standardizes for wider use.
Code and methods are public
To help other groups spin up the same setup, the authors released a step-by-step protocol plus the analysis code and imaging tools on public repositories, so labs interested in HSLCI imaging and the tracking pipeline can get started without reinventing the wheel. According to Nature Protocols, the paper includes links to both a GitHub repository and a Zenodo deposition hosting the code and data.
Where this fits in the clinic
The UCLA team suggests that, in the future, this workflow could be used to test a patient’s own tumor cells before treatment, helping clinicians decide which therapies to try first. For now, though, it is strictly a research tool, not a clinical diagnostic, and it still needs more validation before it can be used to guide care.
Co-senior author Alice Soragni has described the work as a milestone on the road to CLIA-grade implementation through the Colorado Center for Personalized Medicine. At the same time, experts have warned that bringing 3D functional screening to the point of care will not be simple, with operational and validation hurdles still standing in the way, as noted in coverage by The Scientist.
Funding and next steps
The project pulled in support from several corners of the federal research apparatus, including the Air Force Office of Scientific Research, the Department of Defense, the National Science Foundation and the National Institutes of Health, according to the UCLA release.
Next up, the researchers say they plan to validate the platform using more patient samples and to build out the quality systems and cross-site comparisons needed to move the workflow closer to clinical translation.
Because both the protocol and code are public, academic labs do not have to wait for a commercial package; they can start testing and tweaking the workflow immediately, which could speed up comparisons across research centers. For those looking to dig into the technical fine print, the full methods and related resources are collected in the Teitell Lab publications.
