A computer tool tries to read brain MRI scans for lung cancer gene changes
Original title: DriverNet: A Clinical and MRI-Based Framework for Noninvasive Pre-Treatment Molecular Triage in NSCLC Brain Metastases.
How far along is this research?
- Lab cells
- Animals
- Review
- Tested in people
This was tested in people. That is the most reliable kind of research we share.
The tool was tested on scans and records from real patients, but it has not been used to guide anyone's care yet.
The short version
Researchers built a computer tool that looks at brain MRI scans to guess a lung tumor's gene changes, without taking a tissue sample.
What was studied. Researchers built a computer tool called DriverNet and tested it on 374 patients from three centers whose lung cancer had spread to the brain. The tool used each patient's clinical details and brain MRI scans to guess whether the tumor had an EGFR or ALK gene change.
What they found. The tool was trained on patients from one center, then tested on 54 patients at a second center and 96 patients at a third center. It did better at telling which tumors had these gene changes than tools that used only clinical details or only scans. The researchers also saw a link between the tool's guesses and how long patients lived, but they call that part exploratory.
What this means, and what it doesn't
What it could mean: Gene test results guide which targeted pills a doctor can offer. Those results are not always quick or possible to get. If a tool like this works, it might give doctors a hint while they wait. For now it is a research tool, and it does not replace a real gene test.
What it doesn't mean: This does not mean there is a new treatment. The tool does not treat anything. It only tries to guess a lab result from scans that were already taken. It was tested on old records, not used to make real decisions for patients. The researchers say it must be tested again in new patients before anyone can use it in care. This is not a cure and it is not something you can ask for at your next visit.
Source: PubMed, June 26, 2026 · Read the original
This plain-language summary was written by AI and published automatically after passing our automatic safety checks. How we write.