New Neural Disease Treatment Rewires the Brain

Imagine trying to understand a bustling city by looking at satellite images of its buildings. While you may be able to see the shapes, you have no idea which structure is a hospital, a school, or a power plant. That, in essence, has been the challenge for doctors treating complex brain disorders like Parkinson's disease, depression, or Alzheimer's. Conventional MRI scans show the brain's structure — tumors, atrophy, or bleeding — but not the function of its 140 billion neurons.

Now, a team of Chinese scientists has unlocked this puzzle. After years of relentless effort, beset by major setbacks, the team led by Professor Liu Hesheng from the Changping Laboratory and Peking University has developed the world's first non-invasive precision brain stimulation system.

Recently approved for clinical use, this technology could fundamentally alter the way neurological diseases are treated.

Identifying the target

"Radiology cannot reliably diagnose Parkinson's disease," said Liu. "Existing brain maps are too generic for individual treatment."

The team's first breakthrough was identifying a reliable treatment target — the brain's "functional circuitry" responsible for the disease. This required analyzing data from hundreds of Parkinson's disease patients. To do this, researchers worked for months making frame-by-frame comparisons of stimulation signals with clinical outcomes.

Since the existing algorithms performed poorly, Liu led the team to develop a new one. With each algorithm iteration, faint signal features grew increasingly distinct. Ultimately, the common signals became bright enough to form functionally significant areas.

"We spent days and nights in the lab, matching signal after signal," recalled Liu. " Then, a connectivity finally emerged — a brain network called the somato-cognitive action network (SCAN) plays a critical role in the disorder." This discovery pinpointed the "neural target" for the treatment of Parkinson's disease.

Mapping you, not just anyone

Finding the "neural target" was only half the battle. The next challenge was the slow process of creating a personalized brain map for each patient. Using existing software, it took over two days to process a single patient's data — impractical for clinical use.

Moreover, brain injuries or lesions often created "dead zones" in the data, causing the analysis to crash. This led the team to build their own AI solution. They developed a generative model to fill in missing signals, similar to how large language models predict missing words.

"We tested it on ourselves," said Ren Jianxun, researcher at Changping Laboratory. "We erased parts of our own real brain signals and had the AI generate replacements. When the AI-generated signals matched the real ones exactly, we knew we were on the right track."

With the rapid development of AI technology, the team were able to accelerate the model reconstruction. They developed a fast cortical reconstruction algorithm in 2022, followed by a fast cortical registration algorithm in 2024.

By 2025, their efforts had reduced data processing time from over 48 hours to less than 30 minutes, making rapid, individualized brain mapping a reality.

Breaking the implant barrier

The final step was to provide treatment to patients. The most effective existing treatment, deep brain stimulation (DBS), requires invasive surgery to implant electrodes. It's costly and has a less than one percent adoption rate in China.

The team therefore opted for a non-invasive approach, using a specific frequency of electromagnetic fields to "re-energize" damaged neural circuits. However, they faced a massive engineering hurdle: the stimulator must track a patient's head movements with pinpoint accuracy. The imported infrared navigation system was too slow and inefficient.

This meant developing imaging technology such as ultrasound and visible light systems using a domestic camera. But initially, the image processing lagged, causing the system to freeze.

The solution came from noticing repetitive patterns in the image data. "We rewrote the algorithm to batch-process those patterns," explained Ren. The lag vanished. After solving a cascade of other issues — from system calibration to motion tracking — the world's first non-invasive precision brain stimulation system was born.

The device is now being used in dozens of Chinese hospitals, primarily for Parkinson's disease. The total cost of treatment is less than one-tenth that of invasive DBS surgery.

"This breakthrough, achieved by a large team at a national strategic platform, responds to the call for speed in biopharmaceutical innovation," said Xie Xiaoliang, an academician of the Chinese Academy of Sciences and director of the Changping Laboratory. "Most importantly, it brings new hope to patients worldwide."

Source: Science and Technology Daily