Summary: Researchers developed a compact, low-power brain-machine interface ( BMI ) called MiBMI, designed to enhance communication for individuals with severe motor impairments. The system uses a very miniaturized, real-time processor to convert neural activity into words, producing words with 91 % accuracy.
This development opens the door to real-world, implant BMIs, which have the potential to significantly improve Und and spinal cord injury patients’ quality of life.
Important Information:
- MiBMI processes neuronal impulses in real-time, converting ideas into words with 91 % accuracy.
- The device’s severe nanotechnology makes it suitable for implanted use, with minimal invasiveness.
- MiBMI may go beyond writing to include applications like activity control and speech decoding.
Origin: EPFL
Brain-machine interfaces ( BMIs ) have come to be a useful tool for helping people with severe motor impairments regain control and communication. Typically, these systems have been heavy, power-intensive, and limited in their useful software.  ,
Researchers at EPFL have developed the first high-performance,  , Miniaturized Brain-Machine Interface ( MiBMI ), offering , an extremely small, low-power, highly accurate,  , and versatile solution.
Published in , the latest issue of the , IEEE Journal of Solid-State Circuits , and presented at the International Solid-State Wires Conference,  , the MiBMI not just enhances the efficiency and flexibility of , brain-machine interfaces , but also paves the way for useful,  , fully , implant products.
This technology holds the potential to significantly , improve the quality of life for patients with conditions , such as , amyotrophic lateral sclerosis ( ALS ) and spinal cord injuries.
The , MiBMI ‘s , small size and low power are key features, making , the , system , suitable for implantable applications. Its , little invasiveness , ensures safety , and , practicality , for use in therapeutic and real-life options.
Additionally, because it is a fully integrated system, both 8mm in full and two exceedingly little chips are used for recording and running.  , Thisis the latest in a new class of low-power BMI , devices , developed at Mahsa Shoaran’s Integrated Neurotechnologies Laboratory ( INL ) at EPFL’s IEM and Neuro X institutes.  ,
” MiBM I allows us to turn complex neural activity into clear text with high accuracy and low power consumption. According to Shoaran, this development brings us closer to developing useful, implantable solutions that can drastically improve motor function.  ,  ,
Decoding neurological signals created when a man imagines writing letters or words is a process of brain-to-text change. In this way, electrodes placed in the mind report brain-related motor actions triggered by handwriting. The MiBMI chipset then converts these brain-directed hand movements into appropriate electric word as a result of real-time processing.
This technology allows people, especially those with locked-in illness and other significant engine difficulties, to , communicate , by just thinking about writing,  , with , the interface , converting their thoughts , into readable text on a camera.
” While the , chip , has not yet been integrated into a working BMI, it , has  , processed , data , from previous live , recordings, such as those from the Shenoy lab at Stanford, converting , handwriting activity into text , with , an impressive 91 % accuracy”, says lead , author , Mohammed Ali Shaeri.
The chip can currently decode up to 31 different characters,  , an achievement unmatched by , any , other integrated , systems.  ,
” We are confident that , we , can , decode up to 100 characters,  , but , a handwriting dataset with more characters is not yet available” , , adds , Shaeri.  ,
Current BMIs record the data from electrodes placed in the brain before decoding these signals to a different computer for the decoding. The iBMI ch ps ecord time—integrating the data bu also proc a , sses he nf rm tio in r al time—i tegr ting a&nbs p 192-channel , neural recording system with a , 512-channel , neural decoder.
This neurotechnological breakthrough is a feat of extreme miniaturization that combines expertise in integrated circuits, neural engineering, and artificial intelligence. This innovation is especially exciting in the upcoming era of neurotech startups in the BMI domain, where integration and miniaturization are key priorities. The MiBMI of EPFL provides promising insights and potential for the field’s future.
The researchers had to approach data analysis with a completely different approach to processing the enormous amount of information that the electrodes on the miniaturized BMI picked up.
They discovered that the brain activity for each letter, when the patient imagines writing it by hand, contains very specific , markers,  , which the researchers have named  , distinctive neural codes ( DNCs ).  , Instead of , processing , thousands of bytes of data for each letter, the microchip only needs to process the DNCs, which are around a hundred bytes.
This makes the system , fast, accurate,  , and with low-power consumption.  ,  , This , breakthrough , also , allows for faster training times, making learning how to use the BMI easier and more accessible.  ,
Collaborations with other teams at EPFL’s Neuro-X , and IEM Institutes,  , such as with the laboratories of Grégoire Courtine, Silvestro Micera,  , Stéphanie Lacour,  , and David Atienza , promise to create the next generation of integrated BMI systems. Beyond handwriting recognition, Shaeri, and their team are looking into various uses for the MiBMI system.
” We are collaborating with other research groups to test the system in various situations, such as movement control and speech decoding,” according to the statement. Our goal is to create a versatile BMI that can be tailored to a range of neurological conditions, giving patients a wider range of options,” says Shoaran.
About this research in neurotech and BMI.
Author: Michael Mitchell
Source: EPFL
Contact: Michael Mitchell – EPFL
Image: The image is credited to EPFL / Lundi13 – CC-BY-SA 4.0
Original Research: Closed access.
” A 2.46mm2 Miniaturized Brain-Machine Interface ( MiBMI ) Enabling 31-Class Brain-to-Text Decoding” by Mahsa Shoran et al. IEEE Journal of Solid-State Circuits
Abstract
A 2.46mm2 Miniaturized Brain-Machine Interface ( MiBMI ) Enabling 31-Class Brain-to-Text Decoding
Recent advancements in brain-machine interface ( BMI ) technology provide novel treatments for people with motor impairments, with the potential extending to handwriting assistance and speech synthesis. However, current BMIs rely on cumbersome benchtop setups and resource-intensive computing devices, which makes them difficult to use on a daily basis.
We introduce a miniaturized BMI ( MiBMI ) system capable of accurate, multiclass neural decoding and high-density sensing in a millimeter-scale silicon footprint, making it suitable for next-generation implantable BMIs. A 512-channel, 31-class neural decoder employs a novel concept of distinctive neural code ( DNC ) driven by a class saliency model.
This enables the use of a low-complexity linear discriminant analysis ( LDA ) classifier to precisely translate complex neural activity into handwritten characters.
The proposed decoder achieves significant improvements in memory utilization (  , ∼ , 100 , × , ) and computational complexity (  , ∼ , 320 , × , ) compared to a conventional LDA without DNCs. Moreover, MiBMI enables area-efficient 192-channel neural recording through time-division multiplexing, demonstrating its potential for fully integrated BMIs.
Fabricated in a 65-nm CMOS process, the high-channel-count BMI chipset occupies a compact area of 2.46 mm , 2 , and consumes 883 , μ , W. The proposed decoder translated human intracortical neural activity into 31 characters with 91.3 % accuracy, significantly enhancing the task complexity compared to previous on-chip BMIs.
Furthermore, MiBMI achieved 87 % accuracy in decoding the neural responses of a rat to six classes of acoustic stimuli in an in vivo experiment.
