Neurorehabilitation and Sensory Feedback in Amputees: Remarkable Insights into Multisensory Integration

This study explores how sensory feedback—the brain’s ability to integrate touch and vision—changes in amputees, which can help provide crucial insights into neurorehabilitation. Researchers examined how individuals with limb loss process multisensory information compared to healthy participants. They reveal that the brain can flexibly adapt to new sensory inputs.

These findings also provide crucial insights into neuroplasticity and prosthetic development, offering new ways to enhance sensory feedback in artificial limb.

Experimental model in the study

The study involved two men with above-knee amputations and sixteen healthy participants. Researchers wanted to understand how well people combine visual and touch information. To do this, they used a method from a 2002 study by Ernst and Banks.

This method suggests that when we use multiple senses together, our brain gives more importance to the more reliable sense. Which helps us make the best possible decision with the least uncertainty. If the actual results match the predicted “best possible” result, it means the brain is integrating the senses optimally.

To test this in both amputees and healthy participants, the authors adjusted the experiment based on their specific questions regarding sensory feedback differences.

Multisensory platform using VR and electrical stimulation

The study combined cutaneous electrical stimulation (tactile) and VR scenarios (visual stimuli). During the experiment, researchers also collected EEG data.

Participants experienced the VR scenario in first-person view. They saw themselves seated with one virtual leg raised on a stool. A wooden bar behind the foot provided a visual stimulus by vibrating at different frequencies. The tactile stimulus was delivered via electrical stimulation using two surface electrodes.

Additionally, electrical stimulation was applied to different body locations:

For amputees: Stimulation at the stump (L1P) and the intact thigh (L2P).

For healthy subjects: Stimulation at the upper frontal thigh (L1), knee (L2), and foot sole (L3).

Experimental Task: Two-Alternative Forced Choice (2AFC)

Participants performed a 2AFC task. They had to judge which of the two consecutive stimuli had a higher frequency i.e. faster vibration. Stimuli could be:

• Unimodal (only visual / only tactile)
• Bimodal (both visual and tactile together)

Authors introduced a blurred panel in some conditions to control for visual dominance. In the end, we have the following Experimental Conditions:

1. Visual (V) – Only visual stimulus (wooden bar vibration)
2. Tactile (T) – Only tactile stimulus (electrical stimulation)
3. Visuo-Tactile (VT) – Synchronized visual and tactile stimuli
4. Visual Blurred (VB) – Visual stimulus with a blur filter
5. Visual Blurred-Tactile (VTB) – Combined visual blur and tactile stimulus

Optimal integration analysis

In behavioral experiments, the Just-Noticeable Difference (JND) is used to measure how much a stimulus needs to change before a person can notice a difference. It helps determine how much uncertainty exists in each sense.

To predict how well people should combine sensory information, authors used the JND of individual senses (vision and touch) to calculate the expected JND when both are combined.

However, multisensory integration works best when both senses are equally reliable. If one sense (vision) is much stronger than the other (tactile), the brain relies on the stronger. This is why they had to introduce the blur effect for visual stimuli.

For each experimental condition and stimulation site, researchers used a statistical method called maximum likelihood estimation (MLE). It creates a curve that describes how participants judged stimuli vibration speeds. Two key values were extracted from these curves.

• JND;
• Point of Subjective Equality (PSE). the vibration speed that was judged as “faster” exactly 50% of the time (indicating when two stimuli felt the same).

Statistical findings of the multisensory integration

Amputees optimally integrated visuo-tactile stimuli at the level of the amputated leg but not at the healthy leg. Healthy subjects only integrated spatially matched stimuli optimally, meaning visual and tactile inputs had to come from the same location (sole of the foot).

The study suggests that unity assumption is the key factor behind the amputees’ ability to integrate non-matching visuo-tactile cues at the stump. Unity assumption is a key mechanism of multisensory integration. It says that the brain assumes stimuli from different senses belong together IF they seem related in time and space.

So, the study also supports a “software” rather than “hardware” model of plasticity. Meaning, the brain does not just reorganize structurally, but also changes how it processes sensory information.

Mental Workload assessments from EEG data support these findings

For EEG analysis, authors extracted a Mental Workload metric. Frontal and parietal areas are known to be key in cognitive load assessment, so those are electrodes of interest. Power of the EEG signal was extracted in the alpha and theta bands as shown:

The EEG findings confirm that amputees integrate visuo-tactile stimuli differently than healthy individuals.

Specifically, mental workload shows amputees required less cognitive effort to process bimodal stimuli when the tactile input was applied to the amputated limb. This suggests that the nervous system adapts after limb loss by flexibly integrating sensory inputs. Even when they are not spatially matched!

For healthy participants, EEG data aligned with the expectation that integration is strongest when stimuli are spatially aligned.

Reaction time data supported these findings too, showing faster responses in conditions where integration was optimal.

Study findings have direct applications in Neurorehabilitation and prosthetic sensory feedback

Because, they show that after limb loss, the brain can adapt to integrate sensory inputs at the amputated limb. Even when they are not spatially aligned. This suggests that neurostimulation-based therapies —such as electrical stimulation at the stump— can be effectively used. For example, to  restore multisensory processing and improve body awareness in amputees.

Previous studies have shown that synchronous visuo-tactile stimulation at the stump can reduce phantom limb distortions and that electrical feedback from prosthetics can improve motor control. By demonstrating that the brain integrates artificial sensory inputs, this study provides validation for using sensory feedback in prosthetic design. With the goal to enhance prosthetic embodiment, reduce phantom limb discomfort, and improve functional outcomes in amputees.

Read the original study here: https://doi.org/10.1016/j.isci.2024.111685