# BRI-006: High-performance neuroprosthetic control by an individual with tetraplegia

> 本页由本地论文卡片生成。请不要直接编辑本页；修改源卡片后运行 `vp run docs:generate`。

> 中文支持说明：本页是中文站点镜像，保留论文标题、证据状态和来源字段的原始表述，避免未经核实的翻译改变证据边界。

> 内部 wiki 边界：本页提供论文卡片证据和用户提供的 PDF 下载入口，仅用于私有阅读；全文抽取文本不发布。

> 来源: `library/paper_cards/BRI-006.md`

## 论文访问

* 内部 PDF: <a href={"/papers/BRI-006.pdf"} download style={{ display: "inline-flex", alignItems: "center", justifyContent: "center", minHeight: "2.25rem", padding: "0.45rem 0.8rem", borderRadius: "6px", backgroundColor: "#047857", color: "#ffffff", fontWeight: 700, lineHeight: 1, textDecoration: "none", boxShadow: "0 1px 2px rgba(15, 23, 42, 0.22)" }}>下载论文 PDF</a>
* DOI / 官方页面: [10.1016/S0140-6736(12)61816-9](https://doi.org/10.1016/S0140-6736\(12\)61816-9)
* 部署边界: 这些 PDF 链接只适合私有/受保护的内部 wiki。

## BRI-006: High-performance neuroprosthetic control by an individual with tetraplegia

## Metadata

* ID: BRI-006
* Title: High-performance neuroprosthetic control by an individual with tetraplegia
* Year: 2013
* DOI / URL: 10.1016/S0140-6736(12)61816-9
* Local PDF: 见上方论文访问区块
* Text artifact: local-only path withheld from docs site
* Review status: `extracted`

## Study Type

* Track: Intracortical brain-machine interface / high-dimensional neuroprosthetic arm control
* Task: seven-dimensional control of an anthropomorphic modular prosthetic limb for target-based reaching, orientation, grasping, ARAT-style object tasks, and cone stacking
* Participants or dataset: 1 52-year-old woman with tetraplegia; the participant had motor-complete upper-limb impairment and generally intact sensation
* Hardware: Johns Hopkins University Applied Physics Laboratory Modular Prosthetic Limb, NeuroPort acquisition system, and two Blackrock intracortical microelectrode arrays implanted in left motor cortex
* Channels or sensors: two 4 x 4 mm arrays, each with 96 electrode shanks; single-unit and multi-unit events were converted to firing rates in 30 ms bins and low-pass filtered with a 450 ms exponential smoothing window

## Methods

* Paradigm: progression from 3D endpoint translation control to 4D translation-plus-grasp control and then 7D translation, orientation, and grasp control over 13 weeks
* Signal processing or model: manually classified single-unit and multi-unit events; a linear model related firing rate to 7D movement velocity; indirect optimal linear estimation with ridge regression was used for decoding, excluding units with R2 \<= 0.1
* Training/calibration: daily observation-based calibration used 80 automatic MPL trials followed by 80 participant-controlled trials with orthoimpedance; calibration took about 15 minutes; computer assistance was used during early training and removed after day 66
* Online/offline: online invasive BMI control of a prosthetic limb in a registered clinical study

## Results

* Metrics: seven-dimensional sequence-task success rate, chance level, block completion time, path efficiency, selected Action Research Arm Test scores, cone-stacking completion, drops, and adverse events
* Main findings: the participant moved the MPL freely in 3D on the second recording day and later performed routine 7D control; during the last 2 weeks, mean target-task success was 91.6% (SD 4.4) versus median chance 6.2% (95% CI 2.0-15.3); block completion time decreased from 148 s to 112 s and path efficiency increased from 0.30 to 0.38; selected ARAT total scores were 15-17 of 27 compared with 0 without assistive technology; no adverse events were reported
* Reported limitations: single-participant invasive study; substantial multi-week training and daily calibration; early task learning used computer assistance; no noninvasive EEG, SSVEP, YOLO perception, dynamic object selection, or shared-autonomy comparison was tested; future developments such as telemetry, tactile feedback, and richer hand control were left open

## Relevance To This Project

* Supports: a high-performance invasive upper-bound comparator for robotic reaching, orientation, and grasping from neural activity
* Conflicts with: SAH-BRI-Grasp aims at noninvasive SSVEP-MI high-level intent plus robot autonomy rather than direct intracortical 7D continuous prosthetic control
* Design implication: cite BRI-006 to motivate why low-level dexterous continuous control is possible with invasive motor-cortex interfaces, while noninvasive EEG systems should be scoped more conservatively

## Extracted Evidence

| Claim | Status | Evidence Note | Page/Section |
| --- | --- | --- | --- |
| BRI-006 implanted two 96-channel intracortical arrays in the motor cortex of one participant with tetraplegia. | verified | The methods describe two Blackrock 4 x 4 mm arrays with 96 electrode shanks each, implanted about 14 mm apart in left motor cortex. | Summary; Methods, Array implantation |
| The study targeted seven-dimensional prosthetic-limb control. | verified | The control dimensions were 3D translation, 3D orientation, and 1D grasp; the participant controlled all seven dimensions during testing. | Summary; Methods, Observation-based calibration and neural decoding; Brain-machine-interface testing |
| Training combined observation-based calibration, participant-controlled calibration, and a staged progression in dimensionality. | verified | The paper describes progression from 3D to 4D to 7D control, daily calibration, and early computer assistance that was removed after day 66. | Methods; Brain-machine-interface testing |
| High seven-dimensional task performance was achieved late in training. | verified | During the last 2 weeks without computer assistance, mean success was 91.6% versus median chance 6.2%; completion time and path efficiency also improved. | Results; Figure 3 |
| Functional upper-limb assessment improved with the BMI-controlled prosthetic limb. | verified | The participant scored 15-17 of 27 on selected ARAT items while scoring 0 without BMI or other assistive technology; cone stacking was completed on all 4 testing days. | Results; Tables 1-2 |
| BRI-006 is an invasive high-performance comparator, not direct evidence for low-channel noninvasive SAH-BRI-Grasp performance. | inferred | The study uses intracortical arrays and direct 7D prosthetic control, whereas SAH-BRI-Grasp proposes noninvasive SSVEP-MI intent with vision-guided shared control. | Full paper; README-001 |
| Whether noninvasive SSVEP-MI can approach these manipulation metrics is unresolved. | needs confirmation | The local text does not test EEG, visual evoked potentials, motor-imagery EEG, YOLO scene perception, or shared-autonomous grasp execution. | Full paper |

## Open Questions

* Which BRI-006 metrics should be used as aspirational comparators rather than expected noninvasive EEG targets?
* How should SAH-BRI-Grasp report task success so it is not conflated with direct intracortical 7D prosthetic control?
* Would a shared-autonomy pipeline be compared against 3D translation-only, 4D translation-plus-grasp, or full 7D control baselines?
* The local text does not resolve transfer to EEG, low-channel systems, SSVEP target selection, or vision-generated command spaces.