Lecture 22- Neural prostheses

Question 1

What does a spinal cord injury (SCI) produce?

Answer 1

-Produces both paralysis and sensory loss below the level of the lesion Paralysis is of voluntary muscle • Autonomic functions also disturbed: • Bladder and bowel voiding • Sexual function • Cardiovascular function • Sweating

Question 2

What are the features/damage in the SCI?

Answer 2

-most of the nervous system is ok -it is a connectivity problem 1. Cortex, basal ganglia and cerebellum still completely intact 2. All ventral horn motor neurons intact 3. Connections to skeletal muscle intact 4. “Only” cuts connection between brain and ventral horn motor neurons

Question 3

How is movement controlled?

Answer 3

• Postural control from cortex via medial corticospinal tract and reticular formation
• Fine motor control via cortex and lateral corticospinal tract
• Final motor neurons in ventral horn of SC

Question 4

What are the prosthetic devices about?

Answer 4

• Can we use a computer to replace missing link between brain and spinal cord? • Examples of prosthetic neural devices (Neural Interface Systems - NIS) already in widespread use

Question 5

What are two examples of neural prostheses already in use (non-walk related)?

Answer 5

1: Bionic ear, 50,000 in US • For middle ear hearing problems 2: Deep brain stimulators, 30,000 in US • For treatment of Parkinson’s disease (can control the tremor)

Question 6

Do SCI patients produce motor patterns?

Answer 6

• SCI patients still produce motor patterns • Can we intercept signals and use them to aid mobility?

Question 7

How are motor patterns generated? And what is the best place to eavesdrop?

Answer 7

• Motor patterns result of input from large area of frontal lobe
• Distinct regions involved
• Most converge on M1 – last stage before SC and motor neurons (the best place to eavesdrop on)

Question 8

What is the topography of M1?

Answer 8

• Motor pattern spatially encoded
• we can see which part controls which part of the body

Question 9

What is the motor pattern like in M1?

Answer 9

• Motor pattern also electrically encoded • Each movement results from pattern of action potentials in specific subpopulations of neurons distributed through M1 • Movement of an arm involves postural stabilisation along with graded activation of flexor and extensor muscles in a specific sequence -can you just sum up the APs?

Question 10

How can the interface with the cortex work? What are the technical issues?

Answer 10

• Can we listen in to neural activity in the brain and decode the motor patterns? • Need to record signals (thousands?), 1ms duration, 1-300Hz, in mV range, over a significant area • Might substitute local field potentials (sum of all local currents caused by action of multiple neurons)

Question 11

What is Study 1 about?

Answer 11

-Cortical control of a prosthetic arm for self-feeding -training a monkey to use a robotic arm -get reward when using the arm

Question 12

What is the design of study 1?

Answer 12

• Used monkeys (Macaques) • Recorded from M1 and “decoded” motor signal • Use signal to drive a robot arm (4 joints and a set of “fingers”) • Can monkey feed itself using only prosthesis?

Question 13

What was the monkey training like in study 1?

Answer 13

• Monkey trained to feed itself using a joystick to control robot arm • Forms mental image of how the arm can move and get used to feeding with it • NIS implanted in cortex to record motor signals • NIS connected to robot arm • Computer algorithm links brain activity to robot arm movement

Question 14

What is the brain interface that was used in study 1?

Answer 14

• 96 electrode array • Skull mounted multi-pin electrical connector • Samples area 3.5 mm2 • Electrodes penetrate 1 mm below cortical surface

Question 15

Where to put the interface? (study 1)

Answer 15

• Aim to put interface into M1 over hand/forearm representation • Requires brain surgery –put it into the arm region -brain surgery and put it in

Question 16

What is the outcome of study 1?

Answer 16

• Monkey behaves as if it has a third arm • Crude but effective -this is in a monkey not a human!

Question 17

What is study 2 about?

Answer 17

-Neuronal ensemble control of prosthetic devices by a human with tetraplegia • Volunteer (MN) is quadriplegic with high cervical SCI • Device implanted over hand/forearm representation

Question 18

What was the target for the interface in study 2?

Answer 18

• Identified hand and arm region in primary motor cortex on anatomical grounds, -the thing sticking out of his head

Question 19

What is the transform function (study 2)?

Answer 19

-record APs when imagining movement then computer builds an algorithm • Linked brain activity to an output by building a linear filter (transform function) • Asked MN to imagine moving cursor to track screen prompt • “Linear filters were constructed from a response matrix containing the firing rate over a 1-s history for each neuron (twenty 50-ms bins), and regressing this matrix onto technician-cursor position using a pseudoinverse technique.”

Question 20

What was the outcome of study 2?

Answer 20

• 4 min of data analysed • Filter constructed during this time • Put to use immediately (no practice effect!) • Linked to MN driven cursor, asked to track prompt again -successful but crude also can’t have the implant for long

Question 21

What is the study 3 about?

Answer 21

-High-performance neuroprosthetic control by an individual with tetraplegia -• 52 y old women • Spinocerebellar degeneration • Injury motor complete (0/5) (no motor function) -similar to the monkey

Question 22

What is the other approach to trying to repair spinal cord surgery?

Answer 22

• Robots can replace body • But paraplegics have intact muscle/ skeletal system • Can stimulate muscle contraction through skin or by implantable electrodes • Restore movement to paraplegics

Question 23

What is study 4 about?

Answer 23

-Implantable FES system for upright mobility and bladder and bowel function for individuals with spinal cord injury. -the machine on belt and activate muscles

Question 24

What is the system like in study 4?

Answer 24

• From Cochlear Pty Ltd • 22 electrodes surgically implanted on muscle or nerves • Stimulator has pre- programmed muscle stimulation sequences • Push-button control • No feed-back control -muscles to walk, stand, sit -preprogrammed motor patterns in the device on their belt, can push= walk, sit, upstairs etc.

Question 25

What was the outcome of study 4?

Answer 25

• 3 subjects • Complete motor thoracic section • All three could stand, sit and walk at least 6 m • One subject could ascend and descend stairs

Question 26

What is the issue with having feedback in study 4?

Answer 26

• Neopraxis Pty Ltd was developing external devices to provide feedback directly to controller (by-passes brain) • Monitor knee angle by electric goniometers • Accelerometers for vertical acceleration, rate gyroscopes on different parts of limbs to map limb movements -they knew their system is an open loop = no feedback for errors, now looking at giving feedback -put sensors to measure

Question 27

What was the outcome of study 4 when added feedback mechanism?

Answer 27

• Without feedback (open-loop), can stand for 10-20 minutes (lots of stimulation needed) (because of lactic acid ) • With feedback (closed-loop), can stand for 60 minutes with only 10% of muscle activity (use muscles only to balance which is more like what normal people do = so feedback is key) • Improves efficiency of movement

Question 28

What is the study 5 about?

Answer 28

-connecting the reading of motor programme and moving their own muscles • Can you use cortical activity to drive muscle electrodes? • Monkey study • Brain electrodes as previously • Connect via computer to arm muscle electrodes • Anaesthetise brachial plexus with local anaesthetic

Question 29

What does the brain-controlled functional stimulation (FES) (study 5) look like?

Answer 29

-

Question 30

What is the outcome of study 5?

Answer 30

-• Success with FES turned on = 80-90% compared to un- anaesthetised arm

Question 31

What are the human trial like in study 5?

Answer 31

• Cortex to computer interface linked via computer to non- invasive stimulating electrodes in arm • Controlled movement achieved (about level of FES monkey) -temporary implantation= could produce meningitis if contact with bugs -challenge!

Question 32

Could you put in an electrode into S1 so body would get feedback?

Answer 32

• Only feedback is visual – task impossible without it • Can we provide somatosensory feedback? • Recording electrodes can act as stimulating electrodes • Where to put them? -• Normal sensation is the result of processing by many diverse parts of the brain • Topographic maps in various places

Question 33

How could feedback be organised to provide useful information for Matt’s motor cortex?

Answer 33

• Hypothetical robotic arm (like monkey study) with sensors that can detect touch and provide an output to stimulating electrodes as feedback. -we don’t understand the coding in the somatosensory cortex, like what hurts,etc.

Question 34

What is the outcome of study 6?

Answer 34

• Tested in vivo in patients • Stimulate somatosensory thalamus – sensation “tingling, cold, hot, painful, unnatural” • In monkeys, stimulate area 3b (S1) at one of two frequencies • Frequency “tells” monkey which target to select to get reward -if put wrong signals in, then can be really bad

Question 35

What was the owl S1 stimulation about (study 6)?

Answer 35

• Food offered to in one of two closed boxes • Access initially denied, but box indicated by pattern of S1 stimulation • Monkey allowed access to boxes • Which one does it choose? -it is very good at choosing, one frequency= left and one right

Question 36

What was the setup of study 6?

Answer 36

• Monkeys previously trained with vibrator on arm • S1 on one side stimulated through 32 electrode array • Both monkeys stimulated in the hand representation of S1 • No attempt to mimic normal inputs • Learnt quickly

Question 37

What is Stimulation of proprioceptive area 3a like?

Answer 37

• Adult macaque • Monkey steers joy-stick controlled cursor onto target to get reward • Target cued by electrical stimulation of area 3A • Arm moves by M1 recording or by joystick • Region stimulation responded to arm movements

Question 38

Conclusion?

Answer 38

-if could provide feedback= and the prosthetes would be much more effective • Stimulation of sensory cortex may be “safe” • Stimulation detectable by subject • Can discriminate between different patterns of stimulus • No idea of what the subject feels • Responses are learnt, not innate