Saturday, March 22, 2008

7.29 HAL 9000

Of course you all recognize this scene from "2001: A Space Odyssey" with Frank Poole (l) and David Bowman both hiding in a pod, discussing HAL 9000's increasingly erratic behavior. Unfortunately, HAL was programmed to read lips and "he" was therefore able to learn the astronauts' intention of pulling his plug, and took steps to eliminate both humans. You can see the cyclopean HAL outside the oval window, and a close-up of HAL's eye:
Notice it does not has a clearly defined pupil, unnecessary in a controlled environment such as the interior of a spaceship, we suppose. In George Lucas's Star Wars, C3PO does not even have pupils only vertical slits.
On the other hand, the eyes of the Terminator (by James Cameron et al, 1984) are quite interesting:
A dying Terminator was actually shown with pupil reflexes gradually dimming with the ebbing (machine) life (when it was being crushed by a hydraulic press). It is quite obvious that careful thoughts have been put into the movie plot. Finally, in "I, Robot (2004, based on Issac Asimov's original work)", human-like eyes appeared:
The ultimate is of course an android named Commander Data of Star Trek TNG; although "he" is simply too human to be credible. It was probably too costly to construct a life-like robotic prop for the TV series.

In real-life, machine vision (MV) using artificial intelligence (AI) techniques, such as expert systems, fuzzy logic, inductive learning, neural networks, genetic algorithms, and swarm intelligence, is no longer a nascent field. For automation, MV has applications in manufacturing of semiconductors, electronics, pharmaceuticals, medical devices, automotive and consumer goods, and also in packaging and facial recognition.

The next challenge is the design and construction of autonomous robots (as opposed to "robots" remotely controlled by humans). In fact, true robots are beginning to make its presence felt. It is, however, a bit disconcerting to see Asimo (by Honda) possessing no discernible eye structures at all:
What gives, really.

Arthur C Clarke (1917-2008), the sci-fi visionary, was actually right - HAL-like lip-reading machines are now being developed. They will be capable of translating audio speech into visual speech, based on light reflected from the moving parts of human speech and the reflection collected by, naturally, machine vision. And the purpose for these machines? Hmm, a very good question.

Just to add more intrigue: Each letter of HAL (Heuristic Algorithmic Computer) is one alphabet before IBM, and the significance of which is still unclear.

Sunday, March 16, 2008

7.28 Optic ischemia

Did you know that Johann Sebastian Bach (1685-1750; portrait on a stamp, mit one of the Brandenburg Concertos, see above) was blinded by acute angle-closure glaucoma a few months before he died? Also, Messrs Jose Feliciano, Stevie Wonder, and the late Ray Charles all suffered from congenital glaucoma? And former Minnesota Twins baseball all-star, the Hall-of-Famer, Kirby Puckett (1960-2006) was blinded by glaucoma in his right eye?

What kind of damages you ask? Well, the definition of glaucoma especially that of OAG (open-angle glaucoma) keeps changing. It is a sign of progress; although, the answer to "what is glaucoma?" is no longer straightforward. Perhaps in the next few years, we'll know the whole story. For now, we can only work with what we have.

We'll narrow down the discussion a bit to only open-angle glaucoma: OAG has traditionally been regarded as a topical (as opposed to a systemic) issue. In other words, if the intraocular pressure (IOP) is high, then the eye may suffer pressure-induced damages to the retina and the optic nerve, in the form of ischemia. The treatment is then to normalize the pressure using pharmaceuticals. The problem is in some cases, medicine alone seems ineffective, the disease process continues nonetheless. In low-tension glaucoma, the IOP is actually normal, yet the glaucoma-related ocular damages still occur. So, something else apparently is going on, but what is it? Is it associated with systemic circulation? Shall we now institue systemic neuro-protection? And to complicate the situation even further: It also should be noted that a high IOP does not always lead to OAG.

Even more basic: How do we know optic ischemia is involved? Or more accurately, how do we measure the oxygen level in the optic nerve under high IOP? Short of inserting an oxygen-sensing electrode into the optic nerve itself, there isn't really any good way. One can opt to study animal models or examine post-mortal glaucomatous eyes. However, to perform the measurement in the humans in vivo, a special technique is needed. This is known as T2*(pronounced T2-star)-weighted or BOLD (Blood Oxygen Level Dependent) MRI - used in functional MRI of the brain. This technique takes advantage of de-oxygenated hemoglobin being a paramagnetic which disturbs the homogeneity of a magnetic field. And an increase in oxyhemoglobin, i.e., blood oxygenation, reduces this paramagnetic effect, hence an increase in the MR signal intensity.

We have invited some volunteers for a BOLD-MRI study. Using a home-made surface coil for the eye/orbit and T2* MRI performed before and after a bolus injection of Diamox, a vasodilator via its pH lowering effect. The change in MR signals in the optic nerve near the optic disc is then determined. The results are tabulated below:

Change in image intensity in the optic nerve of normal and glaucomatous patients

Diagnosis

Time after Dx

IOP (mmHg) at MRI

Treatment history

Residual Visual Field

Optic Nerve (% change)

Normal

--

15-16

none

--

+12.1 ± 2.2 (N=4)

POAG

atypical, unilateral (IOP=48 at time of Dx)

2 years

17-27

0.25% Timolol

Upper temporal quadrant

+10.6

POAG

3 years

23-29

0.5% Timolol

Lower temporal quadrant

+0.0

POAG

18 years

20

0.5% Timolol

Cetral 10°

+lower temporal quadrant

-10.5

POAG

Secondary
to trauma?

23 years

20

argon laser trabeculotomy + 0.5% Timolol

Central 5-10°

-14.9

Note: The 4 normal volunteers are all ultra-dedicated ophthalmology residents. And the study has been approved by the IRB.

Just look at the last column. Intriguing results, we might add. They essentially confirm that in the optic nerve of the living humans, as the glaucoma advances, blood oxygen level decreases. In other words, ischemia!!

Granted this is a small study. A large scale study will have to await further developments in MRI technology: (1) even higher magnetic fields (e.g., > 1.5 Tesla) to increase the signal intensity and shorten the scan time; and (2) much reduced scanning costs to justify the utilization of MRI of the eye/orbit. There has been a recent proposal that the glaucomatous changes go beyond the eye, all the way through the optic track, the LGN, ultimately to the visual cortex. Then T2* MRI will prove the only practical means for assessing ischemic changes in the whole visual pathway.

Note: For those interested, the seminal paper on fMRI (i.e., Echo-planar BOLD-MRI) is this one:

Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM, Poncelet BP, Kennedy DN, Hoppel BE, Cohen MS, Turner R, Cheng HM, Brady TJ, Rosen BR: Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci (USA) 1992;89:5675-5679.