VISFUN: SEMINARS - amblyopia, SDT, psychophysics Flashcards
Define amblyopia. What does it result in [Amblyopia #1]
neuro-developmental disorder of the visual cortex arising from abnormal visual experience in early life
results in decreased VA in one or both eyes
List and describe the 3 types of Amblyopia [Amblyopia #1]
- Strabismic Amblyopia - caused by active inhibition of a deviated/turned eye
- Anisometropic Amblyopia - caused by inhibition of the eye with the stronger refractive power (poorer vision eye)
- Deprivation Amblyopia - poor vision in one or both eyes due to under-stimulation of the retina via other causes
List 6 possible causes of deprivation amblyopia [Amblyopia #1]
- congenital cataracts
- corneal opacities
- blepharospasm (involuntary tight closure of eyelids)
- surgical lid closure
- unilateral complete ptosis
- prolonged patching or use of atropine drops
How does amblyopia affect contrast sensitivity? [Amblyopia #2]
Generally decreases at all spatial frequencies
Which type of amblyopia has a larger impact on stereopsis? [Amblyopia #2]
‘Strabismic amblyopia’ has a much larger effect on stereopsis than anisometropic
In fact, many anisometropic amblyopes retain some stereopsis
Where do the predominant neurophysiological changes occur in amblyopia development? What other areas are affected? [Amblyopia #3]
Predominant changes occur in V1
There are changes to areas downstream also:
- V2, LGN, MT, Macula and Optic disc
What is Signal Detection Theory? [SDT #1]
The idea that the detection of a stimulus depends on its intensity and the physical/psychological state of the individual
I.e relates to our ability as researchers/optom’s to discriminate signal from noise of stimulus and noise of observer
I.e true vs false positives/negatives
What is the clinical application of signal detection theory? [SDT #1]
Can be used to detect ocular disease (e.g glaucoma) by setting an expected response criterion
I.e probability of a ‘hit’ or ‘miss’
Define ‘sensitivity’ and ‘specificity’ [SDT #2]
Sensitivity = ability to correctly identify those with the target condition (true positives)
Specificity = ability to correctly identify those without the target condition (true negatives)
Describe the receiver-operator characteristic (ROC) curve. What are it’s axis? and how does it relate to sensitivtiy and specificity? [SDT #2]
Plots the Hit rate (sensitivity) against the rate of False alarms (False positives; 1 - specificity)
Each point on ROC curve represents a sensitivity/specificity pair corresponding to a particular decision threshold
How do you calculate sensitivity and specificity [SDT #3]
Sensitivity = True positive/True positive + False negative
Specificity = True negative/True negative + False positive
List the features of a good clinically chosen test [SDT #4]
- appropriate criterion level
- allow for early detection
- constancy in test conditions
- sensitivity + specificity
- high reproducibility + reliability
- test multiple prediction parameters for stronger prediction
- patient comfort
- within time and cost constraints
- low risk
- easy to follow + interpret
- objective + quantitative analysis
List issues for using clinical tests as a screening tool [SDT #5]
- cost + time + funding
- availability of large population
- sensitivity: which stage to screen and with which combination
- screening not absolute confirmation of disease
- screening often not sensitive to detect early stages
- bias: selection, observer/expectation and verification bias
- psychology
- variable time factors
Describe the methods to determine psycho-physical paramaters [Psychophysics #1]
- Method of Constant Stimuli - randomly presenting stimuli. threshold at 50%
- Method of Adjustment - subject increase/decrease until seen/unseen
- Method of limits - ascending or descending steps until seen/unseen
Adaptive psychophysical methods:
4. Staircase Method, 5. Maximum likelihood estimation
What is Bayesian Data Analysis? [Psychophysics #2]
Is a method of statistical inference where you use probability to represent uncertainty in a statistical model
As you get more info, the probability is updated and becomes more accurate for the chosen parameter
Also, the gathering of data gives you a distribution. With enough data, you can obtain a generative model of what to expect
