Biomedical Engineering Reference
In-Depth Information
through.the.reimaging.optics.onto.the.back.aperture.of.the.microscope.objective..For.the.results.used.
later,. the. lens. selected. was. a. Nikon. 0.75. NA. air. lens. with. a. 20. times. magniication.. his. lens. is. used.
frequently.as.it.has.a.long.working.distance.and.the.back.aperture.closely.matches.the.15.mm.diameter.
of.the.AO.mirrors,.thus.removing.the.need.for.further.beam.expansion..he.luorescent.light.is.then.col-
lected.from.the.dichroic.mirror.mounted.close.to.the.objective.and.the.signal.is.digitized.and.recorded.
in.the.computer.using.home-written.custom.sotware..Z-axis.(axial).movement.is.provided.by.mount-
ing.the.objective.on.a.piezo-driven.objective.scanner.(PI.Instruments).again.controlled.when.required.
by.the.computer.
11.3 Determination of the Mirror Shape
Having.discussed.the.design.and.build.of.a.beam-scanned.AO.microscope,.the.next.consideration.is.to.
determine.what.mirror.shape.will.be.used.to.correct.for.the.aberration..In.this.book,.there.are.several.
chapters.that.go.into. signiicant. detail.on. the.diferent. options.that.are. available. and.these.should. be.
read. in. conjunction. with. this. speciic. section. on. nonlinear. imaging.. In. the. work. described. later,. the.
approach.adopted.is.to.use.an.image.metric-based.optimization.routine.to.determine.the.best.possible.
mirror.shape..his.section.will.look.at.the.diferent.metrics.that.may.be.considered.and.which.may.be.
most.suitable.for.diferent.samples,.a.range.of.algorithms.that.can.be.used.to.optimize.the.system,.and.
guidelines.on.how.such.methods.may.be.implemented.as.well.as.some.practical.results..It.is.clear.that.
computer.programs.will.need.to.be.written.to.control.the.mirror..Visual.programming.languages.such.
as.LabView.provide.an.easy.way.to.generate.a.simple.and.elegant.user.interface.but.generally.are.slow.in.
undertaking.mathematical.calculations.where.languages.such.as.Python,.C,.and.C++.hold.the.upper.
hand..In.our.systems,.we.have.used.LabView,.C++,.and.Python,.and.indeed.combinations,.and.all.have.
worked.successfully..he.inal.choice.is.clearly.up.to.the.individual.user.and.their.expertise.and.is.not.
discussed.further.
11.3.1 image-Based Metrics and optimization Algorithms
here.are.a.number.of.metrics.that.can.be.used.to.assess.the.quality.of.an.image.but.in.the.context.of.
nonlinear.beam-scanned.microscopy,.the.considerations.are.clearly.going.to.be.diferent.from.a.con-
ventional.luorescence.single-photon,.wide-ield.microscope..One.crucial.point.is.that.the.time.taken.to.
image.the.full.ield.of.view.is.typically.around.1/10th.of.a.second.for.a.beam-scanned.microscope..Even.
with. a. guided. search,. this. means. that. the. optimization. time. will. be. very. long. if. the. system. needs. to.
undertake.say.50.iterations..To.overcome.this.problem,.the.user.typically.selects.a.small.region.of.inter-
est.and.then.images.only.this.area..he.optimization.metric.then.used.has.to.be.made.robust.against.
two.signiicant.parameters—noise.and.photobleaching—or,.no.matter.how.well.the.sotware.is.written.
or.the.system.adjusted,.the.optimization. will.not. converge.. For. this.reason,. absolute. intensity,. unless.
normalization.is.introduced,.is.not.considered.a.viable.option.for.conventional.luorescence.but.may.be.
very.suitable.in.harmonic.or.CARS.imaging.where.there.is.no.risk.of.photobleaching.
In.this.section,.we.compare.the.suitability.of.ive.metrics.when.used.on.nonlinear.microscope.images..
hese.metrics.are.given.in.
Table.11.1
..hey.were.selected.based.on.providing.a.range.of.diferent.types.
of.metrics.while.also.covering.the.most.common.metrics..he.intensity.squared.metric,.Equation.11.1,.
and.the.image.variance.metric,.Equation.11.2,.apply.directly.onto.the.image,.without.any.preprocessing..
I(x,y)
.is.the.intensity.of.the.(
x
,
y
)th.pixel,.<
I
>.is.the.average.gray.level.of.the.image,.and.
N
.is.the.number.
of.pixels..he.other.metrics.require.some.preprocessing.of.the.image.
he. metric. in. Equation. 11.3. is. the. ratio. of. the. iltered. modulus. of. the. Fourier. amplitude. over. the.
unmasked. modulus. of. the. Fourier. amplitude. with.F. being. the. Fourier. transform. symbol.. he. mask.
used. is. typically. a. square. mask. of. about. 4. pixels. (corresponding. to. around. 240. mm
−1
. in. the. micro-
scope.object.space).centered.on.the.spatial.frequency.domain..he.edge.detection.or.Sobel.ilter.metric,.
