Comparison of predictive accuracy - PowerPoint Presentation

Slide1

Comparison of predictive accuracy on Partial coherence interferometry (PCI) and swpt-source optical coherence tomographry (SS-OCT)

Choun-Ki Joo, MD, PhDSeoul St. Mary’s HospitalCatholic University of Korea, College of MedicineCatholic Institute for Visual Science

Aug 11,

2017

Slide2

Most of the modern day formulae are based upon the theoretical equation formulated by Fyoderov and its modifications. P = (1336/[AL-ELP]) – (1336/[1336/{1000/([1000/DPostRx] - V) + K} - ELP])Net corneal power (K), Axial length (AL), IOL power (P), Effective lens position (ELP), Desired refraction (DPostRx

), Vertex distance (V)The prediction of effective lens position (ELP) is the most important process in IOL power calculation.Shammas et al. JCRS 2015

Introduction

Slide3

The third-generation formulasThe most popular for calculating intraocular lens (IOL) power in cataract surgeryEstimate postoperative effective lens position (ELP) by using various preoperative biometric variables such as central corneal power and axial lengthHaigis, Hoffer Q, Holladay 1, and the SRK/T formulas.

Comparison of immersion ultrasound biometry and partial coherence interferometry for intraocular lens calculation according to Haigis Graefes Arch Clin Exp

Ophthalmol. 2000;238:765–73.Calculation of intraocular lens power: a review.

Acta

Ophthalmol

Scand. 2007;85:472–85.

Introduction

Slide4

The third-generation formulasanterior corneal power (K) anterior corneal radius of curvature (R)axial length (AL)anterior chamber depth (ACD) : the Haigis formula

Introduction

Formula

Variables for ELP prediction

Haigis

ACD,AL

Hoffer Q

pACD

,

K, AL

Holladay 1

SF(surgeon factor), K, AL

SRK/T

A(A-constant), K, AL

Slide5

Norrby S. Sources of error in intraocular lens power calculation. J Cataract Refract Surg 2008; 34: 368–376.

Major sources of prediction error for the IOL power

Slide6

Errors of third generation formulas by ocular biometric factorsDifferent consideration of variables (in each formula)Disparities in ELP estimationDisparities in predicted postoperative

refractionsThe effect of ocular biometric factors on the accuracy of various IOL power calculation formulasBMC Ophthalmology (2017) 17:62

Formula

Variables for ELP prediction

Haigis

ACD,AL

Hoffer Q

pACD

,

K, AL

Holladay 1

SF(surgeon factor), K, AL

SRK/T

A(A-constant), K, AL

Slide7

DevicesPCI(Partial Coherence Interferometry)

Axial length780 nm laser diode infrared light

Anterior chamber

depth

lateral slit-illumination

Corneal power

6 points at 2.5mm

zone

Fail

in determining the AL in the presence of a dense nuclear or posterior

subcapsular

cataract.

Because

The

signal from the retina becomes attenuated or blocked due to the light scattering inside the lens.

Slide8

SS-OCT (swept sources Optical coherence tomography)Axial length

1050~1080 nm swept laser source

Anterior chamber

depth

1050~1080 nm swept laser source

Corneal power

4 points at 3.0 mm zone

Devices

extended axial range

collection of

2-dimensional

data

->

capturing full eye

Improve

the success ratio in measuring the AL as well as the repeatability of its

measurements

whereas

the PCI unit measures ACD through a lateral slit

illumination

2-dimensional display by the OCT unit allows a more

precise

ACD measurement

independent of the subject's fixation angle

Slide9

The Argos (Movu, Inc.),collect 2-dimensional OCT data of the full eye (SS-OCT)high lateral resolution and axial resolutionmeasure not only the AL and the ACD but also the central corneal thickness (CCT), aqueous depth, lens thickness, pupil size, and corneal diameter.high-speed measurement (~30x faster than optical biometry)

SS-OCT (swept sources Optical coherence tomography)

Even in dense cataract

Slide10

Repeatability and Reproducibility of the SS-OCT measurements

Measurements with the new SS-OCT biometer were repeatable and reproducible

Axial

length measurements comparable to PCI

with a faster and higher acquisition

rate

even

in the presence of a dense nuclear or posterior

subcapsular

cataract

Bland-Altman plot

Comparison

of PCI and

SS-OCT

N=42

N=42

Slide11

The purpose of our studyTo evaluate the AL, ACD, and K measurements with the results obtained with the PCI and SS-OCT biometers To compare the refractive outcomes when 2 optical biometers were applied.

Slide12

▪ Design : Retrospective chart review▪ Setting : Seoul St. Mary’s Hospital ▪ Patients

- Between September 2016

and

Mar

2017

- 153

patients

(153 eyes)

- 1 type of IOL (

Precizon

monofocal

560,

Ophtec

, Groningen, Netherland) was implanted in the bag.

- Refractive outcomes were measured at preoperatively and 3months postoperatively

.

Method

Slide13

▪ Examination - Pre operative value : PCI versus SS-OCT Axial length (mm)

Keratometric value (diopter) Anterior chamber depth (mm) - Post operative Manifest

refraction Mean error (diopter)

Mean absolute error (diopter

) - MAE

Median absolute error (diopter

) -

MedAE

Method

Slide14

Protocols for studies of Intraocular Lens Formula AccuracyOptimization (ME zeroed out)Comparing not MAEs but median AEUsing only 1eyeUsing 1~2 IOL3months postoperatively Protocols for studies of intraocular lens formula accuracy.

Am J Ophthalmol. 2015 Sep;160(3):403-405Hoffer KJ, Aramberri

J, Haigis

W,

Olsen

T,

Savini

G,

Shammas

HJ,

Bentow

S.

Optimized IOL constants and

personlaized

IOL constants

for the PCI and SS-OCT

Retrospectively personalized IOL constants

Slide15

Patient characteristics

Result

Eyes

153

Preoperative

UDVA(

logMAR

)

0.59

0.42

CDVA(

logMAR

)

0.43

0.45

Spherical

equivalent(D)

-1.62

4.75

Age (year)

69.09

9.69

IOL

power (D)

18.07

5.43

Postoperative

UDVA(

logMAR

)

0.16

0.20

CDVA(

logMAR

)0.02

0.06

Spherical

equivalent(D)

-1.20

1.03

Eyes

153

Preoperative

UDVA(

logMAR

)

CDVA(

logMAR

)

Spherical

equivalent(D)

Age (year)

IOL

power (D)

Postoperative

UDVA(

logMAR

)

CDVA(

logMAR

)

Spherical

equivalent(D)

Slide16

Comparison of parameter measurements between the 2 biometers

Result

The

SS-OCT’s

ocular biometry measurements shows similar biometric measurements to those of conventional

biometers

.

N=153

Slide17

Intraclass correlation coefficient and 95% CI

Result

Measurements with the new SS-OCT

biometer

were repeatable and

reproducible

Slide18

Kmeany = 0.969x + 0.747R2 = 0.997

y

= 0.982x + 0.757

R

2

= 0.967

ACD

y = 0.937x + 0.159

R

2

= 0.748

Correlation

Result

AL

y

= 0.982x + 0.757

R

2

= 0.967

Slide19

Refractive outcomes at 3months postoperativelyMean error, Mean absolute error, Median absolute error

Result

PCI

SS-OCT

Barret

-Universal II

ME(D)

-0.38

0.44

-0.43

0.45

MAE(D)

0.49

0.33

0.51

0.35

MedAE

(D)

0.44

0.50

Haigis

ME(D)

-0.07

0.45

-0.13

0.44

MAE(D)

0.36

0.28

0.36

0.28

MedAE

(D)

0.29

0.30

HofferQ

ME(D)

-0.16

0.48

-0.20

0.46

MAE(D)

0.41

0.28

0.42

0.28

MedAE

(D)

0.40

0.39

SRK/T

ME(D)

-0.29

0.49

-0.33

0.50

MAE(D)

0.47

0.33

0.49

0.34

MedAE

(D)

0.41

0.46

T2

ME(D)

-0.29

0.47

-0.33

0.48

MAE(D)

0.45

0.32

0.48

0.33

MedAE

(D)

0.41

0.46

PCI

SS-OCT

Barret

-Universal II

ME(D)

MAE(D)

MedAE

(D)

0.44

0.50

Haigis

ME(D)

MAE(D)

MedAE

(D)

0.29

0.30

HofferQ

ME(D)

MAE(D)

MedAE

(D)

0.40

0.39

SRK/T

ME(D)

MAE(D)MedAE(D)0.410.46T2ME(D)MAE(D)MedAE(D)0.410.46

When the optimized IOL constants were applied

Slide20

Refractive outcomes at 3months postoperativelyMean error, Mean absolute error, Median absolute error

Result

PCI

SS-OCT

Barret

-Universal II

ME(D)

00

0.44

0.00

0.45

MAE(D)

0.34

0.29

0.36

0.27

MedAE

(D)

0.27

0.32

Haigis

ME(D)

0.00

0.45

0.00

0.44

MAE(D)

0.35

0.29

0.34

0.27

MedAE

(D)

0.28

0.31

HofferQ

ME(D)

0.00

0.48

0.000

0.46

MAE(D)

0.39

0.27

0.37

0.27

MedAE

(D)

0.35

0.35

SRK/T

ME(D)

0.00

0.49

0.00

0.50

MAE(D)

0.

0.31

0.39

0.31

MedAE

(D)

0.31

0.33

T2

ME(D)

0.00

0.47

0.00

0.48

MAE(D)

0.

0.28

0.38

0.29

MedAE

(D)

0.32

0.34

PCI

SS-OCT

Barret

-Universal II

ME(D)

MAE(D)

MedAE

(D)

0.27

0.32

Haigis

ME(D)

MAE(D)

MedAE

(D)

0.28

0.31

HofferQ

ME(D)

MAE(D)

MedAE

(D)

0.35

0.35

SRK/T

ME(D)

MAE(D)MedAE(D)0.31

0.33

T2

ME(D)

MAE(D)

MedAE(D)0.320.34

When the retrospectively personalized IOL constants were applied

After personalization, the predictive accuracies of the two optical biometers are almost the same overall. In addition, the need for optimization is greater in SS-OCT.

Slide21

MedAE

in each formula in subgroup of Axial length

Result

SRK/T

Hoffer Q

T2

Slide22

MedAE

in each formula in subgroup of Axial length

Result

Barret

-Universal II

Slide23

Best formula in each

subgrops(AL)

Result

BU II

BU II

BU II

BU II

HofferQ

HofferQ

Haigis

Haigis

Slide24

Conclusion

The

recently

introduced

SS-OCT

shows similar

biometric measurements

to those of conventional

biometers

.

After

personalization,

the

predictive accuracies of

the two

optical

biometers

are almost the same overall

.

In

addition, the need for optimization is greater in

SS-OCT.

In medium

-long

eyes, the

predictive accuracy

of the

SS-OCT

is higher, but in the shorter or longer eyes, the

PCI tend

to be more accurate.

The

Barret

-Universal II formula : one of fifth generation formula,

seems to be more appropriate for

PCI.

Slide25

So far, PCI is thought to be more accurate in IOL power prediction.However, the SS-OCT method has the advantage offaster and higher acquisition rate, even in the presence of a dense nuclear or posterior subcapsular cataract. More accurate ACD measurement being able to get image of whole eye Lens thicknessAqueous depthCentral corneal thickness

Corneal diameter

Discussion

Not

only can you get accurate measurements, but ultimately, it can be a good tool for ELP prediction.

Slide26

Thanks