Glaucoma Advanced, LAbel-free High resolution Automated OCT Diagnostics GALAHAD

Transcription

1 Project Overview Glaucoma Advanced, LAbel-free High resolution Automated OCT Diagnostics GALAHAD Jul-2017

2 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 2

3 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 3

4 GALAHAD key facts Horizon 2020 project funded by the European Union ICT : Photonics KET Project No Budget: c. 4 M Ten partners 4 four industrial, 3 academic, 2 clinical, 1 management Start date: 01-Dec-2016 Duration: 36 months Coordinator: Gooch & Housego (Torquay) Project public website: Page 4

5 Consortium Partner Main role in project Optical components and sub-systems for OCT [Coordinator] OCT systems integrator Supercontinuum sources Diffraction gratings and spectrometers Supercontinuum source design and simulation Algorithm development Data collection (new lab data) and analysis Data collection (existing patient data) and analysis Data collection (existing patient data) and analysis Project management and admin Page 5

6 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 6

7 Motivation: potential for glaucoma screening Glaucoma is the second leading cause of blindness globally (after cataracts) Increased intra-ocular pressure Slow onset and gradual degeneration Irreversible damage to the optic nerve head Universal screening could save 4M cases of blindness p.a. worldwide Cost-effective testing is not possible with existing technology Vertical transfoveal OCT of an eye with epiretinal fibrosis Image courtesy of Rigshospitalet (Copenhagen) OCT A candidate screening technology is optical coherence tomography Offers non-contact label-free high-resolution retinal imaging BUT technical developments are required for glaucoma screening Page 7

8 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 8

9 UHR-PS OCT Ultra-high resolution (UHR-) OCT Leading commercial systems achieve c. 3-5 µm axial resolution Glaucoma screening requires an axial resolution of c. 1 µm This requires broader bandwidth source and components Polarisation sensitive (PS-) OCT Can identify birefringent retinal features These may improve glaucoma diagnosis This requires polarisation maintaining (PM) components GALAHAD hardware objective To develop ultra-high resolution polarisation sensitive OCT Page 9

10 Automated algorithms Automated image analysis based on artificial intelligence Could allow cost-effective glaucoma screening Intelligent algorithms will be developed useable by non-experts Automatically process and analyse OCT images Identification of patients requiring closer expert examination GALAHAD software objective Automated algorithms to identify high risk patients from OCT images Page 10

11 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 11

12 Supercontinuum sources Pulsed fibre lasers are coupled with supercontinuum (SC) generating fibres Allows fibre delivery of output spectrum Broadband light from UV to near-ir In GALAHAD, NKT will develop two different sources: 1. Low risk option - To ensure project delivery - Randomly polarised, low noise, incoherent SC source 2. High risk option - New state-of-the-art source - Linearly polarised, ultra-low noise, coherent SC source power [mw/nm] Extreme Red P=500mW wavelength [nm] Page 12

13 SC sources (SCSs) competition The resolution of an OCT system depends largely on the spectral bandwidth of the interrogating source SCSs can comfortably provide sufficient optical power across the nm window suitable for ocular OCT Other options: Lamps: brightness is too low for OCT SLD* and ASE sources have relatively narrow bandwidths Ti:sapphire lasers are large, expensive and not rugged enough Swept sources mainly access the 1300 nm band The noise level of the OCT source is a critical consideration in the GALAHAD project: Ti:sapphire and the incoherent SCS can provide similar results Coherent SCSs provide the lowest possible noise for any OCT technology. * Super Luminescent Diode Amplified Spontaneous Emission Page 13

14 State-of-the-art SCSs The normally-pumped, linearly polarised, ultra-low noise, coherent SCS is the best available source for OCT. New SCSs are pumped with allnormal dispersion fibre. This produces temporally coherent pulses Temporal coherence reduces shotto-shot noise Background noise of OCT measurement is reduced Linearly polarised SCSs allow for polarisation sensitive OCT New image information available on birefringent objects and features Features (like blebs ) can be observed Figure shows the strong birefringence of blebs. (a) Intensity OCT, (b) polarisationsensitive OCT, and (c) highly birefringent regions from (b) overlaid on (a). Page 14

15 Femtosecond laser sources available at DTU Laser description Max. Average power/ W Pulse length/ fs Rep rate/ MHz Wavelength/ nm OCT viability Picosecond mode-locked laser compressed with a hollow-core fiber Fianium femtosecond laser (Commercially available) Yes, but max SC band: nm Yes Max SC band nm Optical parametric amplifier No (bulky) Measured supercontinua from 1 m (red) and 9 m (black) fibre Calculated (dashed red) with 100 kw and 170 fs pumped at 1064 nm Inset: Microscope image of the ANDi (all normal dispersion) fibre. Bandwidth (black) from 950 nm at -10 db w.r.t. 800 nm vs average power measured at the fiber output. Estimated average power (blue) from nm. Pumping with 170 fs at 1064 nm.

16 Conclusions on pump options Calculated spectrally averaged RIN (relative intensity noise) for different pulse lengths along 1 m of fiber. Pumping along the slow-axis with 170 fs at 1064 nm and 44 kw peak power. L B is the optical wave breaking distance. Measured (squares/solid) and calculated (stars/dashed) RIN for different output average powers using 170 fs at 1064 nm. The calculated RIN is for pumping at 20 w.r.t. slow axis. 1. With the current configuration (ANDi fiber and 170 fs, 80 MHz), high average powers are needed to reach 650 nm. 2. Simulations taking into account the two polarizations of the fundamental mode show that pumping along the slow-axis of the ANDi fiber, the RIN is low for short fs pulses even for high power and long fiber length. 3. Simulations and experiments show that noise increases for higher pump powers, when the fiber is NOT pumped along the slow axis.

17 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 17

18 Diffraction gratings Ibsen core technology is planar, surface relief transmission gratings in fused silica Produced by holographic or photo-lithographic patterning and reactive ion etching In GALAHAD Ibsen will work to develop its fundamental grating technology: Cost-effective grating manufacturing process Large, high efficiency broadband transmission Polarisation insensitive GALAHAD targets: Large area gratings Low PDL (polarisation dependent loss) >90% diffraction efficiency 300 nm bandwidth Low grating volume pricing Page 18

19 Spectrometers Ibsen strengths in spectrometer manufacture In-house transmission gratings Robust and athermal design High throughput Flexibility for OEM integration In GALAHAD the objective is a low cost polarisation insensitive broadband spectrometer GALAHAD spectrometer targets: High performance, high resolution Temporally stable, broadband <1800 cm 3 for UHR-OCT Low volume pricing Focus lens Detector array Collimation lens Transmission grating Page 19

20 Fused component development Development of beyond state-of-the-art fused couplers Cost effective and rugged components Suitable for Ultra High Resolution (UHR-) & Polarisation Sensitive (PS-OCT) Specific advances in wavelength-flattened SM & PM couplers Target bandwidth: SM & PM 300 nm Target flatness: SM ±5% PM ±10% Page 20

21 Wideband Optical Delay Line An optical delay line is used in OCT systems to match the reference and sample optical path lengths For optimum performance, the delay line needs to be dynamically adjusted Allows for changes in eye position between patients New design of small form factor motorised optical delay lines Wideband PM device High stability and reliability over travel and temperature Low loss 80 mm/s travel with <10 µm resolution Page 21

22 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 22

23 Algorithm development Development of retinal layer segmentation algorithms Initially based on conventional OCT images Image processing algorithms for significant feature extraction Refinement and testing from new data acquired in GALAHAD GALAHAD target: Development and validation of a glaucoma screening algorithm Automatic, robust and clinically usable Traffic light output to identify high risk patients for further assessment OCT image Processed image Assessment Automated feature extraction Automated image analysis Page 23

24 Algorithm development Page 24

25 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 25

26 Data collection and demonstration Development of test standards including cellular, tissue and animal models for UHR OCT performance analysis and demonstration Data collection from test standards and animal models for algorithm training and demonstration Experimental characterization of optical tissue properties utilizing OCT and digital holographic microscopy OCT tomographic image of a cellular 3D test standard 200 µm OCT image of a 3D artificial model including different tissue layers Characterization of tissue birefringence properties with polarization sensitive interferometric techniques Tissue refractive index maps with corresponding cross-sections (mouse model, ex vivo)

27 Demonstration Establish two independent UHR-PS-OCT systems Evaluate on acquired and inherited small animal glaucoma models GALAHAD targets: Reduced screening costs (cf. current cost of ~300 ) Traffic light algorithm response to alert physicians to potential glaucoma. Hardware Software & analysis Lab demo Alignment, control, and data acquisition Reference arm & fibre optics Machine learning and algorithm training Test demo system 1 Supercontinuum source Big data analysis Test demo system 2 Spectrometer Page 27

28 Presentation outline Project key facts Motivation Project objectives Project technology Photonic sources Components and sub-systems Automated algorithms Demonstration Summary Page 28

29 Project Summary Glaucoma is the second leading cause of blindness globally Universal screening could save 4M cases of blindness p.a. worldwide Ultra-high resolution OCT is a key candidate screening technology Polarisation sensitive OCT could offer new information to clinicians UHT PS OCT requires Very broadband polarised sources: SCS Very broadband polarisation maintaining delay lines and components Automated algorithms could assess OCT images Traffic light output to indicate high risk patients for further testing Demonstration on rodents to be performed at two hospital laboratories Page 29

30 Thank you for your attention!! Technical enquiries Liam Henwood-Moroney Gooch and Housego (Torquay) Other enquiries Bruce Napier Vivid Components Page 30