Skin Microcirculation Imaging
Transcription
Skin Microcirculation Imaging
- LightHOUSE – Centre for Photonics & Imaging June 6 – 12, 2012 www.nbipireland.ie BIGSS’12 –martin.leahy@ul.ie Graduate Summer School Outline • • • • • • • • Microcirculation Imaging Laser Doppler monitoring and imaging Laser Speckle Diffuse Reflectance (TiVi) MI State of the Art PAT OCT Correlation mapping OCT (cmOCT) Why BioPhotonics? Courtesy of Frank Chuang Label-free Imaging Domains 1000 Standard US MRI, X-ray, CT Ultrasound High frequency US LDPI TiVi www.nbipireland.ie/education Resolution (mm) LSPI 100 DOT PET Radionucleotide 10 Optical Coherence Tomography 1 PAT Confocal Microscopy 1 10 Sampling depth (mm) 100 Image from www.biophonticsWorld.com Why microcirculation? Ryan Ryan, TJ, 1966 The microcirculation of the skin in old age. Gerontologia Clinica Motivation • Microcirculation serves key functions within the body: – Exchange nutrients and metabolic waste to body – Regulate body temperature. – Regulate blood pressure. • Structural changes associated with disease – Diabetes – Raynaud’s syndrome – Cancer Sound theoretical base Bonner & Nossal Bonner and Nossal, 1981. Model for laser Doppler measurements of blood flow in tissue, Appl. Opt. 20, 2097–2107. Martin Leahy, DPhil Director of R&D Dai Chaplin, Ph.D. Head of Research and Development & Chief Scientific Officer British Journal of Cancer 74, 260 – 263 (1996) Laser Doppler and Combretastatin Microcirculation Techniques Imaging techniques Laser Doppler perfusion imaging (LDPI) Doppler Optical coherence tomography (OCT) Optical Microangiography (OMAG) source (PAT) detector Photoacoustic Tomography Sequential raster scanning of tissue creating a colour coded “perfusion” image of underlying vasculature Example during brachial Cannot monitor real-time changes in microvasculature artery occlusion skin surface Leahy, M.J., de Mul, F.F.M, Nilsson, G.E., and Maniewski, R Principles and Practice of the laser Doppler perfusion technique, TECHNOLOGY AND HEALTH CARE, pp 143-162 7, 1999 Laser Speckle • The line scanner generates quite good images that look like ordinary LDPI images - a new image can be generated every 10 second or faster. • “the image acquisition times are much shorter - 50 x 64 pixels in 5 seconds!” • The FLPI unit generates realtime images (or close to real time) • What the images really display? - How TiVi Works G B white light CR image Light detector LP 1 2 LP SR≈7% melanin layer DP BS≈46% epidermis capillary loops 1,2 = polarisers LP= Linear Polarised DP= Depolarised SR = surface reflection BS = backscattered CR = cross polarised O’Doherty PhD Thesis. University of Limerick , 2007 R TiVi Algorithm TiViindex k gain Where: Rd ( g ) Rd ( r ) 8m ( ) (8ma ( )) 8ma ( ) R d ( ) 1 a 2 2 3ms ( ) (3ms ( )) 3ms ( ) 2 1 2 ma RBC f ma RBC 1 RBC f maTISSUE ms RBC f ms RBC 1 RBC f msTISSUE J. Biomed. Opt. 14 (3) 2009 2 ( maEPID ( g ) maEPID ( r )) x Rd ( r ) ke Skin Res and Tech – 13 (4) 2007 Kubelka-Munk theory facilitates an algorithm which is sensitive to RBCs only (absorption changes are large between red (λ ≈ 600 - 700 nm) and green (λ ≈ 500 - 600 nm) light. Light in surrounding dermal tissue is absorbed to approximately the same amount in red and green • Radial analysis – Variable isodose diameter – Minimum of 0.1 mm (LDPI = 1 mm) ITC6 O’Doherty, J., et al., 2011 .Arch Derm Res (2010) Minimal erythemal dose Swelling reduces TiVi index value 18% reduction in averaged value of dashed box while the edges increase ITC6 J. Physiological Measurement 31 (11) N79-N83 (2010) Mobile platform J. Biophotonics 4 (5) 293-296. Mobile platform ITC6 Nokia Photoacoustic Computed Tomography (1) Laser pulse (<ANSI limit: e.g., 20 mJ/cm2) (2) Light absorption & heating (~ mK) 1 mK 8 mbar = 800 Pa (4) Ultrasonic detection (optical scatter/1000) Nature Biotech. 21, 803 (2003). (3) Ultrasonic emission (~ mbar) Photoacoustic Microscopy of Human Palm Max amplitude projection 5 4 3 2 1 B-Scan @ 584 nm Epiderm.-derm. junction Stratum corneum 1 mm Epidermis 1 2 Dermis 3 4 Subpapillary plexus 5 1 mm C. Favazza, unpublished. Collaboration: L. Cornelius • Correlation mapping OCT • 8 sequential frames • 2-D correlation map average correlation value for a square grid measuring 7x7 ITC6 Enfield, J. 2011 Biomedical Optics Express 2 (5) 1184-1193. cmOCT Jonathan et al. 2011 J. Biophotonics 4 (5) cmOCT Background • Optical Coherence Tomography (OCT) is a technique that allows imaging of highly scattering mediums with micron resolution. • It is analogous to ultrasound except is uses reflections of light. • The contrast mechanism is scattering. 10 mm x 10 mm x 3 mm 1 mm x 1 mm x 0.1 mm OCT System • All work has been performed on a commercial OCT system (Thorlabs OCS1300). • Specifications: – – – – – Wavelength : 1325 nm Axial Resolution : 9 µm (water) Lateral Resolution : 25 µm Ascan Rate : 16 kHz Volume Capture (1024x1024x512) : 70-80 sec Background • Optical Coherence Tomography (OCT) can visualize structural features within the skin • Several Technologies have been developed to enable extraction of flow information using OCT. – Doppler OCT (DOCT) – Speckle variance OCT (svOCT) – Optical Micro-angiography (OMAG) • However each have associated limitations. Background • Recently a new technique has been developed within the group for flow extraction from OCT datasets. • Based on correlation statistics and called “Cross correlation OCT” or cmOCT. • Addresses key issues with existing technologies – Angle Dependence – Speed of processing Jonathan, E., Enfield, J. and Leahy, M. J. , “Correlation mapping method for generating microcirculation morphology from optical coherence tomography (OCT) intensity images”. Journal of Biophotonics, Online Dec. 2010 . doi: 10.1002/jbio.201000103 Principle of cmOCT 200 µm embedded capillary tube with flowing fluid Excised section of Pig Skin Principle of cmOCT • To compare the images, the correlation between subsequent frames is determined. • Correlation provides a measure of the similarity between datasets. • Correlation values range from -1 to +1. – Higher correlation indicates images are same – Lower correlation indicates images are different Principle of cmOCT Frame A 𝑀 𝑁 Correlation Image Frame B 𝐹𝐴 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐴 (𝑥, 𝑦) 𝐹𝐵 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐵 (𝑥, 𝑦) 𝐶𝐶(𝑥, 𝑦) = 𝑝=0 𝑞=0 𝐹𝐴 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐴 (𝑥, 𝑦) Where M,N are the kernel size 2 + 𝐹𝐵 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐵 (𝑥, 𝑦) 2 Principle of cmOCT Frame A 𝑀 𝑁 Correlation Image Frame B 𝐹𝐴 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐴 (𝑥, 𝑦) 𝐹𝐵 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐵 (𝑥, 𝑦) 𝐶𝐶(𝑥, 𝑦) = 𝑝=0 𝑞=0 𝐹𝐴 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐴 (𝑥, 𝑦) Where M,N are the kernel size 2 + 𝐹𝐵 𝑥 + 𝑝, 𝑦 + 𝑞 − 𝐹𝐵 (𝑥, 𝑦) 2 Principle of cmOCT Static region (>0.7) Flow regions (<0.2) Cross Correlation Background (<0.2) -1 +1 Principle of cmOCT Threshold and binarize image Frame n Correlation Image Mask Image Cross Correlation Mask Image +0.6 -0.6 Frame n+1 cmOCT image -0.6 +0.6 Principle of cmOCT • This cmOCT algorithm is capable of extracting the flow information from the OCT data. Structural Image Flow Image Principle of cmOCT • To generate 3D vessel maps, a spatial separation between frames is required. • If spatial separation is too high, correlation is lost for static structure. Principle of cmOCT 1 Static Region Corellation 0.8 Flow Region 0.6 Axial resolution of lens 0.4 0.2 0 0 5 10 15 20 25 30 35 Frame Separation (µm) 40 45 • A frame separation of < 5 µm will provide sufficient oversampling and is an instrumentation limit. 50 Principle of cmOCT • cmOCT has been implemented using customised java code optimized for multithread processors. • Processing speed for volume of 1024 x 1024 x 512 voxels – 3x3 kernel : 28 s – 5x5 kernel : 71 s – 7x7 kernel : 119 s Multi-layered Phantom • 3D phantom (3 x 3 x 3 mm) • 200 µm capillary tubes embedded in a static scattering matrix. • Tubes filled with intralipid solution moving under Brownian motion. • Brownian motion can be clearly seen. In-vivo Human Results • The technique is capable of mapping large regions of the tissue 2 mm Jonathan, E. Enfield, J., and Leahy, M.J. 2010. Correlation mapping method for generating microcirculation morphology from optical coherence tomography (OCT) intensity images. J. Biophotonics (published online 17 December 2010). http://onlinelibrary.wiley.com/doi/10.1002/jbio.201000103/abstract In-vivo Human Results • Limited results published on in-vivo human imaging to date. 2.5 mm • OCT (gray) and cmOCT of volar forearm is shown. – Capture : 70 s – Processing : 119 s (7x7 ker) 2.5 mm • Clear vessel structure extracted : What vessels are seen? Vascular Supply Papillary Layer Epidermis Dermis Sub-cutaneous Capillaries/superficial plexus (3-10 µm) arterioles/venules (12-35 µm) arteries/veins tissue Image from : http://www.scf-online.com/ In-vivo Human Results • To determine location depth slices can be examined In-vivo Human Results In-vivo Human Results Oral Mucosa Imaging • The following images show preliminary results of imaging the oral mucosa using correlation mapping optical coherence tomography (cmOCT). • Due to the system used, the lip was imaged. • Two regions are shown 1. Oral Mucosa 2. Minor Salivary Gland 2 1 3D Rendering • Oral mucosa sweat gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland Minor Salivary gland • C-Scan for Structural OCT and cmOCT at a depth of 350 µm • cmOCT clearly provides a new contrast mechanism. Human Wound Healing Human Wound Healing Human Wound Healing 1 mm 1 mm Human Optical Clearing Before 400 µm 2 cm 2 mm In Vivo Human Optical Clearing T = 0 min 400 µm 2 cm 2 mm In Vivo Human Optical Clearing T = 20 min 400 µm 2 cm 2 mm In Vivo Human Optical Clearing T = 40 min 400 µm 2 cm 2 mm In Vivo cmOCT and Optical Clearing Before Clearing After Clearing Region shown is a 3x3 mm region, before and after 40 min clearing In Vivo cmOCT and Optical Clearing Before Clearing After Clearing Ex Vivo OCT and Optical Clearing After 30 min clearing Before Clearing 3 mm 3 mm 5 mm 5 mm OCT Reflectance (AU) 1 Before Clearing After Clearing 0.8 0.6 1.50 mm 0.4 2.15 mm 0.2 0 0.00 0.50 1.00 1.50 Position (mm) 2.00 2.50 3.00 In Vivo cmOCT and Optical Clearing Before Clearing After Clearing After Clearing Before Clearing OCT Signal (AU) 1 0.8 0.6 0.4 0.2 0 -200 0 200 400 600 800 1000 Depth (µm) 1200 1400 1600 1800 2000 In Vivo cmOCT and Optical Clearing • Using the model, the measured scattering coefficients (µs) are : – Before : 9.40±0.31 mm -1 – After : 8.29±0.23 mm-1 • From before, the mean free path between scattering events (ls) • This indicates an increase by 13% occurs after clearing. • There is thus an increased penetration depth of OCT signal achieved. Human Reactive hyperaemia 1.5 mm 1.5 mm Human Reactive hyperaemia • To improve this a 256x256 region can be acquired in 5 s. • The scanning area is reduced to 500x500 µm so a small region of microcirculation is imaged. 500 µm 500 µm Human Reactive hyperaemia Depth Resolved RH Summary • MI at clinically relevant speeds and depths is close • cmOCT is a new power technology for flow extraction from structural OCT images. • The algorithm can be applied to any structural OCT images to extract flow. • The technique is still being developed and enhanced, however the initial results show great promise for microcirculation imaging. NBIPI: Tissue Optics and Microcirculation Imaging Facility University of Surrey Dr Jim O’Doherty Imperial College London Dr Neil Clancy Washington U., St. Louis Prof. Lihong Wang OHSU, Portland Profs. Steve Jacques, Ricky Wang University Hospital Linköping Dr Chris Anderson Joachim Henricson Wheelsbridge AB Prof Folke Sjöberg Prof. Gert Nilsson NUI Galway & U. Limerick Prof Martin Leahy Prof Valery Tuchin (adjunct) Prof Terence Ryan (adjunct) Dr Marie-Louise O’Connell Dr Azhar Zam Dr Hrebesh Subhash Dr Sergey Alexandrov Dr Joey Enfield Paul McNamara Dennis Warncke Kate Lawlor Olga Zhernovaya Susan McElligott Roshan Dsouza Haroon Zafar Gillian Lynch August 28 - Sept 1 www.nbipireland.ie BIGSS’12 –martin.leahy@ul.ie Graduate Summer School