Outline Advantages of PACT Photoacoustic Computed Tomography with Applications to Breast Imaging Mark A. Anastasio Department of Biomedical Engineering Washington University in St. Louis St. Louis, MO Introduction to photoacoustic computed tomography (PACT) Imaging models and iterative image reconstruction Success with small animal imaging PACT breast imaging» System design» Preliminary clinical data PACT methods have been recently developed to overcome the limitations of other existing modalities.» Strong (hemoglobin-based) contrast similar to pure optical methods» High spatial resolution similar to pure ultrasonic methods Anatomical structures can be imaged based on endogenous hemoglobin. Hemoglobin can also serve as a functional contrast for imaging of hemoglobin oxygen saturation (so 2 ). Molecular imaging is also possible by use of exogenous agents. 3
Schematic of PAT Schematic of PAT Photoacoustic physics Optical pulse Ultrasound transducers Optical pulse Ultrasound transducers Absorbed optical energy density:» μ a (r) optical absorption coefficient» Φ s (r) optical fluence rate Image reconstruction algorithm Image reconstruction algorithm Photoacoustic wave equation (can generalize to heterogeneous media) Tissue Tissue Absorbed optical energy density Absorbed optical energy density Grueneisen parameter 2
Canonical imaging model Conventional imaging model (assuming idealized transducers)» Solution to PA wave equationa homogeneous lossless me» Continuous-to-continuous mapping (C-C) (mapping between infinite dimensional Hilbert spaces)» Assumes uniform acoustic properties (SOS), constant density, no acoustic attenuation (i.e., input and output functions are defined on continuous domains.) Alternate form: spherical Radon transform (SRT) where PACT reconstruction: Heterogeneous media If SOS variations are known, they can be incorporated into the PACT imaging model. Two general approaches: Geometrical-acoustics ( ray-based ) reconstruction: Full-wave equation-based model: C. Huang, et al. IEEE Tran. Med Imaging, 32, 203 Action of wave equation Challenges in PACT image reconstruction Modeling of acoustic physics» Speed of sound (SOS) (dominant factor for breast imaging)» Mass density» Acoustic attenuation» Shear wave physics (e.g., transcranial brain imaging) Compensating for measurement system response» Acousto-electrical impulse response (EIR)» Spatial impulse response (SIR) Mitigating data incompleteness Inherently a 3D problem computational issues 3
Iterative image reconstruction 3D integrated PACT-USCT imaging Reconstructed 3D images of the mouse body Penalized least-squares objectives are commonly employed A variety of penalty terms have been explored in PACT» Tikhonov regularization» Quadratic smoothness penalties» Sparsity promoted penalties Goal: Incorporate USCT capabilities Collaboration with Tomowave Laboratories, Inc. (A. Oraevsky) Iterative image reconstruction produces images with better spatial resolution and lower noise levels. FBP PLS-Q PLS-TV Depending on cost functions, a variety of optimization algorithms are employed for PACT image reconstruction.» For quadratic cost functions: conjugate gradient, LSQR, etc.» For cost functions containing l-type terms: shrinkage/threshold type algorithms, etc. System developer: Sergey Ermilov and colleagues at TomoWave Labs..-mm 80 tomographic views employed 2 4
3D integrated PACT-USCT imaging: Mouse study 3D integrated PACT-USCT imaging: Mouse study Rational for PACT breast cancer imaging Cryo photograph PACT Speed of sound 8 2 8 Abdominal aorta/caudal vena cava, 2 Right kidney, Intestines/abdominal fat, Vertebrae/back muscles, 8 Left kidney 2 8 2 Cryo photograph PACT Speed of sound 9 Abdominal aorta/caudal vena cava, Vertebrae/back muscles, 9 Urinary bladder, Iliac arteries/veins 9 Without a private supply of new microscopic blood vessels cancerous tumors can not grow larger than the head of a pin and are unlikely to become lethal. Without blood vessels to feed them oxygen and nutrients, these tumors remain tiny and unable to spread Judah Folkman, MD 3 4 5 5
Absorption Capability, µ a (cm - ) x % content Tumor Absorption at 4-nm, /cm Biophotonic imaging of hemoglobin Possibility for tumor classification PACT breast imaging 4 nm 5 nm 8 O2-Hb 800 nm H-Hb 4 2 H 2 O 00 00 800 900 00 0 Wavelength, nm 0.08 0.0 0.0 Benign 0.05 0.04 0.03 Malignant 0.02 0.0 0 0 0.0 0.02 0.03 0.04 0.05 0.0 0.0 0.08 Tumor Absorption at 55-nm, /cm PACT/OAT breast imaging: safe high optical contrast and high ultrasound resolution structural and functional information Problem: illumination of large object, heavy computation burden. We are developing effective system set-ups and efficient iterative methods for PACT breast imaging. Collaboration with Dr. Alexander Oraevsky at TomoWave Laborotories Inc. 8
Clinical evaluation Phantom study Breast phantom results: Backprojection Our collaborators at Tomowave Lab have constructed a prototype PACT imager. It has recently been installed at MD Anderson Cancer Center and is being evaluated in clinical breast imaging studies. Final phantom mm 4nm Sphere mm 0nm Sphere (Work by: Yang Lou) /5/205 9 /5/205 20 2
Preliminary clinical results Clinical PACT images for Patient B Clinical PACT images for Patient A The following slides show some preliminary clinical images obtained by the first-generation PACT breast imaging system built by our collaborators at TomoWave Laboratories Inc. The images are maximum intensity projection (MIP) images. Patient A collected 300 views of data, and the PACT images are reconstructed using an accelerated iterative method incorporating Total-Variation penalty. Patient B collected 800 views of data, and the PACT images are reconstructed using filtered back-projection method. We are actively working on the next-generation PACT breast imaging system with better illumination and acoustic probe design. front-back top-down 3D movie left-right front-back top-down left-right 3D movie 22 23 24 8
Optoacoustic Brightness Summary PACT provides high contrast images based on optical contrast Physiological parameters related to Hb can be obtained by acquiring multi-wavelength measurements. Ultrasound imaging provides complementary contrast and yields high resolution structural information. Information from ultrasound image can be utilized to improve PACT image quality. Non-ionizing and compression free. Fatima Anis Trey Garson Joe Poudel Acknowledgements Work supported by: NIH R0 EB0049 NIH R0 EB093 NIH R0 CA440 Collaborators Alexander Oraevsky/ Tomowave Laboratories Lihong Wang (WUSTL) http://anastasio.wustl.edu Yujia Chen Yang Lou Huifeng Guan Chao Huang Kenji Mitsuhashi Kun Wang 25 /5/205 2 50 40 30 20 Depth limitations noise floor -mm blood vessel 0 in % fat milk 0 20 30 40 50 0 0 80 m a=0.04/cm Depth, mm m s =2/cm m eff=.2/cm 2 9
Summary of advantages of hybrid PACT-US imaging Summary 3D integrated PACT-USCT imaging PACT provides high contrast images based on optical contrast Physiological parameters related to Hb can be obtained by acquiring multi-wavelength measurements. PACT is a rapidly emerging bioimaging modality with great potential for important preclinical and clinical applications.» Whole body small animal imaging» Breast cancer detection and management» Brain imaging Ultrasound imaging provides complementary contrast and yields high resolution structural information. Information from ultrasound image can be utilized to improve PACT image quality. Non-ionizing and compression free. Numerous challenges for image reconstruction exist.» Accurate modeling of physics and instrument response» Shear waves» Computational challenges Enhanced interactions between theoreticians and engineers are needed to address these challenges. LU emitter acrylic half-rod with curved surface painted black Illumination via 00 µm fiberoptic source offset by 5 mm from the flat back surface of the emitter Wide planar (transverse slice) directivity, cylindric wavefront Emit-receive directivity 28 30
PACT imaging model in heterogeneous media Photoacoustic wave equation Ultrasound-informed PACT for breast imaging Hybrid imaging system, collaboration with Tomowave Lab and MD Anderson Cancer Center» Ultrasound imaging and PACT imaging Subject to initial conditions: Measurements are defined as /5/205 32