William Lo

Next-generation acceleration and code optimization for light transport in turbid media using GPUs

A highly optimized Monte Carlo (MC) code package for simulating light transport is developed on the latest graphics processing unit (GPU) built for general-purpose computing from NVIDIA - the Fermi GPU. In biomedical optics, the MC method is the gold standard approach for simulating light transport in biological tissue, both due to its accuracy and its flexibility in modelling realistic, heterogeneous tissue geometry in 3-D. However, the widespread use of MC simulations in inverse problems, such as treatment planning for PDT, is limited by their long computation time. Despite its parallel nature, optimizing MC code on the GPU has been shown to be a challenge, particularly when the sharing of simulation result matrices among many parallel threads demands the frequent use of atomic instructions to access the slow GPU global memory. This paper proposes an optimization scheme that utilizes the fast shared memory to resolve the performance bottleneck caused by atomic access, and discusses numerous other optimization techniques needed to harness the full potential of the GPU. Using these techniques, a widely accepted MC code package in biophotonics, called MCML, was successfully accelerated on a Fermi GPU by approximately 600x compared to a state-of-the-art Intel Core i7 CPU. A skin model consisting of 7 layers was used as the standard simulation geometry. To demonstrate the possibility of GPU cluster computing, the same GPU code was executed on four GPUs, showing a linear improvement in performance with an increasing number of GPUs. The GPU-based MCML code package, named GPU-MCML, is compatible with a wide range of graphics cards and is released as an open-source software in two versions: an optimized version tuned for high performance and a simplified version for beginners (http://code.google.com/p/gpumcml).

Hardware acceleration of a Monte Carlo simulation for photodynamic therapy treatment planning

Monte Carlo (MC) simulations are being used extensively in the field of medical biophysics, particularly for modeling light propagation in tissues. The high computation time for MC limits its use to solving only the forward solutions for a given source geometry, emission profile, and optical interaction coefficients of the tissue. However, applications such as photodynamic therapy treatment planning or image reconstruction in diffuse optical tomography require solving the inverse problem given a desired dose distribution or absorber distribution, respectively. A faster means for performing MC simulations would enable the use of MC-based models for accomplishing such tasks. To explore this possibility, a digital hardware implementation of a MC simulation based on the Monte Carlo for Multi-Layered media (MCML) software was implemented on a development platform with multiple field-programmable gate arrays (FPGAs). The hardware performed the MC simulation on average 80 times faster and was 45 times more energy efficient than the MCML software executed on a 3-GHz Intel Xeon processor. The resulting isofluence lines closely matched those produced by MCML in software, diverging by only less than 0.1 mm for fluence levels as low as 0.00001 cm(-2) in a skin model.

FPGA-based Monte Carlo Computation of Light Absorption for Photodynamic Cancer Therapy

Photodynamic therapy (PDT) is a method of treating cancer that combines light and light-sensitive drugs to selectively destroy cancerous tumours without harming the healthy tissue. The success of PDT depends on the accurate computation of light dose distribution. Monte Carlo (MC) simulations can provide an accurate solution for light dose distribution, but have high computation time that prevents them from being used in treatment planning. To alleviate this problem, a hardware design of an MC simulation based on the gold standard software in biophotonics was implemented on a large modern FPGA. This implementation achieved a 28-fold speedup and 716-fold lower power-delay product compared to the gold standard software executed on a 3 GHz Intel Xeon 5160 processor. The accuracy of the hardware was compared to the gold standard using a realistic skin model. An experiment using 100 million photon packets yielded a light dose distribution that diverged by less than 0.1 mm. We also describe our development methodology, which employs an intermediate hardware description in SystemC prior to Verilog coding that led to significant design effort efficiency.

GPU-accelerated Monte Carlo simulation for photodynamic therapy treatment planning

Recent improvements in the computing power and programmability of graphics processing units (GPUs) have enabled the possibility of using GPUs for the acceleration of scientific applications, including time-consuming simulations in biomedical optics. This paper describes the acceleration of a standard code for the Monte Carlo (MC) simulation of photons on GPUs. A faster means for performing MC simulations would enable the use of MC-based models for light dose computation in iterative optimization problems such as PDT treatment planning. We describe the computation and how it is mapped onto the many parallel computational units now available on the NVIDIA GTX 200 series GPUs. For a 5 layer skin model simulation, a speedup of 277x was achieved on a single GTX280 GPU over the code executed on an Intel Xeon 5160 processor using 1 CPU core. This approach can be scaled by employing multiple GPUs in a single computer - a 1052x speedup was obtained using 4 GPUs for the same simulation.

Longitudinal, 3D Imaging of Collagen Remodeling in Murine Hypertrophic Scars In Vivo Using Polarization-Sensitive Optical Frequency Domain Imaging

Hypertrophic scars (HTS), frequently seen after traumatic injuries and surgery, remain a major clinical challenge because of the limited success of existing therapies. A significant obstacle to understanding HTS etiology is the lack of tools to monitor scar remodeling longitudinally and noninvasively. We present an in vivo, label-free technique using polarization-sensitive optical frequency domain imaging for the 3D, longitudinal assessment of collagen remodeling in murine HTS. In this study, HTS was induced with a mechanical tension device for 4-10 days on incisional wounds and imaged up to 1 month after device removal; an excisional HTS model was also imaged at 6 months after injury to investigate deeper and more mature scars. We showed that local retardation and degree of polarization provide a robust signature for HTS. Compared with normal skin with heterogeneous local retardation and low degree of polarization, HTS was characterized by an initially low local retardation, which increased as collagen fibers remodeled, and a persistently high degree of polarization. This study demonstrates that polarization-sensitive optical frequency domain imaging offers a powerful tool to gain significant biological insights into HTS remodeling by enabling longitudinal assessment of collagen in vivo, which is critical to elucidating HTS etiology and developing more effective HTS therapies.

Preventing Scars after Injury with Partial Irreversible Electroporation

Preventing the formation of hypertrophic scars, especially those that are a result of major trauma or burns, would have enormous impact in the fields of regenerative and trauma medicine. In this report, we introduce a noninvasive method to prevent scarring based on nonthermal partial irreversible electroporation. Contact burn injuries in rats were treated with varying treatment parameters to optimize the treatment protocol. Scar surface area and structural properties of the scar were assessed with histology and non-invasive, longitudinal imaging with polarization-sensitive optical coherence tomography. We found that partial irreversible electroporation using 200 pulses of 250 V and 70 μs duration, delivered at 3 Hz every 20 days during a total of five therapy sessions after the initial burn injury, resulted in a 57.9% reduction of the scar area compared with untreated scars and structural features approaching those of normal skin. Unlike humans, rats do not develop hypertrophic scars. Therefore, the use of a rat animal model is the limiting factor of this work.

Laser thermal therapy monitoring using complex differential variance in optical coherence tomography

Conventional thermal therapy monitoring techniques based on temperature are often invasive, limited by point sampling, and are indirect measures of tissue injury, while techniques such as magnetic resonance and ultrasound thermometry are limited by their spatial resolution.xa0 The visualization of the thermal coagulation zone at high spatial resolution is particularly critical to the precise delivery of thermal energy to epithelial lesions. In this work, an integrated thulium laser thermal therapy monitoring system was developed based on complex differential variance (CDV), which enables the 2D visualization of the dynamics of the thermal coagulation process at high spatial and temporal resolution with an optical frequency domain imaging system. With proper calibration to correct for noise, the CDV-based technique was shown to accurately delineate the thermal coagulation zone, which is marked by the transition from high CDV upon heating to a significantly reduced CDV once the tissue is coagulated, in 3 different tissue types ex vivo: skin, retina, and esophagus. The ability to delineate thermal lesions in multiple tissue types at high resolution opens up the possibility of performing microscopicxa0image-guided procedures in a vast array of epithelial applications ranging from dermatology, ophthalmology, to gastroenterology and beyond.

Accelerated 3D Monte Carlo light dosimetry using a graphics processing unit (GPU) cluster

This paper presents a basic computational framework for real-time, 3-D light dosimetry on graphics processing unit (GPU) clusters. The GPU-based approach offers a direct solution to overcome the long computation time preventing Monte Carlo simulations from being used in complex optimization problems such as treatment planning, particularly if simulated annealing is employed as the optimization algorithm. The current multi- GPU implementation is validated using a commercial light modelling software (ASAP from Breault Research Organization). It also supports the latest Fermi GPU architecture and features an interactive 3-D visualization interface. The software is available for download at http://code.google.com/p/gpu3d.

Feature space optimization for virtual chromoendoscopy augmented by topography.

Optical colonoscopy is the preferred modality for the screening and prevention of colorectal cancer. Chromoendoscopy can increase lesion detection rate by highlighting tissue topography with a colored dye, but is too time-consuming to be adopted in routine colonoscopy screening. We developed a fast and dye-free technique that generates virtual chromoendoscopy images that incorporate topography features acquired from photometric stereo endoscopy. We demonstrate that virtual chromoendoscopy augmented by topography achieves similar image quality to conventional chromoendoscopy in ex-vivo swine colon.

Towards in vivo laser coagulation and concurrent optical coherence tomography through double-clad fiber devices

There is a strong clinical need for an optical coherence tomography (OCT) system capable of delivering concurrent coagulation light enabling image-guided dynamic laser marking for targeted collection of biopsies, as opposed to a random sampling, to reduce false-negative findings. Here, we present a system based on double-clad fiber (DCF) capable of delivering pulsed laser light through the inner cladding while performing OCT through the core. A previously clinically validated commercial OCT system (NVisionVLE, Ninepoint Medical) was adapted to enable in vivo esophageal image-guided dynamic laser marking. An optimized DCF coupler was implemented into the system to couple both modalities into the DCF. A DCF-based rotary joint was used to couple light to the spinning DCF-based catheter for helical scanning. DCF-based OCT catheters, providing a beam waist diameter of 62μm at a working distance of 9.3mm, for use with a 17-mm diameter balloon sheath, were used for ex vivo imaging of a swine esophagus. Imaging results using the DCF-based clinical system show an image quality comparable with a conventional system with minimal crosstalk-induced artifacts. To further optimize DCF catheter optical design in order to achieve single-pulse marking, a Zemax model of the DCF output and its validation are presented.