This paper presents a multiobjective optimization strategy developed in the context of field-programmable gate array-based implementation of medical image registration that can easily be adapted to a wide range of signal processing applications, including areas of image and video processing beyond the medical domain.
Abstract:
With a multitude of technological innovations, one emerging trend in image processing, and medical image processing, in particular, is custom hardware implementation of computationally intensive algorithms in the quest to achieve real-time performance. For reasons of area-efficiency and performance, these implementations often employ limited-precision datapaths. Identifying effective wordlengths for these datapaths while accounting for tradeoffs between design complexity and accuracy is a critical and time consuming aspect of this design process. Having access to optimized tradeoff curves can equip designers to adapt their designs to different performance requirements and target specific devices while reducing design time. This paper presents a multiobjective optimization strategy developed in the context of field-programmable gate array-based implementation of medical image registration. Within this framework, we compare several search methods and demonstrate the applicability of an evolutionary algorithm-based search for efficiently identifying superior multiobjective tradeoff curves. This strategy can easily be adapted to a wide range of signal processing applications, including areas of image and video processing beyond the medical domain.
TL;DR: It is shown that a single core on 412MHz XC5VLX330T FPGA can evaluate a rigid transformation of a 3D image with 16 million voxels in 35ms, over 108 times faster than a multi-threaded implementation running on a 2.5GHz Intel Quad-Core Xeon platform.
TL;DR: This dissertation is geared toward developing high-performance 3D image processing architectures, which will enable improved intraprocedural visualization and navigation capabilities during IGIs.
TL;DR: This paper suggests a non-dominated sorting-based MOEA, called NSGA-II (Non-dominated Sorting Genetic Algorithm II), which alleviates all of the above three difficulties, and modify the definition of dominance in order to solve constrained multi-objective problems efficiently.
TL;DR: The proof-of-principle results obtained on two artificial problems as well as a larger problem, the synthesis of a digital hardware-software multiprocessor system, suggest that SPEA can be very effective in sampling from along the entire Pareto-optimal front and distributing the generated solutions over the tradeoff surface.
TL;DR: The results demonstrate that subvoxel accuracy with respect to the stereotactic reference solution can be achieved completely automatically and without any prior segmentation, feature extraction, or other preprocessing steps which makes this method very well suited for clinical applications.
Q1. What are the contributions mentioned in the paper "Multiobjective optimization of fpga-based medical image registration" ?
This paper presents a multiobjective optimization strategy developed in the context of fieldprogrammable gate array–based implementation of medical image registration. Within this framework, the authors compare several search methods and demonstrate the applicability of an evolutionary algorithm–based search for efficiently identifying superior multiobjective tradeoff curves.
Q2. What is the trend in real-time signal processing systems?
An emerging trend in real-time signal processing systems is to accelerate computationally intensive algorithmic components by mapping them to custom or reconfigurable hardware platforms, such as applicationspecific integrated circuits (ASICs) and fieldprogrammable gate arrays (FPGAs).
Q3. What is the first step in the calculation of a voxel?
The initial step in MI calculation involves applying a candidate transformation (T), to each voxel coordinate ( rv ) in the RI to find the corresponding voxel coordinates in the FI ( fv ).
Q4. What other heuristic techniques are limited to software implementations?
Other heuristic techniques that take into account tradeoffs between hardware cost and implementation error and enable automatic conversion from floating-point to fixed-point representations are limited to software implementations only [26].
Q5. What is the MI value of the FPGA-based architecture?
The authors have developed a parameterized, bit-true emulation of the FPGA-based architecture that is capable of calculating the MI valuecorresponding to any feasible configuration for a given image transformation.
Q6. What is the importance of comparing the Pareto-optimized solution sets?
Quantitative comparison of the Pareto-optimized solution sets is essential in order to compare more precisely the effectiveness of various search methods.
Q7. What is the effect of the MI calculation on registration accuracy?
These factors, along with its execution time, in their experience, may render registration accuracy as an unsuitable objective function, especially if there is nonmonotonic behavior with respect to the wordlength of design variables.
Q8. How many accesses to MH memory for each RI voxel?
Because the MH must be updated (read–modify–write) at these eight locations, this amounts to 16 accesses to MH memory for each RI voxel.