Fast multi-core based multimodal registration of 2D cross-sections and 3D datasets.

Area of Science:

• Bioinformatics
• Computational Biology
• Medical Image Processing

Background:

• Bioinformatics tasks demand substantial computational resources.
• Heterogeneous multi-core processors like the Cell Broadband Engine (CBE) offer power-efficient, high-performance computing.
• Image analysis, specifically registering 2D cross-sections into 3D datasets, is a key application area.

Purpose of the Study:

• To evaluate the PlayStation 3 with a CBE processor as a high-performance, cost-effective computing platform.
• To adapt and optimize a multimodal alignment procedure for the CBE architecture.
• To demonstrate the benefits of multi-core implementation for image analysis tasks.

Main Methods:

• Optimized a multimodal alignment procedure using partitioning, vectorization, branch reduction, and loop unrolling.
• Focused on 32-bit multiplies and limited local storage on computing units.
• Implemented the alignment algorithm on the multi-core CBE processor.

Main Results:

• The CBE-driven PlayStation 3 demonstrated accelerated computational performance for multimodal registration.
• Optimized techniques effectively leveraged the CBE's hardware properties.
• The multi-core implementation showed significant benefits for the image analysis problem.

Conclusions:

• The Cell Broadband Engine processor in the PlayStation 3 effectively accelerates computationally intensive multimodal registration.
• This low-cost platform provides an efficient alternative for biological and medical image processing.
• The PlayStation 3 serves as a viable solution for computational challenges in image processing and bioinformatics.