Computation in Medicine: Medical Image Analysis and Visualization

Abstract

Computation in medicine has recently revolutionized those ideal procedures for translating fundamentally proven mathematical concepts in medical imaging and analysis into relevant routines of algorithms. Modern computational techniques, such as CUDA, a parallel computing platform, enabling direct access to the GPU instruction and parallel processing capability, are currently providing flexibility in the use of high-performance computational approaches. Similarly are the other software optimization procedures that assure low-cost and high-throughput visualization of medical datasets. Without mincing words, significant impact of such hardware and software optimization algorithms in medical image analysis and visualization cannot be overemphasized. In the same vein, acquisition of appropriate clinical datasets plays a great role in the accurate diagnosis of diseases and therapy management. The use of appropriate datasets and suitable image modalities are both important in order to successfully prove the effectiveness of any applied computational approaches in medical image analysis and visualization. Moreover, data reconstruction and representation from 2-D to 3-D usually follow notable mathematical approaches such as Euclidean plane, projective plane, and Cartesian coordinate systems and involve other interactive properties such as rotation, scaling, and translation which are also relying on various renderable concepts of data representation. This chapter documents some of the image procedures for acquiring morphological and functional information of patients with more emphasis on mathematical computations of commonly used techniques, such as X-ray, computed tomography (CT), and magnetic resonance imaging (MRI). Interestingly, a typical framework for medical imaging and visualization has been conceptualized in the course of this documentation. Relevant approaches to medical data representation, restructuring, and modeling procedures such as volume segmentation, classification, shading, gradient computation, interpolation, and resampling are presented along with all the significant processes required before generating informative composition of images. In order to facilitate better understanding of some of the concepts introduced in this chapter, real-world examples of CT and MRI datasets in 2-D and in their 3-D correspondence are showcased to depict the significance of the mapped structures in the 2-D.