Various medical imaging modalities exist, with applications to different clinical needs such as oncology, perfusion studies, angiography, musculoskeletal diagnosis, and others. This paper by Yin et al. describes the current state of the art of two relatively new imaging modalities that support time-series analysis: terahertz (TZz) pulse imaging (TPI) and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The goal of the paper is to give a comprehensive description of each imaging technique and to explore current methods for analysis of the imagery each produces, with special attention to time-based methods.
Simply stated, THz imaging uses the portion of the electromagnetic (EM) spectrum between microwaves and infrared, which has unique properties in tissue response, acceptable resolution, and low biological impact. Recent developments in sensing and post-processing make this band usable for medical imaging. DCE-MRI is based on the introduction of contrast agents that affect the rate of magnetic relaxation, so that image differences in time correspond to varying blood perfusion. Both techniques offer many variables that can be used to design and tune imaging strategies to particular clinical needs. Many of these new approaches are discussed briefly and referred to in the impressively complete list of references.
However, the need exists for processing approaches to maximize the insights from this data. The authors discuss general considerations for typical problems such as preprocessing for registration and denoising, feature extraction, and classification (what tissue type is this, for example). They also explore the use of extreme learning machines (ELMs) and present some results. The length of the paper prevents a thorough discussion of their work in this area, but references are included to more detailed publications on that aspect.
This paper is a good snapshot of current technology and a suitable entry point into the research literature for both THz and DCE-MRI data, and the analysis of such data. One of the key insights that the authors demonstrate is the commonality between processing THz imaging and DCE-MRI data. This suggests the consideration of early versus late fusion for the combining of these two data sources in clinical use. They also correctly raise the issue of slow acquisition times, which further advances in electronics and sensing will likely improve but currently is a challenge from a clinical adoption standpoint.