What constitutes a useful and meaningful terahertz image?

This contribution attempts to answer the questions of how THz images are best processed, analyzed, presented, classified and interpreted, especially within a clinical setting. The solutions to these problems are reviewed within the context of recent imaging work on tumours and dental tissue, largely conducted by the Teravision group. It is concluded that compound images, especially those involving pulse broadening as an image parameter, are highly promising routes to visualisation. The possibility of using computer vision and data compression techniques is also discussed.