Abstract

Volumetric data from medical scans (e.g. CT, MRI) is essential for clinical analysis, medical education, and patient communication. However, achieving photorealistic rendering of this data using traditional methods often either compromises real-time rendering times or requires expensive hardware, limiting its accessibility.

This thesis proposes a novel rendering framework to address this limitation. We will investigate the use of path tracing combined with splatting methods to render medical volumetric data. Together with targeted performance optimizations, we aim to achieve photorealistic and real-time results on consumer-grade GPUs.

The efficacy of the proposed method will be evaluated through its performance and visual fidelity will be benchmarked against traditional volume rendering techniques. The diagnostic quality of the resulting visualizations will be qualitatively assessed in consultation with medical professionals.