This paper designs and implements an algorithm framework for the out-of-core medical data processing and analyzing and integrates it into MITK (medical imaging toolkit), an algorithm toolkit for medical image processing and analyzing accomplished by the group. With the help of this, a processing platform for the out-of-core medical data is set up and fast out-of-core volume rendering algorithms based on volume ray casting and 3D texture are studied in this paper. A semi-adaptive partitioning method is proposed to divide original data sets into sub-blocks and get a better partitioning result without influencing the partitioning speed. Furthermore, the graphics hardware is also used to accelerate the rendering process. The experimental results indicate that the new framework and algorithms are effective and efficient for the processing and visualization of the out-of-core medical data sets.