dvh graph parallel statistical method based on cuda stream and shared memory

Abstract

The invention discloses a DVH (dose volume histogram) parallel statistical method based on a CUDA (compute unified device architecture) stream and a shared memory. The method includes the steps: firstly, sampling organs on a host, transmitting the position of a sampling point into a device and processing dose statistics of each organ by one stream; secondly, loading a dose matrix by the aid of a texture memory; thirdly, fetching a texture according to a position point allocated for each thread, setting a filter mode of the texture into linear interpolation, namely linearly interpolating eight picture elements of the three-dimensional texture according to distances and returning values obtained by linear interpolation; fourthly, storing statistical results by the aid of the shared memory. By developing N sub dose boxes on the shared memory, the problem of bank conflict of the shared memory is solved, and statistical speed is increased.

Description

technical field [0001] The invention relates to the field of medical data image processing, in particular to a DVH graph parallel statistical method based on CUDA stream and shared memory. Background technique [0002] Dose-volume histogram (Dose Volume Histogram, DVH) is a powerful tool to evaluate the quality of radiotherapy planning. In inverse treatment planning systems such as intensity-modulated radiotherapy, higher requirements are placed on the speed of DVH map statistics. [0003] The DVH map is a method of using two-dimensional graphs to count the three-dimensional dose distribution in radiotherapy planning, that is, to represent a region of interest such as a tumor target area or to evaluate how much volume in a key organ (organ at risk, OAR) receives a high dose of irradiation. Because it directly and effectively reflects the dose distribution of the treatment plan and the advantages and disadvantages of the plan, it has become the main basis for evaluating the...

AI Technical Summary

Problems solved by technology

First, for the contour line corresponding to each organ, a large number of if-else statements will be used to determine whether the sampling point position is within its range. This is a very simple task in the CPU, but using the judgment in the GPU statement is likely to cause serial execution of parallel threads, greatly reducing the execution performance inside the streaming multiprocessor

Second, the results of the sampling points will be saved in one hundred bins during statistics, especially for dose distributions with special rules like heavy ion radiotherapy, which will lead to a large number of write conflicts and cause computing bottlenecks

Images