CASPIAN JOURNAL

MANAGEMENT AND HIGH TECHNOLOGIES

Constructing 3D models of human internal organs using supercomputer calculations

Read Kurochkin M. A., Kasimov E. M., Purii A. A., Vostrov A. V. Constructing 3D models of human internal organs using supercomputer calculations // Caspian journal : management and high technologies. — 2018. — №4. — pp. 72-80.

Kurochkin M. A. - Cand. Sci. (Engineering), Peter the Great St. Petersburg Polytechnic University, 29 Politekhnicheskaya St., Saint Petersburg, 195251, Russian Federation, kurochkin.m@gmail.com

Kasimov E. M. - student, Peter the Great St. Petersburg Polytechnic University, 29 Politekhnicheskaya St., Saint Petersburg, 195251, Russian Federation, kasimov.ernest@gmail.com

Purii A. A. - student, Peter the Great St. Petersburg Polytechnic University, 29 Politekhnicheskaya St., Saint Petersburg, 195251, Russian Federation, zloypur@gmail.com

Vostrov A. V. - Senior Lecturer, Peter the Great St. Petersburg Polytechnic University, 29 Politekhnicheskaya St., Saint Petersburg, 195251, Russian Federation, alex.sinkriver@gmail.com

In today’s modern medicine, computer tomography is a multipurpose device for diagnosing the condition of human organism. The principles of its work help to produce three-dimensional final images, that consist of separate slices. When we receive the images, they need to be visualized for further application in patient treatment. The article touches upon the parallelization questions of visualization tasks CT data with the help of chest 3D model example. In order to solve the task the marching cubes algorithm is used. It allows to use not only sequential realization, but also the parallel one. The former is performed using CPU, however the latter uses computing on GPU with Nvidia architecture and CUDA technology. A comparative analysis of execution time and different parameters (such as realization method (CPU-GPU), the quantity of processed data on GPU, kernel parameters settings) is carried out. The visualization of chest 3D models, that are constructed in different quantity of initial nodes, is done as a result.

Key words: компьютерная томография, CUDA, параллельные вычисления, 3D-визуализация данных, формат DICOM, GPU, CPU, шкала Хаунсфилда, computer tomography, CUDA, parallel computing, 3D data visualization, DICOM, GPU, CPU, Hounsfield scale