Detection of computationally-intensive functions in a medical image segmentation algorithm based on an active contour model

Abstract

Image segmentation is one of the most critical operations performed on medical images. These operations require developing optimisation strategies to reduce runtime. Profiling methods can assess algorithm's performance concerning the overall cost of runtime, memory access, and performance bottlenecks. Therefore, we propose an approach for detecting computationally intensive functions in a competent medical image segmentation algorithm based on an active contour model. Our approach applies performance analysis tools commonly available in traditional computer operating systems, requiring no new setup or developing new performance-measuring techniques. The overall cost of execution time, memory accesses, and performance bottlenecks are measured in execution time. In conclusion, a call graph visualisation can suggest to users a quick graphical overview of the execution time of their codes and, therefore, guarantee the shortest possible learning curve by the community of researchers from medical image processing and analysis.