Comparison of bio-inspired algorithms for medical image segmentation using Tsallis entropy on the BRATS dataset (years 2017 and 2018) using MATLAB.

Prof. Guilherme A. Wachs Lopes, Ricardo M. Santos, Prof.Nilson T. Saito and Prof. Paulo S. Rodrigues

The main file for this experiment is "start.m". It is responsible for iterating through each patient in the dataset and each algorithm. The algorithms are executed through the "execute_algorithm.m" file. Also, each algorithm may be executed independently for any kind of image. Follow the instructions in the header of each algorithm file to get to know better about the parameters considered. A description of each algorithm and their respective parameters is suplied below.

All algorithms expect four parameters as input:

The image to be segmented: I

Number of segmentation thresholds: thresholds

Number of generations: generations

Parameter struct: parameters

The names of these parameters may vary for each algorithm. The parameters struct contains dinamic parameters which vary for each algorithm. In this study, we considered the following algorithms and their respective parameters:

Population size (pop_size) = 40;

Probability of host bird discovering the cucoo egg (pa) = 0.5;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.

Population size (pop_size) = 30;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.

Population size (pop_size) = 40;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.

Population size (pop_size) = 100;

Number of clans (numClan) = 5;

Elitism (Keep) = 2;

Alpha (alpha) = 0.5;

Beta (beta) = 0.1;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.

Population size (pop_size) = 30;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.

Population size (pop_size) = 30;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.

Population size (pop_size) = 50;

Upper Bound for image thresholding (UB) = 253;

Lower Bound for image thresholding (LB) = 2.