Download data1/code/evaluation/script_evaluation_scores.m from harsharajkumar273/ReCL-Ultrasound-Dataset: direct link, hf CLI and curl.
- Browser
- Download file 3.12 kB
-
https://huggingface.co/datasets/harsharajkumar273/ReCL-Ultrasound-Dataset/resolve/main/data1/code/evaluation/script_evaluation_scores.m
- Command line
-
hf download hf://datasets/harsharajkumar273/ReCL-Ultrasound-Dataset/data1/code/evaluation/script_evaluation_scores.m
-
curl -L -o script_evaluation_scores.m https://huggingface.co/datasets/harsharajkumar273/ReCL-Ultrasound-Dataset/resolve/main/data1/code/evaluation/script_evaluation_scores.m
3.12 kB
| %-- Script to be used as an example to evaluate the quality of the reconstructed images | |
| %-- After choosing the specific configuration through acquisition_type, | |
| %-- phantom_type, data_type and flag_display parameters, this script allows computing the | |
| %-- chosen evaluation metrics. | |
| %-- By default, the computed metrics are saved in a text file in the folder named "evaluation" | |
| %-- Authors: Olivier Bernard (olivier.bernard@creatis.insa-lyon.fr) | |
| %-- Alfonso Rodriguez-Molares (alfonso.r.molares@ntnu.no) | |
| %-- $Date: 2016/03/01 $ | |
| clear all; | |
| close all; | |
| clc; | |
| addpath(genpath('../src')); | |
| %-- Parameters | |
| acquisition_type = 1; %-- 1 = simulation || 2 = experiments | |
| phantom_type = 1; %-- 1 = resolution & distorsion || 2 = contrast & speckle quality | |
| data_type = 2; %-- 1 = IQ || 2 = RF | |
| flag_display = 0; %-- 0 = do not display || 1 = display intermediate results | |
| %-- Parse parameter choices | |
| switch acquisition_type | |
| case 1 | |
| acquisition = 'simulation'; | |
| acqui = 'simu'; | |
| flag_simu = 1; | |
| case 2 | |
| acquisition = 'experiments'; | |
| acqui = 'expe'; | |
| flag_simu = 0; | |
| otherwise %-- Do deal with bad values | |
| acquisition = 'simulation'; | |
| acqui = 'simu'; | |
| flag_simu = 1; | |
| end | |
| switch phantom_type | |
| case 1 %-- evaluating resolution and distorsion | |
| phantom = 'resolution_distorsion'; | |
| case 2 %-- evaluating contrast and speckle quality | |
| phantom = 'contrast_speckle'; | |
| otherwise %-- Do deal with bad values | |
| phantom = 'resolution'; | |
| end | |
| switch data_type | |
| case 1 | |
| data = 'iq'; | |
| case 2 | |
| data = 'rf'; | |
| otherwise %-- Do deal with bad values | |
| data = 'iq'; | |
| end | |
| %-- Create path to load corresponding files | |
| path_scan = ['../../database/',acquisition,'/',phantom,'/',phantom,'_',acqui,'_scan.hdf5']; | |
| path_phantom = ['../../database/',acquisition,'/',phantom,'/',phantom,'_',acqui,'_phantom.hdf5']; | |
| path_reconstruted_img = ['../../reconstructed_image/',acquisition,'/',phantom,'/',phantom,'_',acqui,'_img_from_',data,'.hdf5']; | |
| path_output_log = ['../../evaluation/',phantom,'/',phantom,'_',acqui,'_evaluation_from_',data,'.txt']; | |
| %-- Perform evaluation for resolution | |
| disp(['Starting evaluation from ',acquisition,' for ',phantom,' using ',data,' dataset']) | |
| %-- Pass online string instances | |
| flag_simu = num2str(flag_simu); | |
| flag_display = num2str(flag_display); | |
| switch phantom_type | |
| case 1 %-- evaluating resolution and distorsion | |
| tools.exec_evaluation_resolution_distorsion(path_scan,path_phantom,path_reconstruted_img,flag_simu,flag_display,path_output_log); | |
| case 2 %-- evaluating contrast and speckle quality | |
| tools.exec_evaluation_contrast_speckle(path_scan,path_phantom,path_reconstruted_img,flag_simu,flag_display,path_output_log); | |
| otherwise %-- Do deal with bad values | |
| tools.exec_evaluation_resolution(path_scan,path_phantom,path_reconstruted_img,flag_simu,flag_display,path_output_log); | |
| end | |
| disp('Evaluation Done') | |
| disp(['Result saved in "',path_output_log,'"']) | |