Skip to content

Files

Latest commit

ef34606 · Feb 22, 2023

History

History

C301

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Feb 22, 2023
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
Dec 22, 2022
1. Dataset:     c301_raw_data

     c301_raw_data: raw_data for EEG signal and label for data label
     Each patient has 20 channels, including 19 EEG and 1 ECG.

     ID Seizure events (Tmin - Tmax) (s) Total Seizure time (s) Total seizure-free time (min)
     1        26       (7.4 - 84.4)             980.0                      34.5
     2        53       (2.0 - 21.8)             300.8                      45.8
     3        20       (12.1 - 45.3)            499.8                      42.5
     4        16       (1.6 - 45.3)             89.4                       49.3
     5        10       (1.6 - 8.3)              45.3                       50.1
     6        6        (5.1 - 14.1)             55.5                       49.9
     7        1        (7.8 - 7.8)              7.8                        50.7
     8        12       (3.5 - 8.6)              78.3                       49.5
     9        5        (3.1 - 18.6)             51.4                       50.0
     10       17       (1.2 - 12.5)             129.2                      48.7
     11       10       (2.3 - 11.3)             80.8                       49.5
     12       8        (2.0 - 7.2)              44.3                       50.1
     13       6        (2.7 - 9.4)              34.4                       50.2
     14       16       (2.3 - 42.1)             179.8                      47.8
     15       20       (2.3 - 13.7)             133.5                      48.6
     16       20       (2.3 - 11.7)             133.5                      48.6
     17       12       (2.0 - 15.6)             95.4                       49.2
     18       8        (1.6 - 7.4)              41.1                       50.1
     19       12       (8.6 - 22.5)             206.4                      47.4

     Sum      258                               3186.6                     912.4

     	We combine all 19 patients from the c301 datasets into seizure, non-seizure-1, and non-seziure-2 by choosing the more clean dataset with 4s windows at
     	the start of seizure/non-seizure by sliding_window.m.             (non-seizure, seizure) = (9992, 757)
	The read_data.m can be used to read and analyze the C301 raw data.
	*We have not uploaded the entire C301 raw data because of space limitation on github. 
	

2. Code:

      Data_preprocessing.py:	c301 datasets are processed by Discrete Wavelet Transform.

                DBM_train.m: 	Please use the DBM model to train C301 data for 2 to 10-dimensional outputs.
                              	Each layer of training is performed in RBM1--RBM2--RBM3--RBM4 ----> *.m, and one can get
                              	the result of two states['transient', 'converged']
                              	All of the results shall be saved under "Different_dimension".

          Contrast_model.py: 	The code in this part includes four training models: ['KPCA', 'Isomap', 't-SNE', 'UMAP'], 
				KPCA, Isomap, and UMAP can generate dimensions from 2 to 10, and t-SNE can generate dimensions from 2 to 3.
                              	All of the results shall be saved under "Different_dimension".

SVM+Fisher_Discriminiant.py: 	SVM and Fisher discriminant are used to evaluate results from Different_dimensions, and all
                             	of the results shall be saved under "Evaluation_results."

        2D_visualization.py: 	The 2D data can be used to visualize the output results for all 6 methods.
                      

                measure_bar: 	In order to evaluate the performance of different dimensions, bar plots are used, 
				which shall save the results in "Evaluation_results/Result_bar".