Skip to content

Latest commit

 

History

History
99 lines (79 loc) · 9.88 KB

Beat_Classification.md

File metadata and controls

99 lines (79 loc) · 9.88 KB
Raw

Beat Classification

This project includes an algorithm for automatic beat classification on ECG signals, described in the following paper:

  • T. Teijeiro, P. Félix, J.Presedo and D. Castro: Heartbeat classification using abstract features from the abductive interpretation of the ECG, IEEE journal of biomedical and health informatics, 2018, vol. 22, no 2, p. 409-420. DOI: 10.1109/JBHI.2016.2631247 .

The algorithm relies on the abductive interpretation of an ECG record to obtain a set of qualitative morphological and rhythm features for each QRS observation in the interpretation result. Then, a clustering task provides a partition of the full set of QRS observations, and finally a label is assigned to each cluster, classifying all the beats in the record.

The classification stage is implemented in the beat_classification.py script, that receives as input a record in MIT-BIH format, a set of annotations in the MIT format with the interpretation results, and the clustering results, consisting of a simple tag assignment to each beat annotation in the interpretation results. As output, this script generates a copy of the input annotations file, with all beats labeled.

Files required for the classification

In short, for the automatic classification of heartbeats using this algorithm, it is required:

  1. An ECG record in the MIT-BIH format (Typically, a .hea file and a .dat file).
  2. A set of annotations resulting from the abductive interpretation of the record (The .iatr files included in the classification_data/ directory).
  3. The results of the clustering task on the beat annotations (The .cluster files included in the classification_data/ directory)

Since the abductive interpretation stage is a very time-consuming task, the annotation files with the results for all records in the MIT-BIH Arrhythmia Database are provided in the classification_data/ directory with the .iatr extension. These results can be obtained with the construe_ecg.py script, taking as initial evidence the .atr reference annotations in the database. Note that the interpretation results depend on the computing capabilities, so they may be slightly different from those provided in this repository.

With respect to the clustering results, unfortunately at this time we can not provide access to the implementation of the clustering algorithm. This algorithm is described in the following paper:

  • D. Castro, P. Félix, and J. Presedo: A Method for Context-Based Adaptive QRS Clustering in Real Time, IEEE Journal of Biomedical and Health Informatics, vol. 19, no. 5, pp. 1660–1671, Sept 2015.

We have also included the clustering results for all records in the MIT-BIH Arrhythmia Database in the classification_data/ directory, in a set of files with the .cluster extension. These files have a simple plain text format, with two columns of integer numbers. The first column contains the index of each beat within the interpretation annotation file, in 0-based indexing, and the second column contains the identifier of the cluster the beat belongs to.

Using a customized clustering algorithm

If you want to use a different clustering strategy before the classification stage, you have to follow these steps:

  1. Get the annotation file resulting from the abductive interpretation of the record (or the already provided .iatr file).
  2. Apply your clustering algorithm to all annotations whose NUM field is one of the following: (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 16, 25, 30, 34, 35, 38, 41).
  3. Generate a cluster file with the same format described in the previous section.

Reproducing the results in the paper

To reproduce the classification results shown in the paper, you have to follow these steps:

  1. Download the MIT-BIH Arrhythmia Database files to your hard disk (for example, in the /tmp/mitdb/ directory).
  2. Copy the .iatr and .cluster files into the same directory.
  3. For all records to be tested, execute the beat_classification.py script: python beat_classification.py -r /tmp/mitdb/$REC -a iatr -c cluster -o cls. (Here $REC stands for the name of the record, and .cls is the selected extension for the annotations file with the classification results).
  4. For all records to be tested, compare the classification results with the manual reference using the bxb application: bxb -r $REC -a atr cls -f 0 -t 0 -L file1 file2 2> /dev/null.
  5. Get the aggregated statistics using the sumstats application: sumstats file1 > mitdb_validation.

Finally, in the mitdb_validation you will find the full statistics of the classification results. Following we show the results for the full database:

Record Nn' Sn' Vn' Fn' On' Ns Ss Vs Fs' Os' Nv Sv Vv Fv' Ov' No' So' Vo' Fo' Q Se Q +P V Se V +P S Se S +P RR err
100 2239 0 0 0 0 0 33 0 0 0 0 0 1 0 0 0 0 0 0 100.00 100.00 100.00 100.00 100.00 100.00 2.23
101 1858 1 0 1 0 2 2 0 0 0 0 0 0 1 0 0 0 0 0 100.00 100.00 - - 66.67 50.00 3.55
102 99 0 0 2084 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 100.00 100.00 100.00 100.00 - - 17.32
103 2082 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 100.00 100.00 - - 50.00 100.00 2.64
104 163 0 1 2063 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 99.96 100.00 0.00 - - - 29.69
105 2513 0 5 4 1 0 0 0 0 0 2 0 35 0 0 11 0 1 1 99.49 99.96 85.37 94.59 - - 24.18
106 1493 0 17 0 0 0 0 0 0 0 14 0 490 0 0 0 0 13 0 99.36 100.00 94.23 97.22 - - 109.08
107 0 0 3 2054 0 0 0 0 0 0 0 0 56 23 0 0 0 0 1 99.95 100.00 94.92 100.00 - - 15.27
108 1685 1 0 2 0 2 4 5 0 0 48 0 11 0 0 4 0 1 0 99.72 100.00 64.71 18.64 80.00 36.36 80.13
109 2491 0 8 2 0 0 0 2 0 0 1 0 28 0 0 0 0 0 0 100.00 100.00 73.68 96.55 - 0.00 6.57
111 2122 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 100.00 100.00 100.00 50.00 - - 9.67
112 2536 0 0 0 0 1 2 0 0 0 0 0 0 0 0 0 0 0 0 100.00 100.00 - - 100.00 66.67 5.01
113 1788 0 0 0 0 0 0 0 0 0 1 6 0 0 0 0 0 0 0 100.00 100.00 - 0.00 0.00 - 5.70
114 1814 3 0 0 0 0 9 0 0 0 4 0 43 4 0 2 0 0 0 99.89 100.00 100.00 91.49 75.00 100.00 45.60
115 1951 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 100.00 100.00 - 0.00 - - 2.75
116 2299 0 0 0 0 0 0 7 0 0 3 1 102 0 0 0 0 0 0 100.00 100.00 93.58 96.23 0.00 0.00 4.75
117 1534 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 100.00 100.00 - - 100.00 100.00 7.88
118 2164 2 0 0 0 1 94 7 0 0 1 0 9 0 0 0 0 0 0 100.00 100.00 56.25 90.00 97.92 92.16 6.48
119 1540 0 0 0 0 0 0 0 0 0 0 0 442 0 0 3 0 2 0 99.75 100.00 99.55 100.00 - - 101.90
121 1861 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 100.00 100.00 100.00 100.00 0.00 - 2.96
122 2475 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 99.96 100.00 - - - - 24.00
123 1515 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 100.00 100.00 100.00 100.00 - - 2.78
124 1526 34 0 3 0 3 2 0 0 0 2 0 46 2 0 0 0 1 0 99.94 100.00 97.87 95.83 5.56 40.00 45.33
200 1688 9 20 1 0 23 21 31 0 0 32 0 773 1 0 0 0 2 0 99.92 100.00 93.58 96.02 70.00 28.00 25.09
201 1612 40 0 0 0 2 76 0 0 0 1 20 197 2 0 10 2 1 0 99.34 100.00 99.49 90.37 55.07 97.44 152.12
202 2047 22 0 1 0 1 28 1 0 0 0 2 17 0 0 13 3 1 0 99.20 100.00 89.47 89.47 50.91 93.33 147.06
203 2278 1 44 4 1 35 0 4 0 1 207 1 384 1 2 9 0 12 0 99.30 99.87 86.49 64.65 0.00 0.00 84.47
205 2569 0 0 10 0 2 2 0 0 0 0 0 70 1 0 0 1 1 0 99.92 100.00 98.59 100.00 66.67 50.00 19.54
207 1446 1 7 0 0 5 106 2 0 0 90 0 201 0 0 2 0 0 0 99.89 100.00 95.71 69.07 99.07 93.81 14.32
208 1570 2 30 348 0 0 0 0 0 0 16 0 959 26 0 0 0 3 1 99.86 100.00 96.67 98.36 0.00 - 33.80
209 2602 20 0 0 0 18 362 0 0 0 1 0 1 0 0 0 1 0 0 99.97 100.00 100.00 50.00 94.52 95.26 13.26
210 2316 0 10 2 1 102 19 14 2 0 5 3 168 6 0 0 0 3 0 99.89 99.96 86.15 95.45 86.36 13.87 25.46
212 2748 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100.00 100.00 - - - - 5.93
213 2634 3 24 333 0 7 25 6 3 0 0 0 189 26 0 0 0 1 0 99.97 100.00 85.91 100.00 89.29 60.98 20.09
214 2003 0 2 3 0 0 0 0 0 0 0 0 254 0 0 0 0 0 0 100.00 100.00 99.22 100.00 - - 8.99
215 3192 0 26 1 0 0 0 0 0 0 3 3 138 0 0 0 0 0 0 100.00 100.00 84.15 95.83 0.00 - 6.34
217 239 0 9 1800 0 0 0 0 0 0 5 0 142 1 0 0 0 11 1 99.46 100.00 87.65 96.60 - - 79.05
219 2040 7 4 0 0 2 0 1 0 0 1 0 58 1 0 39 0 1 0 98.14 100.00 90.62 98.31 0.00 0.00 244.08
220 1947 2 0 0 0 5 92 0 0 0 0 0 0 0 0 2 0 0 0 99.90 100.00 - - 97.87 94.85 59.27
221 2030 0 5 0 0 0 0 0 0 0 0 0 371 0 0 1 0 20 0 99.13 100.00 93.69 100.00 - - 109.40
222 1921 141 0 0 0 134 275 0 0 0 0 0 0 0 0 7 5 0 0 99.52 100.00 - - 65.32 67.24 118.49
223 2027 24 110 12 0 2 37 25 0 0 0 27 336 2 0 0 1 2 0 99.88 100.00 71.04 92.56 41.57 57.81 52.68
228 1684 1 1 0 0 4 2 10 0 0 0 0 351 0 0 0 0 0 0 100.00 100.00 96.96 100.00 66.67 12.50 20.04
230 2255 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 100.00 100.00 100.00 100.00 - - 3.87
231 1568 1 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 100.00 100.00 100.00 100.00 0.00 - 5.37
232 389 5 0 0 0 2 1333 0 0 0 4 43 0 0 0 2 2 0 0 99.78 100.00 - 0.00 96.38 99.85 92.22
233 2216 0 1 3 0 1 7 1 0 0 1 0 821 8 0 12 0 8 0 99.35 100.00 98.80 99.88 100.00 77.78 66.92
234 2699 8 0 0 0 1 42 0 0 0 0 0 3 0 0 0 0 0 0 100.00 100.00 100.00 100.00 84.00 97.67 1.93
Sum 89468 330 327 8731 3 355 2575 116 5 1 445 106 6708 106 2 118 15 85 4
Gross 99.80 99.99 92.70 92.38 85.10 84.51