Add citations

README.md

@@ -68,10 +68,10 @@ We formulate a supervised multi-label image classification task on 12-lead ECGs
 
 Training data combines:
 
-- PTB-XL, a large-scale dataset of raw 12-lead ECG waveforms in WFDB format with 16-bit precision
-- A supplementary ECG image dataset
+- PTB-XL, a large-scale dataset of raw 12-lead ECG waveforms in WFDB format with 16-bit precision [1]
+- A supplementary ECG image dataset [2]
 
-To enable image-based classification, raw PTB-XL waveforms are converted into realistic print-style ECG images using the open-source ECG image generator by Rahimi et al. This yields approximately 21,000 synthetic ECG images, which are combined with 713 real ECG images from the supplementary dataset.
+To enable image-based classification, raw PTB-XL waveforms are converted into realistic print-style ECG images using the open-source ECG image generator by Rahimi et al [3]. This yields approximately 21,000 synthetic ECG images, which are combined with 713 real ECG images from the supplementary dataset.
 
 ---