On-Chip Fuzzy Logic Synthesis of a New Ischemic and Non-Ischemic Heartbeat Classifier

This brief presents a new ischemic and non-ischemic heartbeat classifier, which is highly synthesizable for on-chip applications. The classifier consists of an algorithm analyzing the electrocardiogram (ECG) shape to distinguish ischemia. It emulates medical expertise in cardiac ischemia diagnosis using a digital fuzzy logic system. The proposed fuzzy sets obtain information to classify ischemic and non-ischemic heartbeats, helping in discarding false positives. The work presents the physical implementation and experimental results on-chip in a CMOS 180nm process. The algorithm tests 20 electrocardiogram recordings from the Long-Term ST database in Physionet, verifying the method. The experimental results, with an 8-bit resolution, exhibit a sensitivity (Se) of 96.43 percent and specificity (Sp) of 96.88 percent with a (300× 325) squared microns digital circuit. The implementation of the algorithm, with 14-bit resolution, shows a performance increase in (Se) of 99.03 percent and specificity (Sp) of 99.08 percent, solely using two ECG descriptors.