Evaluation of the accuracy and noise response of an open-source pulse onset detection algorithm on pulsatile waveform databases

Zong's open-source algorithm ‘wabp.c’ (2003) has been widely used for onset detection of arterial blood pressure (ABP) waveforms. This code was subsequently modified by Li and Clifford (2012) to avoid possible double detections in a beat cycle. However, its performance was not systematically validated, especially on a noisy pulse database. This study aimed to evaluate its detection accuracy on both clean and noisy ABP pulse signals. Synchronously recorded ECG and ABP signals in two databases from the PhysioNet/Computing in Cardiology Challenge 2014 were used. Reference QRS positions were used as the benchmarks for pulse onset detection. Three signal quality assessment (SQA) methods, i.e., Sun's jSQI (2006), a modified jSQI (jSQI2) and Gaussian Template Matching (GTM), were performed and the onset detection results were compared with and without each SQA. For the clean set-p database, the algorithm achieved an accuracy of 99.56% without SQA and slightly enhanced its accuracy to 99.97%, 99.84% and 99.79% when using the jSQI, jSQI2 and GTM methods respectively. For the noisy set-p2 database, the algorithm achieved an accuracy of only 76.42% without SQA but significantly increased to 96.73%, 90.60% and 90.79% respectively. The jSQI2 and GTM methods exhibited a higher accuracy for assessing the ABP signal quality compared to the jSQI method. In summary, the open-source pulse onset detection algorithm was found to achieve high detection accuracy in a low noise pulsatile database while relative low detection accuracy was observed when using a relatively noisy database. Combining the algorithm with an appropriate SQA procedure significantly improved beat detection accuracy.