P889Validation of a smartphone-camera based software for the identification of electrodes location on human chest

Abstract

Background: The global diffusion of smartphone devices is producing profound changes in diagnostics, as much relevant data could now be generated locally by the patient rather than centrally by providers. Although the multiple-lead electrocardiogram (ECG) is a cost-effective, valuable and non-invasive test, its use requires trained healthcare workers to correctly record the exam. This is mainly due to the specific position that electrodes need to have on the chest of the patient. Purpose: To validate 'Mobile ECG on other', a smartphone App algorithm that guides electrode self-placement via imaging processing technology by identifying the electrode theoretical location on the patient's own chest, once framed within the smartphone camera. The current software, aimed at guiding users with no medical background to the correct placement of electrodes, is designed to be coupled with DHeartVR smartphone electrocardiograph, in order to enable patient's multiple-lead ECG self-acquisition. Methods: Consecutive healthy volunteers (n=40, men=30) were enrolled from June to July 2017. All patients were > 18 years old and signed informed consent for study participation. Two uncovered chest images were acquired for each patient. The first image was processed and virtual electrodes automatically placed by D-HeartVR App with 'Mobile ECG on other' software. The second image was stored in a database and virtual electrodes were placed on the chest picture by a trained nurse, blinded to the 'Mobile ECG on other' results. The two chest images were then compared for accuracy assessment of the 'Mobile ECG on other' software for electrode placement, assuming the nurse placement as the 'gold standard'. For peripheral electrodes (RA, LA, LL) an error within 6 3 cm was considered acceptable, whereas for precordial electrodes (V2 and V5) an error within 6 2 cm was satisfactory. Results: Median placement error for 'Mobile ECG on other' software was: 0.3 [0.1-1] cm for RA on vertical axis, 0.5 [0.3-1] cm for RA on horizontal axis 0.5 [0.2-1] cm for LA on vertical axis, 0.3 [0.2-1] cm for LA on horizontal axis 1 [0.5-1.2] cm for LL on vertical axis, 0.6 [0.5-1] cm for LL on horizontal axis 0.5 [0.1-1.1] cm for V2 on vertical axis, 0.5 [0.1-0.9] cm for V2 on horizontal axis 0.5 [0.1-1] cm for V5 on vertical axis, 0.5 [0.2-1] cm for V5 on horizontal axis. Six (15%) patients had at least 1 electrode misplaced: 1 isolated V5 misplacement, 1 isolated V2 misplacement, 2 combined V2 and V5 misplacement, 1 combined V2 and RA displacement and 1 case with V2, V5 and RA misplacement. Conclusions: 'Mobile ECG on other' software proved reliable and accurate for correct electrode self-placement, thereby opening new perspectives for accurate patient generated remote diagnostic tests.