World Congress of Perinatal Medicine

SPEAKERS ABSTRACTS The application of super slow review of ultrasound clips to a diagnosis of fetal tachyarrhythmia Name and surname: Jun Yoshimatsu Institution: National Cerebral and Cardiovascular Center Email: yoshimatsu.jun.hp@ncvc.go.jp Objective Fetal arrhythmias occur in 1–2% of all pregnant cases. Premature atrial contraction (PAC), premature ventricular contraction (PVC) which are clinically benign are relatively common and do not require therapy. In contrast supraventricular tachycardias (SVTs), atrial flutter (AFL), ventricular tachycardia which are life-threatening are relatively rare but require treatment. These tachyarrhythmias easily complicate fetal congestive heart failure which can be followed by a fetal death. Thus, prenatal treatment for SVT and AFL is warranted to convert to sinus rhythm. Recently our group demonstrated that protocol-defined transplacental treatment for fetal SVT and AFL was effective and tolerable in 90% of cases. In order to choose the appropriate medication for fetal tachyarrhythmia the accurate diagnosis is important. Because fetal electrocardiography is still not available fetal echocardiography is commonly used for the diagnosis of fetal arrhythmia. The M-mode and Doppler echocardiography may achieve prenatal identification of arrhythmic types. The easier method helps more accurate and precise diagnosis of it. Recent advances of ultrasound examination technology allows a high spatial resolution and high temporal resolution at the same time. The image editing of video clips of four chamber view, which is the most easier image plane, is applicable for the diagnosis of fetal arrhythmias. We conducted the research to innovate the useful video clip editing method including a super slow motion for the precise diagnosis of fetal arrhythmias. Method From our fetal echocardiography database, 55 fetal echocardiography video clips (35 AFL, 20 SVT) were reviewed. Four chamber view clips of each case were selected. Every clips were edited into 10% speed. One examiner inserted the clip maker at the timing of atrial contraction and ventricular contraction. The intervals between ventricular contraction and atrial contraction were measured. Also the number of times of contraction of atrium and ventricle and their relations were recorded. Results Every contractions of each portion (atrium and ventricle) could be detected easily with edited video clips. The differentiation between AFL and SVT could be done completely by counting the number of contractions of atrium and ventricle. Also the V-A intervals and the A-V intervals could be measured in one hundredth seconds. In all cases, the diagnosis is compatible with that had made postnatally. Conclusion The first step of this study was depended on the examiner’s naked eyes helped by editing application. The accuracy of each steps of this study were satisfactory. This is the preliminary step to conduct the deep learning of Artificial Intelligence (AI). The most important point of using AI is setting up what to teach and how to teach. Our study showed the possibility of the application of AI to make a diagnosis of fetal tachyarrhythmia.