Ouriel Barzilay, PHD – CTO
Received his BSc and MSc with honors from the Technion, and his PhD research on 3D Geometry and Biorobotics with Profs. Alon Wolf and Lihi Zelnik-Manor was awarded twice
 
Over the last decade, AI and specifically Deep Neural Networks (DNNs) have revolutionized the industry and the academy in numerous fields. More recently, DNNs are being introduced into more and more safety-critical applications such as autonomous driving, limb rehabilitation or telesurgery. This is accompanied by a quickly evolving acceptance of Deep Learning as what it is: the means to provide results better than any other way, provided high quality data and a decent network architecture. In 2020, regulatory agencies have rapidly adapted in the effort to face COVID-19 and several AI-based products regulation processes have been accelerated. 
Today, skin-related conditions are among the most prevalent reasons for visits to primary care medical centers, and most of them are non-critical cases, suitable for remote treatment. Asynchronous teledermatology includes the collection of images and additional information from the patient's description of his skin condition, and the registration of diagnoses provided by board-certified dermatologists. This constitutes an invaluable and unique dataset for disease classification with machine and deep learning. A learning system can be trained on this dataset to specialize in differential diagnosis, in triage (critical/non-critical diseases), or on the classification of any subset of diseases in the dataset. Such expert-labeled data can be leveraged to produce AI-based tools for physicians towards better diagnoses and treatment plans.
Numerous applications related to skin health and appearance could benefit from deep learning tools trained on this kind of dataset, including but not limited to, management of drug dermatological side effects, wound management, estimation of various skin properties such as skin dryness or skin tones for cosmetics and fashion.