Periodic Reporting for period 1 - SWAN (SWAN: Blitzscaling the early detection of skin cancer)
Reporting period: 2022-07-01 to 2023-06-30
Swan represents a groundbreaking solution that addresses the persistent challenge of early detection of skin cancer. What sets Swan apart is its ability to perform comprehensive full-body scans with dermoscopic resolution, all completed in less than three minutes. This breakthrough technology delivers substantial benefits to a wide range of stakeholders, including patients, physicians, hospitals, and payers.
Combining both hardware and software components, Swan is situated at the intersection of cutting-edge, patented technologies in computer vision, advanced robotic control, image processing, and AI. The core objective of this project is to enable a robot, equipped with a proprietary imaging system, to automatically navigate around a patient's body and capture highly detailed images in a matter of minutes. These images are then utilized to generate the world's first dermoscopic skin maps, which are subsequently reviewed by physicians.
The scope of work involved in this project encompasses the development of software algorithms across various domains. Firstly, sophisticated robotic path planning and control software is devised to ensure real-time navigation around the patient's body within a short timeframe. Secondly, the imaging pipeline is optimized to expedite the generation of comprehensive skin maps, akin to the familiar 'Google' maps, in under a minute. Furthermore, an AI-based lesion detection software is created to identify lesions, with a particular focus on recognizing new and changed lesions. This aspect is crucial, as it is estimated that 80% of melanomas originate from new lesions rather than pre-existing moles.
Additionally, a groundbreaking AI classification software is engineered, capitalizing on the unique insights extracted from these full-body scans. Leveraging the power of AI, this software enables accurate categorization and analysis of skin conditions, providing valuable diagnostic support to healthcare professionals. Moreover, considerable effort is dedicated to designing a delightful experience for both patients and physicians. This involves developing an intuitive graphical user interface (GUI) for the scanning device itself, as well as a GDPR/HIPAA-compliant web application to facilitate scan reviews by physicians.
The scope of work involved in this project encompasses the development of software algorithms across various domains. Firstly, sophisticated robotic path planning and control software is devised to ensure real-time navigation around the patient's body within a short timeframe. Secondly, the imaging pipeline is optimized to expedite the generation of comprehensive skin maps, akin to the familiar 'Google' maps, in under a minute. Furthermore, an AI-based lesion detection software is created to identify lesions, with a particular focus on recognizing new and changed lesions. This aspect is crucial, as it is estimated that 80% of melanomas originate from new lesions rather than pre-existing moles.
Additionally, a groundbreaking AI classification software is engineered, capitalizing on the unique insights extracted from these full-body scans. Leveraging the power of AI, this software enables accurate categorization and analysis of skin conditions, providing valuable diagnostic support to healthcare professionals. Moreover, considerable effort is dedicated to designing a delightful experience for both patients and physicians. This involves developing an intuitive graphical user interface (GUI) for the scanning device itself, as well as a GDPR/HIPAA-compliant web application to facilitate scan reviews by physicians.
The R&D works have yielded exceptional results in terms of generation of the world’s first dermoscopic-skin maps as well as detection of new / modified lesions. Further optimization and refinement are necessary before the solution can be deployed. Once fully realized, Swan has the potential to set a new standard of care in dermatology.