Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A dermatological assistance system for the computer-aided image analysis of nevi

: Dornheim, L.; Dornheim, J.; Kellermann, K.; Gollnick, H.; Göppner, D.; Berndt, D.; Weber, P.

Dermatology practical & conceptual 5 (2015), Nr.2, S.250
ISSN: 2160-9381
International Dermoscopy Society (World Congress) <4, 2015, Vienna>
Fraunhofer IFF ()

To support dermatologists in assessing nevi and their development over time, an image-based assistance system was developed within a close collaboration of companies as well as scientific and dermatological facilities. While the images are acquired and preprocessed using a dermatological whole body scanner, this abstract focuses on the related medical assistance system. This software system supports the dermatologist during the complete oncological diagnosis and treatment process including anamnesis, examinations, diagnosis, treatment, histological confirmation and follow-ups. Therefore, a comprehensive analysis of the clinical workflow in dermatological oncology and an analysis of technical requirements have been performed. The interactive visualizations for high-resolution body images and close-up views of individual nevi enable the user to assess suspicious lesions thoroughly multiple examinations. The dermatologist is supported by intelligent image processing and machine learning algorithms, which automatically detect, segment and evaluate the individual lesions according to dermatological criteria. The automated lesion evaluation is based on a database of image data annotated by dermatological experts and serves for highlighting lesions with suspicious characteristics. Due to the optimized acquisition conditions, the dermatological full body scanner generates spatially registered, colour-calibrated and thus standardized body images which are enriched with additional 3D information. This allows for a faithful and accurate evaluation of lesion size and colour for the first time. The automated registration of nevi from multiple follow-up images enables an objective assessment and monitoring of the specific changes over time and simplifies the user-driven evaluation. The intelligent and intuitive user interface provides several forms whose contents are filled automatically and which are optimized for fast and safe editing. Any visualization functionality is optimized for large image data. Individual reports can be generated to further support the information exchange within the clinical workflow. Together with the dermatological full body scanner, the image processing and enhancement modules, the software system represents a convenient assistance system for the computer-aided image analysis, communication and documentation of the skin and treatment state of risk patients.