Research centre for imaging and non-invasive diagnostic technologies in dermatology
Prof. Dr. med Holger Hänßle
Arbeitsgruppenleiter (Research centre for imaging and non-invasive diagnostic technologies in dermatology)
Facharzt für Dermatologie und Venerologie, Phlebologie, Allergologie, Medikamentöse Tumortherapie, Ultraschalldiagnostik (DEGUM zertifiziert)
Imaging technologies, Non-invasive diagnosis, Prospective clinical studies, Licensing studies in conformity to current legislation/guidelines on medicinal products (MPG)
Human skin is well-accessible for a direct visual inspection and also for a non-invasive examination by physical and opto-acustical technologies. Increasing numbers of non-invasive strategies for the diagnosis of skin cancer but also for inflammatory skin diseases hold the promise of improving the diagnostic accuracy without the requirement of a surgical biopsy. More recently systems using ‘artificial intelligence’ algorithms for the diagnostic assessment of digital images entered the clinical arena. The aim of our research is the assessment and clinical validation of novel and innovative diagnostic technologies within the scope of current legislation/guidelines on medicinal products (MPG).
Training and validation of an automated diagnostic deep-learning-algorithm in dermoscopy for skin cancer detection (AD-LEARN DERMOSCOPY study)
Melanoma has emerged as a major challenge in public health. Novel strategies for an improved accuracy in melanoma diagnosis include the application of deep learning convolutional neural networks (CNN) for analysis of digital images of skin lesions. In this ongoing project we intend to explore the diagnostic performance of a deep learning CNN based on Google’s Inception v4 architecture in comparison to dermatologists.
In a first level we aim to investigate the performance of ‘man against machine’ in a broad set of dermoscopic images including lesions of all different anatomic body sites. In a second level we will focus on anatomic sites that show specific dermoscopic patterns in comparison to common skin (mucosa, nail apparatus, palmoplantar skin, face/scalp).
Haenssle HA, Fink C, Schneiderbauer R, Toberer F, Buhl T, Blum A, Kalloo A, Ben Hadj Hassen A, Thomas L, Enk A, Uhlmann L: Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Annals of Oncology, 2018 (epub ahead of print)
(Open access manuscript: https://doi.org/10.1093/annonc/mdy166)
Mar VJ, Soyer HP: Artificial intelligence for melanoma diagnosis: How can we deliver on the promise? Annals of Oncology, 2018 (epub ahead of print)
(Open access Editorial: https://doi.org/10.1093/annonc/mdy193)
Diagnostic accuracy of dermatofluoroscopy in cutaneous melanoma detection
Early detection is a key factor in improving survival from melanoma. The clinical diagnosis of cutaneous melanoma is currently mostly based on visual inspection and dermoscopy. However, the reported sensitivity for melanoma detection by dermoscopic examination does not exceed 80%. The technology of two-step photon excitation of melanin fluorescence (also termed “dermatofluoroscopy”) may be a suitable tool for improving the diagnostic accuracy. The aim of this ongoing project is to improve and validate the diagnostic performance of dermatofluoroscopy in the diagnosis of all melanin-bearing skin tumors.
Forschner A, Keim U, Hofmann M, Spänkuch I, Lomberg D, Weide B, Tampouri I, Eigentler T, Fink C, Garbe C, Haenssle HA: Diagnostic accuracy of dermatofluoroscopy in cutaneous melanoma detection: Results of a prospective multicentre clinical study in 476 pigmented lesions. British Journal of Dermatology, 2018 (epub ahead of print)
Fink C, Hofmann M, Jagoda A, Spaenkuch I, Forschner A, Tampouri I, Lomberg D, Leupold D, Garbe C, Haenssle HA: Study protocol for a prospective, non-controlled, multicentre clinical study to evaluate the diagnostic accuracy of a stepwise two-photon excited melanin fluorescence in pigmented lesions suspicious for melanoma (FLIMMA study). BMJ Open, 2016, 6(12):e012730
Impact of UV-irradiation on electrical impedance spectroscopy of benign nevi
lectrical impedance spectroscopy (EIS) is a non-invasive diagnostic technique for evaluation of tissue structures by applying alternating electrical current and measuring the tissue impedance. This method is based on differences in electrical impedance between benign, well-organized and malignant, chaotic tissues. The in-vivo electrical impedance of a tissue sample is shaped by tissue properties such as the formation of the intra- and extracellular environment, cell shape and size, and cell membrane structure. While the clinical and histological changes of nevi after UV-irradiation have been studied in detail, the impact of UV-irradiation on electrical impedance spectroscopy scores of nevi has not been investigated. However, for physicians it is essential to know to what extent changes in electrical impedance spectroscopy scores of nevi may be attributed to seasonal effects of UV-irradiation. The aim of this ongoing prospective, controlled, clinical study is to investigate the impact of UV-irradiation on the electrical impedance spectroscopy scores of benign nevi in 50 patients undergoing phototherapy.
Fink C, Schweizer A, Uhlmann L, Haenssle HA: Impact of UV-irradiation on electrical impedance spectroscopy of benign nevi: study protocol for a prospective, controlled, clinical study. BMJ Open, 2017, 7(11):e018730
A prospective, non-controlled, multicenter clinical study to evaluate the diagnostic accuracy of in vivo Multiphoton Laser Tomography in potentially malignant pigmented lesions
as compared to the histolopathological diagnosis (Melanoma Identification by Multiphoton Tomography [MI-MulTo])
MPTflex (JenLab GmbH, Jena, Germany), is a CE-certified non-invasive Multiphoton Laser Tomograph using a Ti:Sapphire femtosecond laser with lowgrade in situ laser power (2-50 mW). The aim of this ongoing European multicenter clinical study within the scope of the European Horizon 2020 program is to measure sensitivity, specificity, diagnostic accuracy, positive predictive value, negative predictive value, and number needed to excise for the diagnosis of melanoma by an algorithm based on the results of Multiphoton Laser Tomography.