07th March 2024: Dermalyser AI-driven support tool achieved remarkable sensitivity and specificity values of 95% and 85%, respectively, with these figures increasing to 100% and 93% for invasive melanomas.
Stockholm, Sweden, March 6, 2024. Results from a prospective, multicentre, clinical trial published today in the peer-reviewed British Journal of Dermatology showed that AI helps significantly improve melanoma detection rates.
The trial was designed to evaluate Dermalyser, a smartphone-based diagnostic decision support system empowered by advanced artificial intelligence (AI) and was conducted in 37 Swedish primary care facilities.
Procedure: 228 patients with skin lesions of concern when examined by their primary care physician, were recruited to participate. In case the physician was not fully sure that the lesion wasn’t a melanoma, the lesion was included and underwent the following diagnostic procedures:
1. A smartphone with a dermatoscope mounted in front of the camera was used to photograph skin lesions of concern. Within seconds, the built-in Dermalyser app produced an output – either indicating or contraindicating the presence of melanoma.
2. Regardless of Dermalyser’s output, all lesions examined in step 1 underwent further diagnostic investigation, either by surgical excision, or by referral to a dermatologist, to establish whether it was a melanoma or not. Lesion diagnoses were collected from the patients’ medical records and compared to Dermalyser’s output.
Results: In total, 253 lesions of concern were identified in 228 patients. 21 proved to be melanomas, with 11 invasive melanomas and 10 melanoma in situ (precancerous melanocytic lesions).
To gauge Dermalyser’s accuracy in identifying melanomas, an Area Under the Receiver Operating Characteristic (AUROC) curve was generated, which for invasive melanomas had a value of 0.988 (0.965–0.997), corresponding to a maximum sensitivity and specificity of 100% and 92.6%, respectively. When also including melanomas in situ, the AUROC was 0.960, corresponding to a maximum sensitivity and specificity of 95.2% and 84.5%, respectively.
Magnus Falk, Associate Professor in General Practice and Principal Investigator, Department of Health, Medicine and Caring Sciences, Linköping University, comments: “The clinical decision support tool evaluated in this investigation showed high diagnostic accuracy when used prospectively on primary care patients, which could add significant clinical value for PCPs in assessing skin lesions to detect melanoma. Additionally, a further clear impression from this trial is that the medical personnel involved in the trial were willing to put a high level of trust in Dermalyser, strongly indicating it can be easily worked into current clinical routines.”
Sam Polesie, Adjunct Senior Lecturer, Department of Dermatology and Venereology, Sahlgrenska University Hospital, Gothenburg comments: “In recent years, the precision of AI in analysing dermatoscopic images to identify melanoma has reached or even surpassed experienced dermatologists. Yet, only a limited number of studies have tested these systems under real-world conditions. This study is notable for its execution in a real clinical setting, which makes its findings especially compelling. I am enthusiastic about the results and anticipate being a part of a future where systems like Dermalyser, supported by rigorous research and in full compliance with all necessary regulations, are integrated into routine healthcare. This includes both general dermatology and specifically the diagnosis of skin cancer. The fundamental goal is, of course, to improve clinical decision-making processes for more precise and, therefore, safer diagnoses.
Christoffer Ekström, CEO of AI Medical Technology adds: “Just as in analysing mammograms for diagnosing breast cancer or neuroimaging for Alzheimer’s disease, we believe this trial has decisively proven that using AI can revolutionise melanoma diagnosis. This will save lives and also remove the anxiety and often devasting consequences of false results. The European market is still our first priority for Dermalyser, but the recently approved AI-powered device for detecting skin cancer by the FDA has also established a clear regulatory pathway for us and we are now accelerating our plans for entering the US market.”