Dermoscopy training combined with AI significantly improves skin cancer detection

Empowering general practitioners through dermoscopy training augmented with artificial intelligence (AI) enhances skin cancer screening accuracy and lesion characterisation, boosting their confidence. It promotes early detection, enabling prompt therapeutic strategies contributing to improved skin cancer survival rates.

The scarcity of adequate health professionals and the strain on primary care systems necessitate general practitioners to engage in skin cancer screening. However, their limited dermoscopy skills highlight the need for essential training. "We evaluated the impact of training sessions on the diagnostic accuracy and therapeutic strategies proposed by general practitioners in the treatment of skin cancer," explained Dr Christian Dorado Cortez (University Hospital of Saint-Etienne, France) [1].

The study consisted of 2 training sessions: the first employing the dermoscopy 2-step algorithm and the second incorporating the 2-step algorithm augmented by AI software. The 2-step algorithm consists of 2 levels. First, the observer must decide whether a lesion is of melanocytic or non-melanocytic origin. When the lesion is identified to be of melanocytic origin, step 2 entails the identification of the melanocytic lesion as benign, suspect, or malignant [2].

The participants were general practitioners from 2 universities, who completed 3 online tests before and after each training session. Additionally, they underwent a final evaluation a month after their second training session. Each online test comprised 51 cases, assessing diagnostic accuracy, lesion characterisation, the capability to recommend therapeutic strategies, and their confidence level for each case.

The mean score of the initial test, conducted before any training, was 4.71 out of 51 (n=41). Subsequently, significant improvements in the second test were seen to 6.72 (following the first training session) and 25.6 (following the second training session; P<0.001). The results from the fourth test demonstrated a notably persistent improvement in the average score, particularly when utilising AI analysis (mean score 27.18 out of 51).

“The study results confirmed that dermoscopy training combined with AI considerably improves the diagnostic accuracy, the ability of general practitioners to determine the nature of the lesion, and the management of pigmented lesions,” Dr Dorado Cortez concluded.

"We strongly encourage universities to offer dermoscopy training to all general practitioners," he emphasised. Such training initiatives could substantially enhance skin cancer screening in primary care, ultimately improving overall skin cancer survival rates.

1. 
   
   1. Dorado Cortez C, et al. Assessment of the impact of a dermoscopy training associated with an artificial intelligence on general practitioner residents. P1417, EADV Congress 2023, 11–14 October, Berlin, Germany.
   2. Argenziano G, et al. J Am Acad Dermatol 2003;48:679–693.

Copyright ©2023 Medicom Medical Publishers



Posted on 28 November 202328 November 2023