Clinical validation study of a CAD system with artificial intelligence algorithms for early non-invasive detection of in vivo cutaneous melanoma. (LEGIT_MC_EVCDAO_2019)

Study type

• Study type:
  Non-interventional study

Non-interventional study

• Non-interventional study design:
  Cohort

  Cross-sectional

Study drug and medical condition

• Additional medical condition(s):
  Any kind of melanoma

Population studied

• Short description of the study population:
  Patients with skin lesions with suspected malignancy seen at the Dermatology Department of the Hospital Universitario Cruces and Hospital Universitario Basurto.
• Age groups:
  Adults (18 to < 46 years)

  Adults (46 to < 65 years)

  Adults (65 to < 75 years)

  Adults (75 to < 85 years)

  Adults (85 years and over)
• Estimated number of subjects:
  200

Study design details

• Study design:
  The goal of this Cross-sectional analytical observational study of clinical case series is to validate a Computer-aided diagnosis software developed by AI Labs Group for the identification of cutaneous melanoma in images of lesions taken with a dermatoscopic camera.
• Main study objective:
  This study aims to validate our CAD system's capability, utilizing machine vision, for the early and non-invasive in-vivo diagnosis of cutaneous melanoma. The primary objective is to confirm that the AI algorithm developed for identifying cutaneous melanoma in dermoscopic images achieves: AUC > 0.8 Sensitivity ≥ 80% Specificity ≥ 70%
• Outcomes:
  Skin Lesion Recognition: Top-1 Image-Level Accuracy: 52.86% (initial) / 48.25% (initial + extension). Top-3 Image-Level Accuracy: 81.20% (initial) / 74.74% (initial + extension). Top-5 Image-Level Accuracy: 88.83% (initial) / 83.16% (initial + extension). Malignancy Prediction: AUC for Malignancy: 87.28% (initial) / 88.26% (initial + extension). Adverse Events and Adverse Reactions to the Product: No adverse events or reactions related to the investigated device were observed. Product Deficiencies: No deficiencies in the device were observed, it demonstrated consistent performance. Subgroup Analysis for Special Populations: No special population subgroups were identified in the context of the analyzed pathologies.
• Data analysis plan:
  For the purpose of estimating the device's performance, we use different metrics depending on the task: Melanoma detection: Top-K precision Top-K sensitivity, Top-K specifity, AUC Malignancy prediction: AUC Skin lesion recognition: Top-K accuracy For this study, we set the value of K to 1, 3, and 5.

Documents

Study results

Study results_EUPAS108254

Study, other information

CEIM_LEGIT_MC_EVCDAO_2019 Modificacion FavorablePIB-5737d3c2e56f73d97a641783577c20ac