<p>The accurate segmentation of skin lesions in dermoscopic images is crucial for effective dermatological diagnosis and treatment, particularly in the early detection of melanoma. This study proposes a novel method that employs association rules within the HSV (Hue, Saturation, Value) colour space, which is designed to be robust to illumination variations, to optimise the segmentation process. Utilising the Apriori algorithm, we developed a rule-based system that improves the precision in identifying and delineating affected areas compared to traditional methods, generating data-driven rules that show the actual colour relationships within lesions and lead to a more accurate segmentation. Our approach was validated using a dataset from the International Skin Imaging Collaboration, where it demonstrated a segmentation accuracy of 98%. This significant improvement over existing K-means and RGB colour space methods illustrates the potential of association rules in medical image analysis. The method’s robustness against common imaging challenges in dermatoscopy, such as irregular lighting and occlusions, makes it a promising tool for enhancing diagnostic procedures and potentially reducing the need for invasive diagnostic methods. To evaluate the method’s robustness, we used the Jaccard index and Dice alongside accuracy to compare with state-of-the-art methods. </p>