Psoriasis is a chronic skin condition characterized by the rapid growth of skin cells, leading to the formation of thick, scaly patches. Detecting these lesions accurately is crucial for early diagnosis and effective treatment. However, the variability in lesion size, shape, and appearance across different skin types presents significant challenges for automated detection systems. Traditional methods often struggle with these variations, leading to reduced accuracy and efficiency. In recent years, convolutional neural networks (CNNs) have emerged as a powerful tool for medical image analysis, particularly in the detection of skin conditions. This paper introduces a novel approach for psoriasis lesion detection using convolutional neural networks (CNNs). The challenge of accurately identifying lesions across various skin types and lesion sizes is addressed by leveraging a scale-aware strategy. In this method, a CNN first estimates the scale distribution of psoriasis lesions within an image by predicting a scale histogram. This histogram serves as a guide for image scaling operations, ensuring that lesions are brought to a uniform scale for subsequent detection. By applying zoom-in and zoom-out transformations based on the predicted scale histogram, the model reduces computational overhead while maintaining high detection accuracy. The proposed approach demonstrates improved efficiency and accuracy in identifying psoriasis lesions, as compared to traditional multi-scale testing techniques, with reduced computational demands. Extensive qualitative and quantitative evaluations on various skin datasets show that the scale-aware strategy significantly enhances lesion detection performance, making it a promising tool for automated psoriasis diagnosis.