皮肤黑色素瘤癌的检测和诊断对人类生命的拯救至关重要。本文的主要目标是对皮肤癌症在皮肤镜图像中进行检测和诊断。皮肤癌症检测和诊断系统都使用深度学习架构以实现有效的性能提升。检测过程通过识别癌细胞影响的皮肤镜图像来进行，而诊断过程是通过估计皮肤图像中分割癌区域的严重程度来进行。本文提出了并行CNN架构，以将皮肤图像分类为黑色素瘤或健康。最初，本文提出了颜色映射直方图均衡化（CMHE）方法来增强源皮肤图像，然后使用模糊系统从增强的皮肤图像中检测出粗和细边缘。从检测出的边缘图像中提取灰度共生矩阵（GLCM）和Law's纹理特征，并利用遗传算法（GA）方法对这些特征进行优化。进一步地，优化后的特征将通过开发的管道内部模块架构（PIMA）的深度学习结构进行分类。分类为黑色素瘤皮肤图像中的癌症区域将使用数学形态学处理进行分割，这些分割的癌症区域将使用提出的PIMA结构诊断为轻度或严重。提出的基于PIMA的皮肤癌症分类系统已应用并测试于ISIC和HAM 10000皮肤图像数据集。研究亮点：利用皮肤镜图像检测和分类黑色素瘤皮肤癌。使用颜色映射直方图均衡化增强皮肤镜图像。从增强的皮肤图像中提取GLCM和Law's纹理特征。提出了管道内部模块架构（PIMA）以对皮肤图像进行分类。© 2023 Wiley Periodicals LLC.

Detection and diagnosis of melanoma skin cancer is important to save the life of humans. The main objective of this article is to perform both detection and diagnosis of the skin cancers in dermoscopy images. Both skin cancer detection and diagnosis system uses deep learning architectures for the effective performance improvement as the main objective. The detection process involves by identifying the cancer affected skin dermoscopy images and the diagnosis process involves by estimating the severity levels of the segmented cancer regions in skin images. This article proposes parallel CNN architecture for the classification of skin images into either melanoma or healthy. Initially, color map histogram equalization (CMHE) method is proposed in this article to enhance the source skin images and then thick and thin edges are detected from the enhanced skin image using the Fuzzy system. The gray-level co-occurrence matrix (GLCM) and Law's texture features are extracted from the edge detected images and these features are optimized using genetic algorithm (GA) approach. Further, the optimized features are classified by the developed pipelined internal module architecture (PIMA) of deep learning structure. The cancer regions in the classified melanoma skin images are segmented using mathematical morphological process and these segmented cancer regions are diagnosed into either mild or severe using the proposed PIMA structure. The proposed PIMA-based skin cancer classification system is applied and tested on ISIC and HAM 10000 skin image datasets. RESEARCH HIGHLIGHTS: The melanoma skin cancer is detected and classified using dermoscopy images. The skin dermoscopy images are enhanced using color map histogram equalization. GLCM and Law's texture features are extracted from the enhanced skin images. To propose pipelined internal module architecture (PIMA) for the classification of skin images.© 2023 Wiley Periodicals LLC.