Diagnosis of melanoma from dermoscopic images using a deep depthwise separable residual convolutional network

Abstract

Melanoma is one of the four major types of skin cancers caused by malignant growth in the melanocyte cells. It is the rarest one, accounting to only 1% of all skin cancer cases. However, it is the deadliest among all the skin cancer types. Owing to its rarity, efficient diagnosis of the disease becomes rather difficult. Here, a deep depthwise separable residual convolutional algorithm is introduced to perform binary melanoma classification on a dermoscopic skin lesion image dataset. Prior to training the model with the dataset noise removal from the images using non‐local means filter is performed followed by enhancement using contrast‐limited adaptive histogram equilisation over discrete wavelet transform algorithm. Images are fed to the model as multi‐channel image matrices with channels chosen across multiple color spaces based on their ability to optimize the performance of the model. Proper lesion detection and classification ability of the model are tested by monitoring the gradient weighted class activation maps and saliency maps, respectively. Dynamic effectiveness of the model is shown through its performance in multiple skin lesion image datasets. The proposed model achieved an ACC of 99.50% on international skin imaging collaboration (ISIC), 96.77% on PH2, 94.44% on DermIS and 95.23% on MED‐NODE datasets.

Deep Learning in Skin Disease Image Recognition: A Review

“…One-third of the collected literature use the mixture of multiple datasets [38], [42], [43], [46], [47], [49], [51], [52], [53], [58], [73], [76], [80], [85]. [49], [53], [54], [56], [57], [66], [71], [76], [79], [84], and [85] ISIC Archive [30] 23906 7 √ [46], [58], [59], [70], [77], [79], [80], and [81] DermIS [32] --× [46], [51], [58], and [80] DermQuest [33] --× [46], [51], and [80] DermNZ [34] --× [46] and [51] ISIC 2016 [35] 1279 2 √ [47], [52], …”

“…Singhal et al used filters to reduce the influence of noise [57]. Rahul et al used nonlocal means deoiling method to remove noise [58]. Xiaoyu added noise to the skin disease image to study image noise's influence on skin disease recognition [38].…”

“…Table III shows the details. [47], [50], [51], [52], [53], [58], and [65] Clipping [38], [44], [49], [53], [55], [56], [57], and [75] - [41], [42], [59], and [70] Normalization Data normalization [41], [47], [48], [52], [53], [54], [56], [58], [65], [70], and [75] This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ This article has been accepted for publication in a future issue of this journal, but has not been fully edited.…”

“…Twenty-seven pieces of literature used a single model in the literature, among which 10 papers ( [38], [38], [43], [57], [59], [64], [73], [75], [76], and [77]) has no model improvement, and the other 15 articles have optimized the model. The optimization methods mainly include: modifying the fully connected layer (five articles: [41], [44], [61], [74], and [80]), using a machine learning classifier (two articles: [65] and [81]), applying (1 article: [51]) the model to the mobile application, introducing attention mechanism (two articles: [50] and [55]), self-building network model (two articles: [58] and [70]), and multilevel classification (three articles: [46], [56], and [78]). Kawahara et al [41] and Yuexiang et al [44] formed a fully CNN based on CNNs and removed the fully connected layers.…”

“…The specific performance indicators are shown in Table VIII. [38], [41], [45], [50], [53], [56], [57], [59], [64], [65], [42], [44], [51], [58], [62], [71], [72], [74], [79], [80], [82], [84],and [85] [38], [44], [50], [51], [54], [61], [62], [64], [65], [71], [75], [76], [77], [78], [79], [80], [81], [83], [84],and [85] Among the negative samples, the correct proportion of samples was predicted [44], [45], [50], [51], [64], [65],…”