International Skin Imaging Collaboration Research Articles

A novel approach toward skin cancer classification through fused deep features and neutrosophic environment.

Variations in the size and texture of melanoma make the classification procedure more complex in a computer-aided diagnostic (CAD) system. The research proposes an innovative hybrid deep learning-based layer-fusion and neutrosophic-set technique for identifying skin lesions. The off-the-shelf networks are examined to categorize eight types of skin lesions using transfer learning on International Skin Imaging Collaboration (ISIC) 2019 skin lesion datasets. The top two networks, which are GoogleNet and DarkNet, achieved an accuracy of 77.41 and 82.42%, respectively. The proposed method works in two successive stages: first, boosting the classification accuracy of the trained networks individually. A suggested feature fusion methodology is applied to enrich the extracted features' descriptive power, which promotes the accuracy to 79.2 and 84.5%, respectively. The second stage explores how to combine these networks for further improvement. The error-correcting output codes (ECOC) paradigm is utilized for constructing a set of well-trained true and false support vector machine (SVM) classifiers via fused DarkNet and GoogleNet feature maps, respectively. The ECOC's coding matrices are designed to train each true classifier and its opponent in a one-versus-other fashion. Consequently, contradictions between true and false classifiers in terms of their classification scores create an ambiguity zone quantified by the indeterminacy set. Recent neutrosophic techniques resolve this ambiguity to tilt the balance toward the correct skin cancer class. As a result, the classification score is increased to 85.74%, outperforming the recent proposals by an obvious step. The trained models alongside the implementation of the proposed single-valued neutrosophic sets (SVNSs) will be publicly available for aiding relevant research fields.

Skin Lesion Segmentation in Dermoscopic Images with Noisy Data.

We propose a deep learning approach to segment the skin lesion in dermoscopic images. The proposed network architecture uses a pretrained EfficientNet model in the encoder and squeeze-and-excitation residual structures in the decoder. We applied this approach on the publicly available International Skin Imaging Collaboration (ISIC) 2017 Challenge skin lesion segmentation dataset. This benchmark dataset has been widely used in previous studies. We observed many inaccurate or noisy ground truth labels. To reduce noisy data, we manually sorted all ground truth labels into three categories - good, mildly noisy, and noisy labels. Furthermore, we investigated the effect of such noisy labels in training and test sets. Our test results show that the proposed method achieved Jaccard scores of 0.807 on the official ISIC 2017 test set and 0.832 on the curated ISIC 2017 test set, exhibiting better performance than previously reported methods. Furthermore, the experimental results showed that the noisy labels in the training set did not lower the segmentation performance. However, the noisy labels in the test set adversely affected the evaluation scores. We recommend that the noisy labels should be avoided in the test set in future studies for accurate evaluation of the segmentation algorithms.

Prediction of melanoma Breslow thickness using deep transfer learning algorithms.

The distinction between in situ melanoma (MIS) and invasive melanoma is challenging even for expert dermatologists. The use of pretrained convolutional neural networks (CNNs) as ancillary decision systems needs further research. To develop, validate and compare three deep transfer learning (DTL) algorithms to predict MIS vs. invasive melanoma and melanoma with a Breslow thickness (BT) of < 0.8 mm vs. ≥ 0.8 mm. A dataset of 1315 dermoscopic images of histopathologically confirmed melanomas was created from Virgen del Rocio University Hospital and open repositories of the International Skin Imaging Collaboration archive and PolesieS et al.(Dermatol Pract Concept 2021; 11:e2021079). The images were labelled as MIS or invasive melanoma and < 0.8 mmor ≥ 0.8 mm of BT. We conducted three trainings, and overall means for receiver operating characteristic (ROC) curves, sensitivity, specificity, positive and negative predictive value, and balanced diagnostic accuracy outcomes were evaluated on the test set with ResNetV2, EfficientNetB6 and InceptionV3. The results of 10 dermatologists were compared with the algorithms. Grad-CAM gradient maps were generated, highlighting relevant areas considered by the CNNs within the images. EfficientNetB6 achieved the highest diagnostic accuracy for the comparison between MIS vs. invasive melanoma (61%) and BT < 0.8 mm vs. ≥ 0.8 mm (75%). For the BT comparison, ResNetV2 with an area under the ROC curve of 0.76 and InceptionV3 with an area under the ROC curve of 0.75, outperformed the results obtained by the dermatologist group with an area under the ROC curve of 0.70. EfficientNetB6 recorded the best prediction results, outperforming the dermatologists for the comparison of 0.8 mm of BT. DTL could be an ancillary aid to support dermatologists' decisions in the near future.

A Hybrid Stacked Restricted Boltzmann Machine with Sobel Directional Patterns for Melanoma Prediction in Colored Skin Images

Melanoma, a kind of skin cancer that is very risky, is distinguished by uncontrolled cell multiplication. Melanoma detection is of the utmost significance in clinical practice because of the atypical border structure and the numerous types of tissue it can involve. The identification of melanoma is still a challenging process for color images, despite the fact that numerous approaches have been proposed in the research that has been done. In this research, we present a comprehensive system for the efficient and precise classification of skin lesions. The framework includes preprocessing, segmentation, feature extraction, and classification modules. Preprocessing with DullRazor eliminates skin-imaging hair artifacts. Next, Fully Connected Neural Network (FCNN) semantic segmentation extracts precise and obvious Regions of Interest (ROIs). We then extract relevant skin image features from ROIs using an enhanced Sobel Directional Pattern (SDP). For skin image analysis, Sobel Directional Pattern outperforms ABCD. Finally, a stacked Restricted Boltzmann Machine (RBM) classifies skin ROIs. Stacked RBMs accurately classify skin melanoma. The experiments have been conducted on five datasets: Pedro Hispano Hospital (PH2), International Skin Imaging Collaboration (ISIC 2016), ISIC 2017, Dermnet, and DermIS, and achieved an accuracy of 99.8%, 96.5%, 95.5%, 87.9%, and 97.6%, respectively. The results show that a stack of Restricted Boltzmann Machines is superior for categorizing skin cancer types using the proposed innovative SDP.

Classification of skin lesion images using modified Inception V3 model with transfer learning and augmentation techniques

In this article, a methodological approach to classifying malignant melanoma in dermoscopy images is presented. Early treatment of skin cancer increases the patient’s survival rate. The classification of melanoma skin cancer in the early stages is decided by dermatologists to treat the patient appropriately. Dermatologists need more time to diagnose affected skin lesions due to high resemblance between melanoma and benign. In this paper, a deep learning based Computer-Aided Diagnosis (CAD) system is developed to accurately classify skin lesions with a high classification rate. A new architecture has been framed to classify the skin lesion diseases using the Inception v3 model as a baseline architecture. The extracted features from the Inception Net are then flattened and are given to the DenseNet block to extracts more fine grained features of the lesion disease. The International Skin Imaging Collaboration (ISIC) archive datasets contains 3307 dermoscopy images which includes both benign and malignant skin images. The dataset images are trained using the proposed architecture with the learning rate of 0.0001, batch size 64 using various optimizer. The performance of the proposed model has also been evaluated using confusion matrix and ROC-AUC curves. The experimental results show that the proposed model attains a highest accuracy rate of 91.29 % compared to other state-of-the-art methods like ResNet, VGG-16, DenseNet, MobileNet. A confusion matrix and ROC curve are used to evaluate the performance analysis of skin images. The classification accuracy, sensitivity, specificity, testing accuracy, and AUC values were obtained at 90.33%, 82.87%, 91.29%, 87.12%, and 87.40%.

Category-aware feature attribution for Self-Optimizing medical image classification

Convolutional Neural Networks (CNNs) have achieved state-of-the-art performance for automatic medical image classification. Although remarkable advancement has been made, they have not demonstrated sufficiently accurate and robust results for clinical use. In addition, they are limited by the unavoidable hard samples and minority class samples. Meanwhile, it is hard to understand the prediction mechanism and debug the wrong prediction. To address these issues, this paper proposes a Self-Optimizing medical image classification framework based on category-aware feature attribution, which automatically performs model diagnosis, repairing, and enhancement. On one hand, we introduce selective filter restraining to alleviate the disturbance of misclassified features for hard samples. On the other hand, we introduce sample-specific fine-tuning to promote category-wise feature learning, especially for minority classes. The Self-Optimizing framework is a versatile module for other medical image classifiers based on mainstream CNN structures. We extensively validated the proposed method on three typical yet challenging medical image classification tasks: (i) skin lesion classification from dermoscopic images in the International Skin Imaging Collaboration (ISIC) 2019 dataset, (ii) breast cancer classification from ultrasound images in Breast Ultrasound Images (BUSI) dataset, and (iii) breast tumour classification from histopathologic scans of lymph node sections in PatchCamelyon (PCam) dataset. Experimental results demonstrate that the proposed framework achieves 1%∼5% performance gain for various classifiers, brings better explainability, and can be applied to incremental learning scenarios.

GFANet: Gated Fusion Attention Network for skin lesion segmentation

Automatic segmentation of skin lesions is crucial for diagnosing and treating skin diseases. Although current medical image segmentation methods have significantly improved the results of skin lesion segmentation, the following major challenges still affect the segmentation performance: (i) segmentation targets have irregular shapes and diverse sizes and (ii) low contrast or blurred boundaries between lesions and background. To address these issues, this study proposes a Gated Fusion Attention Network (GFANet) which designs two progressive relation decoders to accurately segment skin lesions images. First, we use a Context Features Gated Fusion Decoder (CGFD) to fuse multiple levels of contextual features, and then a prediction result is generated as the initial guide map. Then, it is optimized by a prediction decoder consisting of a shape flow and a final Gated Convolution Fusion (GCF) module, where we iteratively use a set of Channel Reverse Attention (CRA) modules and GCF modules in the shape flow to combine the features of the current layer and the prediction results of the adjacent next layer to gradually extract boundary information. Finally, to speed up network convergence and improve segmentation accuracy, we use GCF to fuse low-level features from the encoder and the final output of the shape flow. To verify the effectiveness and advantages of the proposed GFANet, we conduct extensive experiments on four publicly available skin lesion datasets (International Skin Imaging Collaboration [ISIC] 2016, ISIC 2017, ISIC 2018, and PH2) and compare them with state-of-the-art methods. The experimental results show that the proposed GFANet achieves excellent segmentation performance in commonly used evaluation metrics, and the segmentation results are stable. The source code is available at https://github.com/ShiHanQ/GFANet.

Improving Artificial Intelligence−Based Diagnosis on Pediatric Skin Lesions

Artificial intelligence algorithms to classify melanoma are dependent on their training data, which limits generalizability. The objective of this study was to compare the performance of an artificial intelligence model trained on a standard adult-predominant dermoscopic dataset before and after the addition of additional pediatric training images. The performances were compared using held-out adult and pediatric test sets of images. We trained two models: one (model A) on an adult-predominant dataset (37,662 images from the International Skin Imaging Collaboration) and the other (model A+P) on an additional 1,536 pediatric images. We compared performance between the two models on adult and pediatric held-out test images separately using the area under the receiver operating characteristic curve. We then used Gradient-weighted Class Activation Maps and background skin masking to understand the contributions of the lesion versus background skin to algorithm decision making. Adding images from a pediatric population with different epidemiological and visual patterns to current reference standard datasets improved algorithm performance on pediatric images without diminishing performance on adult images. This suggests a way that dermatologic artificial intelligence models can be made more generalizable. The presence of background skin was important to the pediatric-specific improvement seen between models. Our study highlights the importance of carefully curated and labeled data from diverse inputs to improve the generalizability of AI models for dermatology, in this case applied to dermoscopic images of adult and pediatric lesions to improve melanoma detection.

Improving Automatic Melanoma Diagnosis Using Deep Learning-Based Segmentation of Irregular Networks.

Deep learning has achieved significant success in malignant melanoma diagnosis. These diagnostic models are undergoing a transition into clinical use. However, with melanoma diagnostic accuracy in the range of ninety percent, a significant minority of melanomas are missed by deep learning. Many of the melanomas missed have irregular pigment networks visible using dermoscopy. This research presents an annotated irregular network database and develops a classification pipeline that fuses deep learning image-level results with conventional hand-crafted features from irregular pigment networks. We identified and annotated 487 unique dermoscopic melanoma lesions from images in the ISIC 2019 dermoscopic dataset to create a ground-truth irregular pigment network dataset. We trained multiple transfer learned segmentation models to detect irregular networks in this training set. A separate, mutually exclusive subset of the International Skin Imaging Collaboration (ISIC) 2019 dataset with 500 melanomas and 500 benign lesions was used for training and testing deep learning models for the binary classification of melanoma versus benign. The best segmentation model, U-Net++, generated irregular network masks on the 1000-image dataset. Other classical color, texture, and shape features were calculated for the irregular network areas. We achieved an increase in the recall of melanoma versus benign of 11% and in accuracy of 2% over DL-only models using conventional classifiers in a sequential pipeline based on the cascade generalization framework, with the highest increase in recall accompanying the use of the random forest algorithm. The proposed approach facilitates leveraging the strengths of both deep learning and conventional image processing techniques to improve the accuracy of melanoma diagnosis. Further research combining deep learning with conventional image processing on automatically detected dermoscopic features is warranted.

An optimal diagnosis system for melanoma dermoscopy images based on enhanced design of horse herd optimizer

AbstractSkin cells grow in an orderly and controlled manner, in which old cells are pushed to the skin surface by healthy new cells, wherein they finally perish. However, when several cells damage DNA, new ones may start growing disorderly and can finally create cancer cells mass. Different types of skin cancer have been observed by experts. Although, melanoma, as one of the rarely happened cancers, is the most threatening type. Therefore, early detection of this type of cancer can be so useful for avoiding melanoma dangers and even helps to treat this type of cancer. The present paper proposes a new hierarchical procedure for the optimum diagnosis of melanoma cancer from dermoscopy images. Here, after image preprocessing, image segmentation is applied for the segmentation of the ROI. Then, the selection of features from the segmented images is performed and injected to a radial basis function‐based classifier to provide the final diagnosis of cancer. To deliver efficient results, the feature selection and the classifier have been optimized by a new design of Horse herd optimization algorithm (HHOA). The method is then implemented on the Society for Imaging Informatics in Medicine‐The International Skin Imaging Collaboration (SIIM‐ISIC) Melanoma dataset to validate its effectiveness and its achievements and also put in comparison with several different latest techniques. The results show that the proposed method with 95.46% precision provides the maximum performance among the others.

A Novel Multi-Task Learning Network Based on Melanoma Segmentation and Classification with Skin Lesion Images.

Melanoma is known worldwide as a malignant tumor and the fastest-growing skin cancer type. It is a very life-threatening disease with a high mortality rate. Automatic melanoma detection improves the early detection of the disease and the survival rate. In accordance with this purpose, we presented a multi-task learning approach based on melanoma recognition with dermoscopy images. Firstly, an effective pre-processing approach based on max pooling, contrast, and shape filters is used to eliminate hair details and to perform image enhancement operations. Next, the lesion region was segmented with a VGGNet model-based FCN Layer architecture using enhanced images. Later, a cropping process was performed for the detected lesions. Then, the cropped images were converted to the input size of the classifier model using the very deep super-resolution neural network approach, and the decrease in image resolution was minimized. Finally, a deep learning network approach based on pre-trained convolutional neural networks was developed for melanoma classification. We used the International Skin Imaging Collaboration, a publicly available dermoscopic skin lesion dataset in experimental studies. While the performance measures of accuracy, specificity, precision, and sensitivity, obtained for segmentation of the lesion region, were produced at rates of 96.99%, 92.53%, 97.65%, and 98.41%, respectively, the performance measures achieved rates for classification of 97.73%, 99.83%, 99.83%, and 95.67%, respectively.

Regularised Layerwise Weight Norm Based Skin Lesion Features Extraction and Classification

Melanoma is the most lethal malignant tumour, and its prevalence is increasing. Early detection and diagnosis of skin cancer can alert patients to manage precautions and dramatically improve the lives of people. Recently, deep learning has grown increasingly popular in the extraction and categorization of skin cancer features for effective prediction. A deep learning model learns and co-adapts representations and features from training data to the point where it fails to perform well on test data. As a result, overfitting and poor performance occur. To deal with this issue, we proposed a novel Consecutive Layerwise weight Constraint MaxNorm model (CLCM-net) for constraining the norm of the weight vector that is scaled each time and bounding to a limit. This method uses deep convolutional neural networks and also custom layer-wise weight constraints that are set to the whole weight matrix directly to learn features efficiently. In this research, a detailed analysis of these weight norms is performed on two distinct datasets, International Skin Imaging Collaboration (ISIC) of 2018 and 2019, which are challenging for convolutional networks to handle. According to the findings of this work, CLCM-net did a better job of raising the model’s performance by learning the features efficiently within the size limit of weights with appropriate weight constraint settings. The results proved that the proposed techniques achieved 94.42% accuracy on ISIC 2018, 91.73% accuracy on ISIC 2019 datasets and 93% of accuracy on combined dataset.

HI-MViT: A lightweight model for explainable skin disease classification based on modified MobileViT.

To develop an explainable lightweight skin disease high-precision classification model that can be deployed to the mobile terminal. In this study, we present HI-MViT, a lightweight network for explainable skin disease classification based on Modified MobileViT. HI-MViT is mainly composed of ordinary convolution, Improved-MV2, MobileViT block, global pooling, and fully connected layers. Improved-MV2 uses the combination of shortcut and depth classifiable convolution to substantially decrease the amount of computation while ensuring the efficient implementation of information interaction and memory. The MobileViT block can efficiently encode local and global information. In addition, semantic feature dimensionality reduction visualization and class activation mapping visualization methods are used for HI-MViT to further understand the attention area of the model when learning skin lesion images. The International Skin Imaging Collaboration has assembled and made available the ISIC series dataset. Experiments using the HI-MViT model on the ISIC-2018 dataset achieved scores of 0.931, 0.932, 0.961, and 0.977 on F1-Score, Accuracy, Average Precision (AP), and area under the curve (AUC). Compared with the top five algorithms of ISIC-2018 Task 3, Marco's average F1-Score, AP, and AUC have increased by 6.9%, 6.8%, and 0.8% compared with the suboptimal performance model. Compared with ConvNeXt, the most competitive convolutional neural network architecture, our model is 5.0%, 3.4%, 2.3%, and 2.2% higher in F1-Score, Accuracy, AP, and AUC, respectively. The experiments on the ISIC-2017 dataset also achieved excellent results, and all indicators were better than the top five algorithms of ISIC-2017 Task 3. Using the trained model to test on the PH2 dataset, an excellent performance score is obtained, which shows that it has good generalization performance. The skin disease classification model HI-MViT proposed in this article shows excellent classification performance and generalization performance in experiments. It demonstrates how the classification outcomes can be applied to dermatologists' computer-assisted diagnostics, enabling medical professionals to classify various dermoscopic images more rapidly and reliably.

A Deep CNN Transformer Hybrid Model for Skin Lesion Classification of Dermoscopic Images Using Focal Loss.

Skin cancers are the most cancers diagnosed worldwide, with an estimated > 1.5 million new cases in 2020. Use of computer-aided diagnosis (CAD) systems for early detection and classification of skin lesions helps reduce skin cancer mortality rates. Inspired by the success of the transformer network in natural language processing (NLP) and the deep convolutional neural network (DCNN) in computer vision, we propose an end-to-end CNN transformer hybrid model with a focal loss (FL) function to classify skin lesion images. First, the CNN extracts low-level, local feature maps from the dermoscopic images. In the second stage, the vision transformer (ViT) globally models these features, then extracts abstract and high-level semantic information, and finally sends this to the multi-layer perceptron (MLP) head for classification. Based on an evaluation of three different loss functions, the FL-based algorithm is aimed to improve the extreme class imbalance that exists in the International Skin Imaging Collaboration (ISIC) 2018 dataset. The experimental analysis demonstrates that impressive results of skin lesion classification are achieved by employing the hybrid model and FL strategy, which shows significantly high performance and outperforms the existing work.

Automated Malignant Melanoma Classification Using Convolutional Neural Networks

This research is proposed a design of architecture for melanoma (a kind of skin cancer) recognition by using a Convolutional Neural Network (CNN), work that will be useful for researchers in future projects in areas like biomedicine, machine learning, and others related moving forward with their studies and improving this proposal. CNN is mostly used in computer vision (a branch of artificial intelligence), applied to pattern recognition in skin moles and to determine the existence of malignant melanoma, or not, with a limited dataset. The CNN classifier designed and trained in this case was built through a couple of layers of convolution and pooling stacked to form a neural network of 6 layers followed by the fully connected to complete the architecture with an output classifier. The proposed database to train our CNN is the largest publicly collection of dermoscopic images of melanomas and other skin lesions, provided by the International Skin Imaging Collaboration (ISIC), sponsored by International Society for Digital Imaging of the Skin (ISDIS), an international effort to improve melanoma diagnosis. The purpose of this research was to design a Convolutional Neural Network with a high level of accuracy to help professionals in medicine with a melanoma diagnosis, in this case, it was possible to get accuracy up to 88.75 %.

Image segmentation of skin lesions based on dense atrous spatial pyramid pooling and attention mechanism

The skin is the largest organ of the human body, and many visceral diseases will be directly reflected on the skin, so it is of great clinical significance to accurately segment the skin lesion images. To address the characteristics of complex color, blurred boundaries, and uneven scale information, a skin lesion image segmentation method based on dense atrous spatial pyramid pooling (DenseASPP) and attention mechanism is proposed. The method is based on the U-shaped network (U-Net). Firstly, a new encoder is redesigned to replace the ordinary convolutional stacking with a large number of residual connections, which can effectively retain key features even after expanding the network depth. Secondly, channel attention is fused with spatial attention, and residual connections are added so that the network can adaptively learn channel and spatial features of images. Finally, the DenseASPP module is introduced and redesigned to expand the perceptual field size and obtain multi-scale feature information. The algorithm proposed in this paper has obtained satisfactory results in the official public dataset of the International Skin Imaging Collaboration (ISIC 2016). The mean Intersection over Union (mIOU), sensitivity (SE), precision (PC), accuracy (ACC), and Dice coefficient (Dice) are 0.901 8, 0.945 9, 0.948 7, 0.968 1, 0.947 3, respectively. The experimental results demonstrate that the method in this paper can improve the segmentation effect of skin lesion images, and is expected to provide an auxiliary diagnosis for professional dermatologists.

Squeeze-MNet: Precise Skin Cancer Detection Model for Low Computing IoT Devices Using Transfer Learning

Cancer remains a deadly disease. We developed a lightweight, accurate, general-purpose deep learning algorithm for skin cancer classification. Squeeze-MNet combines a Squeeze algorithm for digital hair removal during preprocessing and a MobileNet deep learning model with predefined weights. The Squeeze algorithm extracts important image features from the image, and the black-hat filter operation removes noise. The MobileNet model (with a dense neural network) was developed using the International Skin Imaging Collaboration (ISIC) dataset to fine-tune the model. The proposed model is lightweight; the prototype was tested on a Raspberry Pi 4 Internet of Things device with a Neo pixel 8-bit LED ring; a medical doctor validated the device. The average precision (AP) for benign and malignant diagnoses was 99.76% and 98.02%, respectively. Using our approach, the required dataset size decreased by 66%. The hair removal algorithm increased the accuracy of skin cancer detection to 99.36% with the ISIC dataset. The area under the receiver operating curve was 98.9%.

Cascaded Hough Transform-Based Hair Mask Generation and Harmonic Inpainting for Automated Hair Removal from Dermoscopy Images.

Restoring information obstructed by hair is one of the main issues for the accurate analysis and segmentation of skin images. For retrieving pixels obstructed by hair, the proposed system converts dermoscopy images into the L*a*b* color space, then principal component analysis (PCA) is applied to produce grayscale images. Afterward, the contrast-limited adaptive histogram equalization (CLAHE) and the average filter are implemented to enhance the grayscale image. Subsequently, the binary image is generated using the iterative thresholding method. After that, the Hough transform (HT) is applied to each image block to generate the hair mask. Finally, the hair pixels are removed by harmonic inpainting. The performance of the proposed automated hair removal was evaluated by applying the proposed system to the International Skin Imaging Collaboration (ISIC) dermoscopy dataset as well as to clinical images. Six performance evaluation metrics were measured, namely the mean squared error (MSE), the peak signal-to-noise ratio (PSNR), the signal-to-noise ratio (SNR), the structural similarity index (SSIM), the universal quality image index (UQI), and the correlation (C). Using the clinical dataset, the system achieved MSE, PSNR, SNR, SSIM, UQI, and C values of 34.7957, 66.98, 42.39, 0.9813, 0.9801, and 0.9985, respectively. The results demonstrated that the proposed system could satisfy the medical diagnostic requirements and achieve the best performance compared to the state-of-art.

A multi-stage melanoma recognition framework with deep residual neural network and hyperparameter optimization-based decision support in dermoscopy images

This paper developed a novel melanoma diagnosis model from dermoscopy images using a novel hybrid model. Melanoma is the most dangerous and rarest type of skin cancer. It is seen because of the uncontrolled proliferation of melanocyte cells that give color to the skin. Dermoscopy is a critical auxiliary diagnostic method in the differentiation of pigmented moles, which show moles by magnifying 10–20 times from skin cancers. This paper proposes a multi-stage melanoma recognition framework withskin lesion images obtained from dermoscopy. This model developed a practical pre-processing approach that includes dilation andpooling layers to remove hair details and reveal details in dermoscopy images. A deep residual neural network was then utilized as the feature extractor for processed images.Additionally, the Relief algorithm selected practical and distinctive features from these features. Finally, these selected features were fed to the input of the support vector machine (SVM) classifier. In addition, the Bayesian optimization algorithm was used for the optimum parameter selection of the SVM method. The International Skin Imaging Collaboration (ISIC-2019 and ISIC-2020) datasets were used to test the performance of the proposed model. As a result, the proposed model produced approximately 99% accuracy forclassifying melanoma or benign from skin lesion images. These results show that the proposed model can help physicians to automatically identify melanoma based on dermatological imaging.

Handling dataset dependence with model ensembles for skin lesion classification from dermoscopic and clinical images

AbstractDataset dependence affects many real‐life applications of machine learning: the performance of a model trained on a dataset is significantly worse on samples from another dataset than on new, unseen samples from the original one. This issue is particularly acute for small and somewhat specific databases in medical applications; the automated recognition of melanoma from skin lesion images is a prime example. We document dataset dependence in dermoscopic skin lesion image classification using three publicly available medium size datasets. Standard machine learning techniques aimed at improving the predictive power of a model might enhance performance slightly, but the gain is small, the dataset dependence is not reduced, and the best combination depends on model details. We demonstrate that simple differences in image statistics account for only 5% of the dataset dependence. We suggest a solution with two essential ingredients: using an ensemble of heterogeneous models, and training on a heterogeneous dataset. Our ensemble consists of 29 convolutional networks, some of which are trained on features considered important by dermatologists; the networks' output is fused by a trained committee machine. The combined International Skin Imaging Collaboration dataset is suitable for training, as it is multi‐source, produced by a collaboration of a number of clinics over the world. Building on the strengths of the ensemble, it is applied to a related problem as well: recognizing melanoma based on clinical (non‐dermoscopic) images. This is a harder problem as both the image quality is lower than those of the dermoscopic ones and the available public datasets are smaller and scarcer. We explored various training strategies and showed that 79% balanced accuracy can be achieved for binary classification averaged over three clinical datasets.