Unlabeled skin lesion classification by self-supervised topology clustering network

Wang, Dan; Pang, Na; Wang, Yanying; Zhao, Hongwei

doi:10.1016/j.bspc.2021.102428

Cited by 38 publications

(14 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this work, we compare the relative performance of ML and DL models for the analysis of dermoscopic images of skin lesions. While the analysis of such images had long been thought to be challenging and of limited accuracy [8][9][10], recent ML and DL studies have drawn considerable interest and shown tremendous promise [7,[12][13][14][15][16][17][18][19][20]. Here, we focus on developing ML and DL models to identify whether a tumor is malignant or benign on the sole basis of dermoscopic images.…”

Section: Discussionmentioning

confidence: 99%

“…This approach had been previously deemed to be challenging and of limited accuracy [8][9][10] because of the wide variety of types of skin cancers and the resulting images [11]. In recent years, work based on machine learning (ML) and deep learning (DL) has led to renewed interest in such diagnostic tools [12][13][14][15][16][17][18][19][20]. Recent studies have employed preprocessing and segmentation to extract geometric information on the skin lesion, e.g., size and shape, to classify skin cancer images [6,21].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Machine Learning and Deep Learning Algorithms for Skin Cancer Classification from Dermoscopic Images

Bechelli

Delhommelle

2022

Bioengineering

View full text Add to dashboard Cite

We carry out a critical assessment of machine learning and deep learning models for the classification of skin tumors. Machine learning (ML) algorithms tested in this work include logistic regression, linear discriminant analysis, k-nearest neighbors classifier, decision tree classifier and Gaussian naive Bayes, while deep learning (DL) models employed are either based on a custom Convolutional Neural Network model, or leverage transfer learning via the use of pre-trained models (VGG16, Xception and ResNet50). We find that DL models, with accuracies up to 0.88, all outperform ML models. ML models exhibit accuracies below 0.72, which can be increased to up to 0.75 with ensemble learning. To further assess the performance of DL models, we test them on a larger and more imbalanced dataset. Metrics, such as the F-score and accuracy, indicate that, after fine-tuning, pre-trained models perform extremely well for skin tumor classification. This is most notably the case for VGG16, which exhibits an F-score of 0.88 and an accuracy of 0.88 on the smaller database, and metrics of 0.70 and 0.88, respectively, on the larger database.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Machine Learning and Deep Learning Algorithms for Skin Cancer Classification from Dermoscopic Images

Bechelli

Delhommelle

2022

Bioengineering

View full text Add to dashboard Cite

show abstract

“…Another method was a Self-supervised Topology Clustering Network (STCN) given by Wang. et al [ 28 ] to classify unlabelled data without requiring any prior class information. The clustering algorithm was used to organize anonymous data into clusters by maximizing modularity.…”

Section: Related Workmentioning

confidence: 99%

Melanoma Classification Using a Novel Deep Convolutional Neural Network with Dermoscopic Images

Kaur

GholamHosseini

Sinha

et al. 2022

Sensors

View full text Add to dashboard Cite

Automatic melanoma detection from dermoscopic skin samples is a very challenging task. However, using a deep learning approach as a machine vision tool can overcome some challenges. This research proposes an automated melanoma classifier based on a deep convolutional neural network (DCNN) to accurately classify malignant vs. benign melanoma. The structure of the DCNN is carefully designed by organizing many layers that are responsible for extracting low to high-level features of the skin images in a unique fashion. Other vital criteria in the design of DCNN are the selection of multiple filters and their sizes, employing proper deep learning layers, choosing the depth of the network, and optimizing hyperparameters. The primary objective is to propose a lightweight and less complex DCNN than other state-of-the-art methods to classify melanoma skin cancer with high efficiency. For this study, dermoscopic images containing different cancer samples were obtained from the International Skin Imaging Collaboration datastores (ISIC 2016, ISIC2017, and ISIC 2020). We evaluated the model based on accuracy, precision, recall, specificity, and F1-score. The proposed DCNN classifier achieved accuracies of 81.41%, 88.23%, and 90.42% on the ISIC 2016, 2017, and 2020 datasets, respectively, demonstrating high performance compared with the other state-of-the-art networks. Therefore, this proposed approach could provide a less complex and advanced framework for automating the melanoma diagnostic process and expediting the identification process to save a life.

show abstract

“…The framework of [13] proposes the Self-supervised Topology Clustering Network (STCN) to segment the skin images automatically. The STCN consists of a transformation-invariant network that comprises a feature extraction function, a self-expression function, and a self-supervision deep topology clustering algorithm.…”

Section: Literature Reviewmentioning

confidence: 99%

An automatic feature extraction technique from the images of granular parakeratosis disease

Janthakal¹,

Hosalli²

2022

Int. j. electr. comput. eng. syst. (Online)

View full text Add to dashboard Cite

The largest and most vital part of the human body is skin and any change in the features of skin is termed as a skin lesion. The paper considers granular parakeratosis lesion that is an epidermal reaction occurring due to the disorder of keratinization, and mainly seen in intertriginous areas. The manual inspection of the lesion features is a bit cumbersome due to which an automated system is proposed in this paper. The main goal is to determine the size and depth of granular parakeratosis lesions using the proposed ensemble algorithm, partition clustering and region properties method. As a flow of the proposed model, segmentation is done using U-net with binary cross entropy, features are extracted using partition clustering and region properties method, and classification is done using SVM 10-fold model. The proposed feature extraction method estimates the depth and absolute size of K lesions in each image by predicting the absolute height and width of the lesion in terms of pixel square. After extracting the features, classification is done, thereby obtaining an accuracy of 95%, sensitivity and specificity of 100%. The proposed model provides better performance compared to state-of-the-art models. The main application of this automated system is in dermatology field where some skin lesions have same features which makes the experts to diagnose the disease incorrectly. If the proposed system is incorporated, diagnosis can be done in an effective manner considering all the relevant features.

show abstract

Unlabeled skin lesion classification by self-supervised topology clustering network

Cited by 38 publications

References 15 publications

Machine Learning and Deep Learning Algorithms for Skin Cancer Classification from Dermoscopic Images

Machine Learning and Deep Learning Algorithms for Skin Cancer Classification from Dermoscopic Images

Melanoma Classification Using a Novel Deep Convolutional Neural Network with Dermoscopic Images

An automatic feature extraction technique from the images of granular parakeratosis disease

Contact Info

Product

Resources

About