An improved transformer network for skin cancer classification

Xin, Chao; Liu, Zhifang; Zhao, Ke; Miao, Linlin; Ma, Yizhao; Zhu, Xiaoxia; Zhou, Qiongyan; Wang, Songting; Li, Lingzhi; Yang, Feng; Xu, Suling; Chen, Haijiang

doi:10.1016/j.compbiomed.2022.105939

Cited by 101 publications

(47 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Xin et al [ 176 ] explore Vision Transformer (VIT) and apply on skin lesion classification using multi-scale and overlapping sliding windows. Current decades have seen significant advancements in the initial detection and medication of melanoma, whose prevalence is rising annually throughout the world and pose potential a serious risk to human health.…”

Section: Deep Learning For Medical Image Analysis and Cadmentioning

confidence: 99%

On the Analyses of Medical Images Using Traditional Machine Learning Techniques and Convolutional Neural Networks

Iqbal

Qureshi

et al. 2023

Arch Computat Methods Eng

View full text Add to dashboard Cite

Convolutional neural network (CNN) has shown dissuasive accomplishment on different areas especially Object Detection, Segmentation, Reconstruction (2D and 3D), Information Retrieval, Medical Image Registration, Multi-lingual translation, Local language Processing, Anomaly Detection on video and Speech Recognition. CNN is a special type of Neural Network, which has compelling and effective learning ability to learn features at several steps during augmentation of the data. Recently, different interesting and inspiring ideas of Deep Learning (DL) such as different activation functions, hyperparameter optimization, regularization, momentum and loss functions has improved the performance, operation and execution of CNN Different internal architecture innovation of CNN and different representational style of CNN has significantly improved the performance. This survey focuses on internal taxonomy of deep learning, different models of vonvolutional neural network, especially depth and width of models and in addition CNN components, applications and current challenges of deep learning.

show abstract

Section: Deep Learning For Medical Image Analysis and Cadmentioning

confidence: 99%

On the Analyses of Medical Images Using Traditional Machine Learning Techniques and Convolutional Neural Networks

Iqbal

Qureshi

et al. 2023

Arch Computat Methods Eng

View full text Add to dashboard Cite

show abstract

“…In the last three years, researches started to apply Transformer‐based DL approaches to pattern recognition of tumor images and tumor classification. [ 121–123 ] One vision transformer (VIT) model by Xin et al. [ 121 ] had superior performance in skin cancer classification using benchmarking dermatoscopy data set in.…”

Section: Artificial Intelligencementioning

confidence: 99%

Application of Artificial Intelligence to In Vitro Tumor Modeling and Characterization of the Tumor Microenvironment

Lee

Goh

et al. 2023

Adv Healthcare Materials

View full text Add to dashboard Cite

In vitro tumor models have played vital roles in enhancing the understanding of the cellular and molecular composition of tumors, as well as their biochemical and biophysical characteristics. Advances in technology have enabled the evolution of tumor models from two-dimensional cell cultures to three-dimensional printed tumor models with increased levels of complexity and diverse output parameters. With the increase in complexity, the new generation of models is able to replicate the architecture and heterogeneity of the tumor microenvironment more realistically than their predecessors. In recent years, artificial intelligence (AI) has been used extensively in healthcare and research, and AI-based tools have also been applied to the precise development of tumor models. The incorporation of AI facilitates the use of high-throughput systems for real-time monitoring of tumorigenesis and biophysical tumor properties, raising the possibility of using AI alongside tumor modeling for personalized medicine. Here, the integration of AI tools within tumor modeling is reviewed, including microfluidic devices and cancer-on-chip models.

show abstract

“…After the ViT got researchers' attention because of its strong performance, there was some work in the field of skin cancer classification/segmentation based on the ViT model in 2022. In the work [30], researchers proposed a new method in the image feature embedding block of the original ViT model combined with a contrastive learning method. [32] conducted experiments addressing the bottleneck of the original ViT with their improved position encoding method.…”

Section: Related Workmentioning

confidence: 99%

A Novel Vision Transformer Model for Skin Cancer Classification

Yang

Luo

Greer

2023

Neural Process Lett

View full text Add to dashboard Cite

Skin cancer can be fatal if it is found to be malignant. Modern diagnosis of skin cancer heavily relies on visual inspection through clinical screening, dermoscopy, or histopathological examinations. However, due to similarity among cancer types, it is usually challenging to identify the type of skin cancer, especially at its early stages. Deep learning techniques have been developed over the last few years and have achieved success in helping to improve the accuracy of diagnosis and classification. However, the latest deep learning algorithms still do not provide ideal classification accuracy. To further improve the performance of classification accuracy, this paper presents a novel method of classifying skin cancer in clinical skin images. The method consists of four blocks. First, class rebalancing is applied to the images of seven skin cancer types for better classification performance. Second, an image is preprocessed by being split into patches of the same size and then flattened into a series of tokens. Third, a transformer encoder is used to process the flattened patches. The transformer encoder consists of N identical layers with each layer containing two sublayers. Sublayer one is a multihead self-attention unit, and sublayer two is a fully connected feed-forward network unit. For each of the two sublayers, a normalization operation is applied to its input, and a residual connection of its input and its output is calculated. Finally, a classification block is implemented after the transformer encoder. The block consists of a flattened layer and a dense layer with batch normalization. Transfer learning is implemented to build the whole network, where the ImageNet dataset is used to pretrain the network and the HAM10000 dataset is used to fine-tune the network. Experiments have shown that the method has achieved a classification accuracy of 94.1%, outperforming the current state-of-the-art model IRv2 with soft attention on the same training and testing datasets. On the Edinburgh DERMOFIT dataset also, the method has better performance compared with baseline models.

show abstract

An improved transformer network for skin cancer classification

Cited by 101 publications

References 24 publications

On the Analyses of Medical Images Using Traditional Machine Learning Techniques and Convolutional Neural Networks

On the Analyses of Medical Images Using Traditional Machine Learning Techniques and Convolutional Neural Networks

Application of Artificial Intelligence to In Vitro Tumor Modeling and Characterization of the Tumor Microenvironment

A Novel Vision Transformer Model for Skin Cancer Classification

Contact Info

Product

Resources

About