Skin Disease Datasets Research Articles

Boosting few-shot rare skin disease classification via self-supervision and distribution calibration.

Due to the difficulty in obtaining clinical samples and the high cost of labeling, rare skin diseases are characterized by data scarcity, making training deep neural networks for classification challenging. In recent years, few-shot learning has emerged as a promising solution, enabling models to recognize unseen disease classes by limited labeled samples. However, most existing methods ignored the fine-grained nature of rare skin diseases, resulting in poor performance when generalizing to highly similar classes. Moreover, the distributions learned from limited labeled data are biased, severely impairing the model's generalizability. This paper proposes a self-supervision distribution calibration network (SS-DCN) to address the above issues. Specifically, SS-DCN adopts a multi-task learning framework during pre-training. By introducing self-supervised tasks to aid in supervised learning, the model can learn more discriminative and transferable visual representations. Furthermore, SS-DCN applied an enhanced distribution calibration (EDC) strategy, which utilizes the statistics of base classes with sufficient samples to calibrate the bias distribution of novel classes with few-shot samples. By generating more samples from the calibrated distribution, EDC can provide sufficient supervision for subsequent classifier training. The proposed method is evaluated on three public skin disease datasets(i.e., ISIC2018, Derm7pt, and SD198), achieving significant performance improvements over state-of-the-art methods.

Masked autoencoders with generalizable self-distillation for skin lesion segmentation.

In the field of skin lesion image segmentation, accurate identification and partitioning of diseased regions is of vital importance for in-depth analysis of skin cancer. Self-supervised learning, i.e., MAE, has emerged as a potent force in the medical imaging domain, which autonomously learns and extracts latent features from unlabeled data, thereby yielding pre-trained models that greatly assist downstream tasks. To encourage pre-trained models to more comprehensively learn the global structural and local detail information inherent in dermoscopy images, we introduce a Teacher-Student architecture, named TEDMAE, by incorporating a self-distillation mechanism, it learns holistic image feature information to improve the generalizable global knowledge learning of the student MAE model. To make the image features learned by the model suitable for unknown test images, two optimization strategies are, Exterior Conversion Augmentation (EC) utilizes random convolutional kernels and linear interpolation to effectively transform the input image into one with the same shape but altered intensities and textures, while Dynamic Feature Generation (DF) employs a nonlinear attention mechanism for feature merging, enhancing the expressive power of the features, are proposed to enhance the generalizability of global features learned by the teacher model, thereby improving the overall generalization capability of the pre-trained models. Experimental results from the three public skin disease datasets, ISIC2019, ISIC2017, and PH indicate that our proposed TEDMAE method outperforms several similar approaches. Specifically, TEDMAE demonstrated optimal segmentation and generalization performance on the ISIC2017 and PH datasets, with Dice scores reaching 82.1% and 91.2%, respectively. The best Jaccard values were 72.6% and 84.5%, while the optimal HD95% values were 13.0% and 8.9%, respectively.

Enhancing Skin Disease Classification: A Novel Approach With Tailored Loss Functions And SMOTE Sumeet Ghumare

Although skin disease like skin cancer is a common health problem worldwide, early identification is crucial to its recovery. By using medical photos, the rise of artificial intelligence (AI) has transformed the detection of skin diseases. Despite the reality that various deep learning models have been created with this goal in mind, class imbalance in multi class classification all poses a serious difficulty and imbalance among the classes. This work offers a novel strategy to reduce class imbalance in the categorization of skin diseases, leading to significant accuracy gains. One of the largest skin cancer datasets currently accessible, it consists of 33,106 dermoscopy photos from seven different kinds of skin diseases as a starting point. The main goal is to enhance skin disease classification algorithms’ performance on this imbalance dataset. We are going to use a two-pronged strategy to combat the class disparity and GAN (Generative adversarial network) Based Oversampling or Meta-learning in order to successfully equalize the representation of each class during training, we first design a balanced Dynamic Mini-Batch Size logic, SMOTE (Synthetic Minority Oversampling Technique) and Augmenter. Second, we provide a new loss function designed to take into account the unique features of the dataset. The outcomes are startling. These findings underscore the efficacy of our hybrid approach in training deep convolutional neural networks for imbalanced skin disease datasets. By combining data-level balancing techniques with a carefully designed loss function at the algorithm level, we address the challenge of among the classes, paving the way for more accurate and reliable skin cancer diagnoses. Key Words: Dermoscopy images, DenseNet169, Convolution neural network, SMOTE (Synthetic Minority Over-sampling Technique, GAN(Generative adversarial network), Mini-Batch Logic, Augmenter

A novel deep learning framework based swin transformer for dermal cancer cell classification

Recent studies have demonstrated the efficacy of deep learning architectures in enhancing the interpretation of skin images, thereby aiding in the classification and segmentation of skin cancer. However, the existing deep learning techniques predominantly focus on either segmentation or classification and are designed for normally distributed data. Addressing this limitation, the present study introduces a hybrid ensemble deep learning approach for skin cancer classification and segmentation. This proposed model amalgamates Residual Learning Machines, Swin Transformers, and Fast Neural Networks (FNN) to proficiently manage diverse sets of non-uniformly distributed data, thereby augmenting diagnostic accuracy. The efficacy of this approach was evaluated using various skin cancer datasets, including ISIC-2008, PH-2, and HM007. The assessment involved several performance metrics, such as the Matthew correlation coefficient (MCC), recall, F1-score, specificity, accuracy, and precision. Moreover, to underscore the superiority of the proposed model, its performance was juxtaposed with that of previous efforts. Results from repeated trials indicate that the proposed model achieved 98.78% classification accuracy, 98.7% precision, 98.7% F1-score, 98.64% average recall, and an MCC of 0.9863 across different skin disease datasets. Demonstrating consistent superiority over existing methods, the proposed approach shows considerable potential in revolutionizing the diagnostics of skin cancer.

Enhancing the fairness of AI prediction models by Quasi-Pareto improvement among heterogeneous thyroid nodule population

Artificial Intelligence (AI) models for medical diagnosis often face challenges of generalizability and fairness. We highlighted the algorithmic unfairness in a large thyroid ultrasound dataset with significant diagnostic performance disparities across subgroups linked causally to sample size imbalances. To address this, we introduced the Quasi-Pareto Improvement (QPI) approach and a deep learning implementation (QP-Net) combining multi-task learning and domain adaptation to improve model performance among disadvantaged subgroups without compromising overall population performance. On the thyroid ultrasound dataset, our method significantly mitigated the area under curve (AUC) disparity for three less-prevalent subgroups by 0.213, 0.112, and 0.173 while maintaining the AUC for dominant subgroups; we also further confirmed the generalizability of our approach on two public datasets: the ISIC2019 skin disease dataset and the CheXpert chest radiograph dataset. Here we show the QPI approach to be widely applicable in promoting AI for equitable healthcare outcomes.

A Multi-Stage Faster RCNN-Based iSPLInception for Skin Disease Classification Using Novel Optimization.

Nowadays, skin cancer is considered a serious disorder in which early identification and treatment of the disease are essential to ensure the stability of the patients. Several existing skin cancer detection methods are introduced by employing deep learning (DL) to perform skin disease classification. Convolutional neural networks (CNNs) can classify melanoma skin cancer images. But, it suffers from an overfitting problem. Therefore, to overcome this problem and to classify both benign and malignant tumors efficiently, the multi-stage faster RCNN-based iSPLInception (MFRCNN-iSPLI) method is proposed. Then, the test dataset is used for evaluating the proposed model performance. The faster RCNN is employed directly to perform image classification. This may heavily raise computation time and network complications. So, the iSPLInception model is applied in the multi-stage classification. In this, the iSPLInception model is formulated using the Inception-ResNet design. For candidate box deletion, the prairie dog optimization algorithm is utilized. We have utilized two skin disease datasets, namely, ISIC 2019 Skin lesion image classification and the HAM10000 dataset for conducting experimental results. The methods' accuracy, precision, recall, and F1 score values are calculated, and the results are compared with the existing methods such as CNN, hybrid DL, Inception v3, and VGG19. With 95.82% accuracy, 96.85% precision, 96.52% recall, and 0.95% F1 score values, the output analysis of each measure verified the prediction and classification effectiveness of the method.

EfficientSkinDis: An EfficientNet-based classification model for a large manually curated dataset of 31 skin diseases

Skin diseases are a common health issue, affecting nearly one-third of the global population, but they are often underestimated in terms of their impact despite being highly visible. Accurately diagnosing skin diseases can be challenging sometimes. Fortunately, deep learning algorithms have shown great potential for a variety of tasks, including the diagnosis of skin diseases. This research presents a novel dataset of 31 skin diseases by blending two different datasets. Three different CNN types, which were EfficientNet, ResNet, and VGG with different architectures, were used for transfer learning on the skin disease dataset. EfficientNet had the highest testing accuracy, thus, it was further trained for fine-tuning. Initially, using the training split of 70%, the EfficientNet model achieved a testing accuracy of 71%. Therefore, as this was considered low, the 70% training split comprising 3,424 samples was further augmented. As a result, the model's accuracy improved to 72%. However, it was observed that the data split of 70/30 was not effective, thus the experiment was re-conducted with a train-test split of 80/20 and an improvement in accuracy score i.e., 74% was observed. To further improve the model's accuracy, augmentations were applied to the data and were able to achieve an accuracy of 87.15%. Finally, the trained model with the best accuracy was deployed on a Streamlit webserver. The proposed model can help society in diagnosing skin disease early, therefore it can be treated timely. The webserver of EfficientSkinDis can be accessed at (https://abdulrafay97-efficientskindis-app-ooncon.streamlit.app/).

RMMLP:Rolling MLP and matrix decomposition for skin lesion segmentation

Accurate skin lesion segmentation is of great significance for improving the quantitative analysis of skin cancer. However, due to the irregular shapes and ambiguous boundaries of lesions, automatic segmentation is still very difficult. Although the emergence of CNN and Transformer hybrid models recently complements their respective shortcomings, the use of transformer results in heavy parameters, and the complex structure requires a lot of data and resources for training. In contrast, Multi-Layer Perceptron (MLP) is a cost-effective alternative to complex self-attention. To this end, we propose an MLP-based model using matrix decomposition and rolling tensors for skin lesion segmentation. Specifically, we can dynamically calculate the correlation matrix according to the size of the input image and use it as the weight to guide the segmentation. The way of rolling tensors can fully mix the feature information and sequentially extract the information under different receptive fields. We also employ a two-layer decoding structure using matrix decomposition to fuse the feature extracted in the parallel MLP encoders. Extensive experiments are conducted on the public skin disease datasets ISIC2016, ISIC2017, ISIC2018 and PH2. After comparison, we found that although the addition of Transformer makes up for the lack of global information on CNN, the complex structure of Transformer itself will lead to huge parameters and is difficult to train. The experimental results show that we achieved higher results with fewer parameters compared to other state-of-the-art models which mean that the self-attention mechanism of Transformer is not irreplaceable. Our code is released at https://github.com/Lingglesymphony/RMMLP.

An improved hierarchical variational autoencoder for cell-cell communication estimation using single-cell RNA-seq data.

Analysis of cell-cell communication (CCC) in the tumor micro-environment helps decipher the underlying mechanism of cancer progression and drug tolerance. Currently, single-cell RNA-Seq data are available on a large scale, providing an unprecedented opportunity to predict cellular communications. There have been many achievements and applications in inferring cell-cell communication based on the known interactions between molecules, such as ligands, receptors and extracellular matrix. However, the prior information is not quite adequate and only involves a fraction of cellular communications, producing many false-positive or false-negative results. To this end, we propose an improved hierarchical variational autoencoder (HiVAE) based model to fully use single-cell RNA-seq data for automatically estimating CCC. Specifically, the HiVAE model is used to learn the potential representation of cells on known ligand-receptor genes and all genes in single-cell RNA-seq data, respectively, which are then utilized for cascade integration. Subsequently, transfer entropy is employed to measure the transmission of information flow between two cells based on the learned representations, which are regarded as directed communication relationships. Experiments are conducted on single-cell RNA-seq data of the human skin disease dataset and the melanoma dataset, respectively. Results show that the HiVAE model is effective in learning cell representations, and transfer entropy could be used to estimate the communication scores between cell types.

Hybrid optimization based deep neuro fuzzy network for skin cancer detection

SummaryMelanoma is the most prevalent type of skin cancer, affecting skin surface cells. The risk features of the disease are minimized by finding the disease at time or in the initial stage. Even through, various skin cancer detection methods are developed, detecting the disease using discriminative features still results a challenging task. An efficient detection strategy based on the proposed Lion Cat Swarm Optimization‐based Deep Neuro Fuzzy Network (LCSO‐based DNFN) is created to identify skin cancer in its early stages. The LCSO is the integration of the lion optimization algorithm (LOA) and cat swarm optimization (CSO) algorithm. To make treatment measure at right time increases the probability of survival and so detecting the disease using deep learning classifier is performed with hybrid optimization algorithm. The segmentation result is effective through the fusion model and it ensures to achieve accurate detection process. Rather than extracting the features, the proposed method generates optimal result through data augmentation. The performance of the proposed LCSO‐based DNFN is determined using skin disease dataset, and the introduced approach obtained outstanding performance with the measures of accuracy, sensitivity, and specificity with the values of 93.10%, 92.95%, and 92.87%, respectively.

A multimodal transformer to fuse images and metadata for skin disease classification.

Skin disease cases are rising in prevalence, and the diagnosis of skin diseases is always a challenging task in the clinic. Utilizing deep learning to diagnose skin diseases could help to meet these challenges. In this study, a novel neural network is proposed for the classification of skin diseases. Since the datasets for the research consist of skin disease images and clinical metadata, we propose a novel multimodal Transformer, which consists of two encoders for both images and metadata and one decoder to fuse the multimodal information. In the proposed network, a suitable Vision Transformer (ViT) model is utilized as the backbone to extract image deep features. As for metadata, they are regarded as labels and a new Soft Label Encoder (SLE) is designed to embed them. Furthermore, in the decoder part, a novel Mutual Attention (MA) block is proposed to better fuse image features and metadata features. To evaluate the model’s effectiveness, extensive experiments have been conducted on the private skin disease dataset and the benchmark dataset ISIC 2018. Compared with state-of-the-art methods, the proposed model shows better performance and represents an advancement in skin disease diagnosis.

FusionM4Net: A multi-stage multi-modal learning algorithm for multi-label skin lesion classification.

Skin disease is one of the most common diseases in the world. Deep learning-based methods have achieved excellent skin lesion recognition performance, most of which are based on only dermoscopy images. In recent works that use multi-modality data (patient's meta-data, clinical images, and dermoscopy images), the methods adopt a one-stage fusion approach and only optimize the information fusion at the feature level. These methods do not use information fusion at the decision level and thus cannot fully use the data of all modalities. This work proposes a novel two-stage multi-modal learning algorithm (FusionM4Net) for multi-label skin diseases classification. At the first stage, we construct a FusionNet, which exploits and integrates the representation of clinical and dermoscopy images at the feature level, and then uses a Fusion Scheme 1 to conduct the information fusion at the decision level. At the second stage, to further incorporate the patient's meta-data, we propose a Fusion Scheme 2, which integrates the multi-label predictive information from the first stage and patient's meta-data information to train an SVM cluster. The final diagnosis is formed by the fusion of the predictions from the first and second stages. Our algorithm was evaluated on the seven-point checklist dataset, a well-established multi-modality multi-label skin disease dataset. Without using the patient's meta-data, the proposed FusionM4Net's first stage (FusionM4Net-FS) achieved an average accuracy of 75.7% for multi-classification tasks and 74.9% for diagnostic tasks, which is more accurate than other state-of-the-art methods. By further fusing the patient's meta-data at FusionM4Net's second stage (FusionM4Net-SS), the entire FusionM4Net finally boosts the average accuracy to 77.0% and the diagnostic accuracy to 78.5%, which indicates its robust and excellent classification performance on the label-imbalanced dataset. The corresponding code is available at: https://github.com/pixixiaonaogou/MLSDR.

Enriched Optimization Algorithm for Effective Skin Disease Prediction Using Soft Computing Techniques

In recent years, a usual worldwide problem is a skin disease—the diagnostics of infection and skin disease prediction based on the data mining techniques. The precise and cost-effective treatments obtain a technologybased data mining system that can consider making the right decision. Depends on data, there are 34 UCI datasets have in the skin disease prediction. All of the datasets are not much important when predicting the skin disease problem. In this study, the essential datasets to be analyzed because they only give the best accuracy in skin disease prediction. For an outstanding selection of allocation, to propose a novel feature selection approaches, Enriched Fruitfly Optimization Algorithm (EFOA), and Ensemble Classifiers that are helps for an early stage of skin disease prediction. A hybrid technique through the three essential hybrid feature selection approaches such as Chi-Square method, Information Gain method, and Principal Component Analysis (PCA) methods that are combined for better feature selection results. Based on the skin disease dataset, the resultant feature selection approach generated the reduced data subset. Then, the Enriched Fruitfly Optimization Algorithm (EFOA) offers the optimization of reduced data subset. Here, the accuracy estimation is the vital factor to optimize the effective and best prediction of skin disease affected regions. Afterward, the classification performs to classify the EFOA based optimized result by using the six different classification methods. Where, the classification helps to analyze the optimized results, which offers the better classification procedure. To predict the base learner’s performance, to utilize the Naive Bayesian, K-Nearest Neighbour, Decision Tree, Support Vector Machine, Random Forest, and Multilayer Perceptron (MLP) to classify the optimized result. Then, the ensemble techniques used to analyze the classifier’s results through the 3 different methods like Bagging, Boosting, Stacking, added on the base learners to improve the proposed work. Based on the performance, the base learners’ performance is larger than the input dataset. The base learner’s parameters are essential to calculate the accuracy of skin disease prediction performance. The performance of the proposed method will take and compare to each base learner, and the performance shows the accurate skin disease prediction improvement with other existing methods.

A Study on Multiple Factors Affecting the Accuracy of Multiclass Skin Disease Classification

Diagnosis of skin diseases by human experts is a laborious task prone to subjective judgment. Aided by computer technology and machine learning, it is possible to improve the efficiency and robustness of skin disease classification. Deep transfer learning using off-the-shelf deep convolutional neural networks (CNNs) has huge potential in the automation of skin disease classification tasks. However, complicated architectures seem to be too heavy for the classification of only a few skin disease classes. In this paper, in order to study potential ways to improve the classification accuracy of skin diseases, multiple factors are investigated. First, two different off-the-shelf architectures, namely AlexNet and ResNet50, are evaluated. Then, approaches using either transfer learning or trained from scratch are compared. In order to reduce the complexity of the network, the effects of shortening the depths of deep CNNs are investigated. Furthermore, different data augmentation techniques based on basic image manipulation are compared. Finally, the choice of mini-batch size is studied. Experiments were carried out on the HAM10000 skin disease dataset. The results show that the ResNet50-based model is more accurate than the AlexNet-based model. The transferred knowledge from the ImageNet database helps to improve the accuracy of the model. The reduction in stages of the ResNet50-based model can reduce complexity while maintaining good accuracy. Additionally, the use of different types of data augmentation techniques and the choice of mini-batch size can also affect the classification accuracy of skin diseases.

Intelligent System for Skin Disease Prediction using Machine Learning

Skin is an extraordinary human structure. It frequently suffered from many known and unknown disease. Therefore, diagnosis of human skin diseases is the most uncertain and complicated branch of science. It has been observed that most of the cases remain unnoticed because of the lack of better medical infrastructure and facilities. This paper is devoted to solve this challenge. Therefore, this paper effectively proposed (CNN-SVM-MAA) system which combines Convolutional Neural Network with Support Vector Machine classifier to develop a Mobile Android Application. Thus, to evaluate the performance of the proposed system several experiments are conducted on our dataset. This dataset consists around 3000 images which collected from a lot of sources like Beni-Suef University Hospital, Cairo University Hospital and various websites as well to be more accurate and realistic. A comparative study of applying different Feature extraction algorithms with different classifiers was accomplished. The results obtained showed the adequacy of the proposed (CNN-SVM-MAA) system how many skin diseases images have been detected from skin disease dataset. Which lead to detect skin disease and provide the user with the disease name and treatment related prescription with high accuracy.

Rider Chicken Optimization Algorithm-Based Recurrent Neural Network for Big Data Classification in Spark Architecture

Abstract This paper proposes an effective classification method named Rider Chicken Optimization Algorithm-based Recurrent Neural Network (RCOA-based RNN) to perform big data classification in spark architecture. Initially, the input data are collected from the network by the master node and then forwarded to the slave node. These nodes are responsible for storing the data and performing computations. The features are effectively selected in the slave node using the proposed RCOA. The selected features are forwarded to the master node. The big data classification is achieved in the master node by using the RNN classifier, and the training of the classifier is done using the proposed RCOA algorithm, which is the integration of the Rider optimization algorithm (ROA) with the standard Chicken Swarm Optimization (CSO). The experimentation is done by using the Switzerland dataset, Cleveland dataset, Hungarian dataset and Skin disease dataset, in which the proposed RCOA-based RNN attained better performance based on the quantitative properties, such as sensitivity, accuracy and specificity with the values of 9.3E+01%, 9.4E+01% and 9.3E+01% using Hungarian dataset. The existing learning methods failed to address the complex classification problems at a reasonable time, which is overcome by the proposed method.

Oppositional Butterfly Optimization Algorithm with Multilayer Perceptron for Medical Data Classification

Medical data classification can be assumed to be a crucial process in the domain of medical informatics. Generally, medical data comprises a set of medical records and literature which are considered as the essential healthcare data sources. But the existence of medical data includes complicated medical vocabulary and medical metrics makes the classification process challenging. Though several models are available in the literature, there is still needed to improve the classification performance. In this view, this paper devises a novel oppositional based learning with butterfly optimization algorithm (OBLBOA)and multilayer perceptron (MLP) called OBLBOA-MLP for medical data classification. The presented OBLBOA-MLP model involves three stages of operation such as preprocessing, classification, and parameter tuning. Primarily, data preprocessing is carried out to remove the unwanted data and raise the data quality to a certain extension. Besides, MLP model is applied as a classifier to determine the existence of the diseases. In addition, OBLBOA is employed for the hyperparameter optimization of the MLP model. The application of OBL helps to increase the performance of the BOA. A detailed set of simulation analysis was performed to determine the appropriate detection results of the OBLBOA-MLP model. The obtained experimental values pointed out the improved classification performance by attaining a higher accuracy of 98.23% and 92.67% on the applied CKD and skin disease dataset respectively.

Level set approach based on Parzen Window and floor of log for edge computing object segmentation in digital images

Scientific research into methodologies and algorithms to improve support for medical diagnoses remains in high demand on the agenda. Computer-aided diagnostic systems that use the Internet of Things (IoT) enable greater accessibility and integration between patients and specialists. One of the steps covered by IoT systems is the segmentation of Regions of Interest (ROI) in digital images. The challenge to segment these ROI in IoT systems is to apply methods that have low computational and storage costs with reliable results. Thus this work proposes the use of edge computing with a new approach based on active contours to target the ROI in medical images called FLog Parzen Level Set (FPLS). The method can be divided into two stages. First is the initialization of the seed point using Parzen Window and clusterization using Floor of Log. Second the growth and refinement of the region with probabilistic estimates of the regional contour using the Parzen Window. The proposed method was evaluated using the metrics of Accuracy, Precision, Sensitivity, Specificity, Matthews Coefficient Correlation, Hausdorff distance, Dice, and Jaccard Similarity Coefficient. The stable and satisfactory results can be highlighted despite the low computational times and costs. The proposed method was submitted to stroke, lung, and skin disease datasets. The proposed method achieved the fastest mean segmentation time of 1.64s for the three datasets used in this work. The method obtained the highest values for Sensitivity (98.57%), Accuracy (98.77%) and MCC (94.73%) in the stroke dataset, the lowest value for Hausdorff distance (4.24) in the lung dataset, and the highest Dice Coefficient value (92.49%) in the skin dataset. In conclusion, the proposed method is a promising tool for an edge computing system that segments regions of interest with the high precision of a level set and a fast convergence rate.

Temperature network for few-shot learning with distribution-aware large-margin metric

Few-shot learning learns to classify unseen data with few training samples in hand and has attracted increasing attentions recently. In this paper, we propose a novel Temperature Network to tackle few-shot learning tasks motivated by three crucial factors that are seldom considered in the existing literature. First, to encourage compact intra-class distribution, a general improvement for prototype-based methods is proposed to ensure compact intra-class distribution and the effectiveness is theoretically and experimentally validated. Second, the proposed Temperature Network can implicitly generate query-specific prototypes and thus enjoys a more effective distribution-aware metric. Third, to further strengthen the generalization ability of the proposed model, a novel and simple large-margin based method is developed by leveraging the temperature function and we gradually tune the learning temperature to stabilize the training process. Moreover, we note that the commonly used datasets in few-shot learning are actually contrived from large-scale datasets, and thus may not represent a real few-shot problem. We propose a real-life few shot problem, i.e., Dermnet skin disease, to comprehensively evaluate the performance of few-shot learning methods. Experiments conducted on conventional datasets as well as the proposed skin disease dataset demonstrate the superiority of the proposed method over other state-of-the-art methods. The source code of our method is available.11https://github.com/zwvews/TemperatureNetwork.git

Auxiliary Decoder and Classifier for Imbalanced Skin Disease Diagnosis

To date, deep learning has been widely adopted in medical diagnosis systems and made great achievements in real-world applications. However, in medical image-based intelligent diagnosis, the phenomenon of class imbalance often appears due to the substantially smaller available training data for rare diseases compared to common diseases, which usually degrade the performance of classification dramatically. In this paper, we propose a novel learning framework to effectively alleviate the impact of class imbalance, adding an auxiliary decoder to reconstruct original images and reusing the original CNN classifier to help the classifier more likely extract disease-relevant features for both rare and common diseases. Throughout experiments on two skin disease datasets support that the proposed framework outperforms strong baselines with a very explicit margin. The proposed method is also independent of network architecture so that it can be flexibly combined with different model structures. Besides, applying our proposed method with existing training strategies designed for class imbalance can still improve the classification performance.