Ammara Masood

Computer Aided Diagnostic Support System for Skin Cancer: A Review of Techniques and Algorithms

Image-based computer aided diagnosis systems have significant potential for screening and early detection of malignant melanoma. We review the state of the art in these systems and examine current practices, problems, and prospects of image acquisition, pre-processing, segmentation, feature extraction and selection, and classification of dermoscopic images. This paper reports statistics and results from the most important implementations reported to date. We compared the performance of several classifiers specifically developed for skin lesion diagnosis and discussed the corresponding findings. Whenever available, indication of various conditions that affect the techniques performance is reported. We suggest a framework for comparative assessment of skin cancer diagnostic models and review the results based on these models. The deficiencies in some of the existing studies are highlighted and suggestions for future research are provided.

Self-supervised learning model for skin cancer diagnosis

Automated diagnosis of skin cancer is an active area of research with different classification methods proposed so far. However, classification models based on insufficient labeled training data can badly influence the diagnosis process if there is no self-advising and semi supervising capability in the model. This paper presents a semi supervised, self-advised learning model for automated recognition of melanoma using dermoscopic images. Deep belief architecture is constructed using labeled data together with unlabeled data, and fine tuning done by an exponential loss function in order to maximize separation of labeled data. In parallel a self-advised SVM algorithm is used to enhance classification results by counteracting the effect of misclassified data. To increase generalization capability and redundancy of the model, polynomial and radial basis function based SA-SVMs and Deep network are trained using training samples randomly chosen via a bootstrap technique. Then the results are aggregated using least square estimation weighting. The proposed model is tested on a collection of 100 dermoscopic images. The variation in classification error is analyzed with respect to the ratio of labeled and unlabeled data used in the training phase. The classification performance is compared with some popular classification methods and the proposed model using the deep neural processing outperforms most of the popular techniques including KNN, ANN, SVM and semi supervised algorithms like Expectation maximization and transductive SVM.

Level Set Initialization Based on Modified Fuzzy C Means Thresholding for Automated Segmentation of Skin Lesions

Segmentation of skin lesion is an important step in the overall automated diagnostic systems used for early detection of skin cancer. Skin lesions can have various different forms which makes segmentation a difficult and complex task. Different methods are present in literature for improving results for skin lesion segmentation. Each method has some pros and cons and it is observed that none of them can be regarded as a generalized method working for all types of skin lesions. The paper proposes an algorithm that combines the advantages of clustering, thresholding and active contour methods currently being used independently for segmentation purposes. A modified algorithm for thresholding based on fusion of Fuzzy C mean clustering and histogram thresholding is applied to initialize level set automatically and also for estimating controlling parameters for level set evolution. The performance of level set segmentation is subject to appropriate initialization, so the proposed algorithm is being compared with some other state-of-the-art initialization methods. The work has been tested on clinical database of 270 images. Parameters for performance evaluation are presented in detail. Increased true detection rate and reduced false positive and false negative errors confirm the effectiveness of the proposed method for skin cancer detection.

Texture Analysis Based Automated Decision Support System for Classification of Skin Cancer Using SA-SVM

Early diagnosis of skin cancer is one of the greatest challenges due to lack of experience of general practitioners (GPs). This paper presents a clinical decision support system aimed to save lives, time and resources in the early diagnostic process. Segmentation, feature extraction, and lesion classification are the important steps in the proposed system. The system analyses the images to extract the affected area using a novel proposed segmentation method H-FCM-LS. A set of 45 texture based features is used. These underlying features which indicate the difference between melanoma and benign images are obtained through specialized texture analysis methods. For classification purpose, self-advising SVM is adapted which showed improved classification rate as compared to standard SVM. The diagnostic accuracy obtained through the proposed system is around 90% with sensitivity 91% and specificity 89%.

Automated segmentation of skin lesions: Modified Fuzzy C mean thresholding based level set method

Accurate segmentation of skin lesion can play a vital role in early detection of skin cancer. Taking the complexity and varieties of skin lesion images into consideration, we propose a new algorithm that combines the advantages of clustering, thresholding and active contour methods currently being used independently for segmentation purposes. A modified Fuzzy C mean thresholding algorithm is applied to initialize level set automatically and also for estimating controlling parameters for level set evolution. The performance of level set segmentation is subject to appropriate initialization, so the proposed initialization method is compared to some other state of the art initialization methods present in literature. The work has been tested on a clinical database of 238 images. Parameters for performance evaluation are presented in detail. Increased true detection rate and reduced false positive and false negative errors confirm the effectiveness of the proposed method for skin cancer detection.

Semi-advised learning model for skin cancer diagnosis based on histopathalogical images

Computer aided classification of skin cancer images is an active area of research and different classification methods has been proposed so far. However, the supervised classification models based on insufficient labeled training data can badly influence the diagnosis process. To deal with the problem of limited labeled data availability this paper presents a semi advised learning model for automated recognition of skin cancer using histopathalogical images. Deep belief architecture is constructed using unlabeled data by making efficient use of limited labeled data for fine tuning done the classification model. In parallel an advised SVM algorithm is used to enhance classification results by counteracting the effect of misclassified data using advised weights. To increase generalization capability of the model, advised SVM and Deep belief network are trained in parallel. Then the results are aggregated using least square estimation weighting. The proposed model is tested on a collection of 300 histopathalogical images taken from biopsy samples. The classification performance is compared with some popular methods and the proposed model outperformed most of the popular techniques including KNN, ANN, SVM and semi supervised algorithms like Expectation maximization algorithm and transductive SVM based classification model.

Adaptive Differential Evolution Based Feature Selection and Parameter Optimization for Advised SVM Classifier

This paper proposes a pattern recognition model for classification. Adaptive differential evolution based feature selection is used for dimensionality reduction and a new advised version of support vector machine is used for evaluation of selected features and for the classification. The tuning of the control parameters for differential evolution algorithm, parameter value optimization for support vector machine and selection of most relevant features form the datasets all are done together. This helps in dealing with their interdependent effect on the overall performance of the learning model. The proposed model is tested on some latest machine learning medical datasets and compared with some well-developed methods in literature. The proposed model provided quite convincing results on all the test datasets.

Differential evolution based advised SVM for histopathalogical image analysis for skin cancer detection.

Automated detection of cancerous tissue in histopathological images is a big challenge. This work proposed a new pattern recognition method for histopathological image analysis for identification of cancerous tissues. It comprised of feature extraction using a combination of wavelet and intensity based statistical features and autoregressive parameters. Moreover, differential evolution based feature selection is used for dimensionality reduction and an efficient self-advised version of support vector machine is used for evaluation of selected features and for the classification of images. The proposed system is trained and tested using a dataset of 150 histopathological images and showed promising comparative results with an average diagnostic accuracy of 89.1%.

SCALED CONJUGATE GRADIENT BASED DECISION SUPPORT SYSTEM FOR AUTOMATED DIAGNOSIS OF SKIN CANCER

Melanoma is the most deathful form of skin cancer but early diagnosis can ensure a high rate of survival. Early diagnosis is one of the greatest challenges due to lack of experience of general practitioners (GPs). This paper presents a clinical decision support system designed for the use of general practitioners, aiming to save time and resources in the diagnostic process. Segmentation, pattern recognition, and lesion detection are the important steps in the proposed decision support system. The system analyses the images to extract the affected area using a novel proposed segmentation method. It determinates the underlying features which indicate the difference between melanoma and benign images and makes a decision. Considering the efficiency of neural networks in classification of complex data, scaled conjugate gradient based neural network is used for classification. The presented work also considers analyzed performance of other efficient neural network training algorithms on the specific skin lesion diagnostic problem and discussed the corresponding findings. The best diagnostic rates obtained through the proposed decision support system are around 92%.

Integrating soft and hard threshold selection algorithms for accurate segmentation of skin lesion

Accurate segmentation of skin lesion is one of the most important step for automated diagnosis of skin cancer. Various characteristics of skin lesions and intensity variations in images can make it a highly challenging task. A new histogram analysis based fuzzy C mean thresholding method is presented here. It unifies the advantages of soft and hard thresholding algorithms along with reducing the computational complexity. Appropriate threshold value can be calculated even in the presence of abrupt intensity variations. This algorithm shows significantly improved performance for the segmentation of skin lesions. Experimental verification is done on a large set of skin lesion images having almost all types of expected artifacts that may badly affect the segmentation results. Performance evaluation is done by comparing the diagnosis results based on this method with other state of the art thresholding methods. Results show that the proposed approach performs reasonably well and can form a basis of expert diagnostic systems for skin cancer.