Search

The interest in image dermoscopy has been significantly increased recently and skin lesion images are nowadays routinely acquired for a number of skin disorders. An important finding in the assessment of a skin lesion severity is the existence of dark dots and globules, which are hard to locate and count using existing image software tools. In this work we present a novel methodology for detecting/segmenting and count dark dots and globules from dermoscopy images. Segmentation is performed using a multi-resolution approach based on inverse non-linear diffusion. Subsequently, a number of features are extracted from the segmented dots/globules and their diagnostic value in automatic classification of dermoscopy images of skin lesions into melanoma and non-malignant nevus is evaluated. The proposed algorithm is applied to a number of images with skin lesions with known histo-pathology. Results show that the proposed algorithm is very effective in automatically segmenting dark dots and globules. Furthermore, it was found that the features extracted from the segmented dots/globules can enhance the performance of classification algorithms that discriminate between malignant and benign skin lesions, when they are combined with other region-based descriptors. [ABSTRACT FROM AUTHOR]

Abstract: Taking into account that heart auscultation remains the dominant method for heart examination in the small health centers of the rural areas and generally in primary healthcare set-ups, the enhancement of this technique would aid significantly in the diagnosis of heart diseases. In this context, the present paper initially surveys the research that has been conducted concerning the exploitation of heart sound signals for automated and semi-automated detection of pathological heart conditions. Then it proposes an automated diagnosis system for the identification of heart valve diseases based on the Support Vector Machines (SVM) classification of heart sounds. This system performs a highly difficult diagnostic task (even for experienced physicians), much more difficult than the basic diagnosis of the existence or not of a heart valve disease (i.e. the classification of a heart sound as ‘healthy’ or ‘having a heart valve disease’): it identifies the particular heart valve disease. The system was applied in a representative global dataset of 198 heart sound signals, which come both from healthy medical cases and from cases suffering from the four most usual heart valve diseases: aortic stenosis (AS), aortic regurgitation (AR), mitral stenosis (MS) and mitral regurgitation (MR). Initially the heart sounds were successfully categorized using a SVM classifier as normal or disease-related and then the corresponding murmurs in the unhealthy cases were classified as systolic or diastolic. For the heart sounds diagnosed as having systolic murmur we used a SVM classifier for performing a more detailed classification of them as having aortic stenosis or mitral regurgitation. Similarly for the heart sounds diagnosed as having diastolic murmur we used a SVM classifier for classifying them as having aortic regurgitation or mitral stenosis. Alternative classifiers have been applied to the same data for comparison (i.e. back-propagation neural networks, k-nearest-neighbour and naïve Bayes classifiers), however their performance for the same diagnostic problems was lower than the SVM classifiers proposed in this work. [Copyright &y& Elsevier]

Evaluating the interpretability of Deep Learning models is crucial for building trust and gaining insights into their decision-making processes. In this work, we employ class activation map based attribution methods in a setting where only High-Resolution Class Activation Mapping (HiResCAM) is known to produce faithful explanations. The objective is to evaluate the quality of the attribution maps using quantitative metrics and investigate whether faithfulness aligns with the metrics results. We fine-tune pre-trained deep learning architectures over four medical image datasets in order to calculate attribution maps. The maps are evaluated on a threefold metrics basis utilizing well-established evaluation scores. Our experimental findings suggest that the Area Over Perturbation Curve (AOPC) and Max-Sensitivity scores favor the HiResCAM maps. On the other hand, the Heatmap Assisted Accuracy Score (HAAS) does not provide insights to our comparison as it evaluates almost all maps as inaccurate. To this purpose we further compare our calculated values against values obtained over a diverse group of models which are trained on non-medical benchmark datasets, to eventually achieve more responsive results. This study develops a series of experiments to discuss the connection between faithfulness and quantitative metrics over medical attribution maps. HiResCAM preserves the gradient effect on a pixel level ultimately producing high-resolution, informative and resilient mappings. In turn, this is depicted in the results of AOPC and Max-Sensitivity metrics, successfully identifying the faithful algorithm. In regards to HAAS, our experiments yield that it is sensitive over complex medical patterns, commonly characterized by strong color dependency and multiple attention areas. • A framework for evaluating interpretability methods, under faithfulness, is proposed. • The proposed algorithm is applied to four medical imaging datasets. • Area over Perturbation Curve and Max-Sensitivity favor the faithful method • Heatmap Assisted Accuracy Score fails to identify the faithful method • Heatmap Assisted Accuracy Score is sensitive over complex patterns characterizing medical images. [ABSTRACT FROM AUTHOR]