Euijoon Ahn

Automated saliency-based lesion segmentation in dermoscopic images.

The segmentation of skin lesions in dermoscopic images is considered as one of the most important steps in computer-aided diagnosis (CAD) for automated melanoma diagnosis. Existing methods, however, have problems with over-segmentation and do not perform well when the contrast between the lesion and its surrounding skin is low. Hence, in this study, we propose a new automated saliency-based skin lesion segmentation (SSLS) that we designed to exploit the inherent properties of dermoscopic images, which have a focal central region and subtle contrast discrimination with the surrounding regions. The proposed method was evaluated on a public dataset of lesional dermoscopic images and was compared to established methods for lesion segmentation that included adaptive thresholding, Chan-based level set and seeded region growing. Our results show that SSLS outperformed the other methods in regard to accuracy and robustness, in particular, for difficult cases.

Saliency-based Lesion Segmentation via Background Detection in Dermoscopic Images

The segmentation of skin lesions in dermoscopic images is a fundamental step in automated computer-aided diagnosis of melanoma. Conventional segmentation methods, however, have difficulties when the lesion borders are indistinct and when contrast between the lesion and the surrounding skin is low. They also perform poorly when there is a heterogeneous background or a lesion that touches the image boundaries; this then results in under- and oversegmentation of the skin lesion. We suggest that saliency detection using the reconstruction errors derived from a sparse representation model coupled with a novel background detection can more accurately discriminate the lesion from surrounding regions. We further propose a Bayesian framework that better delineates the shape and boundaries of the lesion. We also evaluated our approach on two public datasets comprising 1100 dermoscopic images and compared it to other conventional and state-of-the-art unsupervised (i.e., no training required) lesion segmentation methods, as well as the state-of-the-art unsupervised saliency detection methods. Our results show that our approach is more accurate and robust in segmenting lesions compared to other methods. We also discuss the general extension of our framework as a saliency optimization algorithm for lesion segmentation.

Dermoscopic Image Segmentation via Multistage Fully Convolutional Networks

Objective: Segmentation of skin lesions is an important step in the automated computer aided diagnosis of melanoma. However, existing segmentation methods have a tendency to over- or under-segment the lesions and perform poorly when the lesions have fuzzy boundaries, low contrast with the background, inhomogeneous textures, or contain artifacts. Furthermore, the performance of these methods are heavily reliant on the appropriate tuning of a large number of parameters as well as the use of effective preprocessing techniques, such as illumination correction and hair removal. Methods: We propose to leverage fully convolutional networks (FCNs) to automatically segment the skin lesions. FCNs are a neural network architecture that achieves object detection by hierarchically combining low-level appearance information with high-level semantic information. We address the issue of FCN producing coarse segmentation boundaries for challenging skin lesions (e.g., those with fuzzy boundaries and/or low difference in the textures between the foreground and the background) through a multistage segmentation approach in which multiple FCNs learn complementary visual characteristics of different skin lesions; early stage FCNs learn coarse appearance and localization information while late-stage FCNs learn the subtle characteristics of the lesion boundaries. We also introduce a new parallel integration method to combine the complementary information derived from individual segmentation stages to achieve a final segmentation result that has accurate localization and well-defined lesion boundaries, even for the most challenging skin lesions. Results: We achieved an average Dice coefficient of 91.18% on the ISBI 2016 Skin Lesion Challenge dataset and 90.66% on the PH2 dataset. Conclusion and Significance: Our extensive experimental results on two well-established public benchmark datasets demonstrate that our method is more effective than other state-of-the-art methods for skin lesion segmentation.

Automated skin lesion segmentation via image-wise supervised learning and multi-scale superpixel based cellular automata

Segmentation of skin lesions is considered as an important step in computer aided diagnosis (CAD) for automated melanoma diagnosis. Existing methods however have problems with over-or under-segmentation and do not perform well when a lesion is partially connected to the background or when the image contrast is low. To overcome these limitations, we propose a new automated skin lesion segmentation method via image-wise supervised learning (ISL) and multi-scale superpixel based cellular automata (MSCA). We propose using ISL to derive a probabilistic map for automated seeds selection, which removes the reliance on user-defined seeds as in conventional methods. The probabilistic map is then further used with the MSCA model for skin lesion segmentation. This map enables the inclusion of additional structural information and when compared to single-scale pixel-based CA model, it produces higher capacity to segment skin lesions with various sizes and contrast. We evaluated our method on two public skin lesion datasets and showed that it was more accurate and robust when compared to the state-of-the-art skin lesion segmentation methods.

Automatic melanoma detection via multi-scale lesion-biased representation and joint reverse classification

Dermoscopy image as a non-invasive diagnosis technique plays an important role for early diagnosis of malignant melanoma. Even for experienced dermatologists, however, diagnosis by human vision can be subjective, inaccurate and non-reproducible. This is attributed to the challenging image characteristics including varying lesion sizes and their shapes, fuzzy lesion boundaries, different skin color types and presence of hair. To aid in the image interpretation, automatic classification of dermoscopy images have been shown to be a valuable aid in the clinical decision making. Existing methods however have problems in representing and differentiating skin lesions due to high degree of similarities between melanoma and non-melanoma images and large variations inherited from skin lesion images. To overcome these limitations, this study proposes a new automatic melanoma detection method for dermoscopy images via multi-scale lesion-biased representation (MLR) and joint reverse classification (JRC). Our proposed MLR representation enable us to represent skin lesions using multiple closely related histograms derived from different rotations and scales while traditional methods can only represent skin lesion using a single-scale histogram. The MLR representation was then used with JRC for melanoma detection. The proposed JRC model allows us to use a set of closely related histograms to derive additional information for melanoma detection, where existing methods mainly rely on histogram itself. Our method was evaluated on a public dataset of dermoscopy images, and we demonstrate superior classification performance compared to the current state-of-the-art methods.

X-ray image classification using domain transferred convolutional neural networks and local sparse spatial pyramid

The classification of medical images is a critical step for imaging-based clinical decision support systems. Existing classification methods for X-ray images, however, generally represent the image using only local texture or generic image features (e.g. color or shape) derived from predefined feature spaces. This limits the ability to quantify the image characteristics using general data-derived features learned from image datasets. In this study we present a new algorithm to improve the performance of X-ray image classification, where we propose a late-fusion of domain transferred convolutional neural networks (DT-CNNs) with sparse spatial pyramid (SSP) features derived from a local image dictionary. Our method is robust as it exploits the rich generic information provided by the DT-CNNs and uses the specific local features and characteristics inherent in the X-ray images. Our method was evaluated on a public dataset of X-ray images and was compared to several state-of-the-art approaches. Experimental results show that our method was the most accurate for classification.

Semi-automatic skin lesion segmentation via fully convolutional networks

Segmentation of skin lesions is considered as an important step in computer aided diagnosis (CAD) for melanoma diagnosis. There have many attempts to segment skin lesions in a semi- or fully-automated manner. Existing methods, however, have problems with over- or under-segmentation and do not perform well with challenging skin lesions such as when a lesion is partially connected to the background or when image contrast is low. To overcome these limitations, we propose a new semi-automated skin lesion segmentation method that incorporates fully convolutional networks (FCNs) with multi-scale integration. We leverage the use of FCNs to derive high-level semantic information with simple user interaction e.g., a single click to accurately segment skin lesions of various complexity. Our experiments with 379 skin lesion images show that our proposed method achieves better segmentation results when compared to the state-of-the-art skin lesion segmentation methods for challenging skin lesions.

Step-wise integration of deep class-specific learning for dermoscopic image segmentation

Abstract The segmentation of abnormal regions on dermoscopic images is an important step for automated computer aided diagnosis (CAD) of skin lesions. Recent methods based on fully convolutional networks (FCN) have been very successful for dermoscopic image segmentation. However, they tend to overfit to the visual characteristics that are present in the dominant non-melanoma studies and therefore, perform poorly on the complex visual characteristics exhibited by melanoma studies, which usually consists of fuzzy boundaries and heterogeneous textures. In this paper, we propose a new method for automated skin lesion segmentation that overcomes these limitations via a novel deep class-specific learning approach which learns the important visual characteristics of the skin lesions of each individual class (melanoma vs. non-melanoma) on an individual basis. We also introduce a new probability-based, step-wise integration to combine complementary segmentation results derived from individual class-specific learning models. We achieved an average Dice coefficient of 85.66% on the ISBI 2017 Skin Lesion Challenge (SLC), 91.77% on the ISBI 2016 SLC and 92.10% on the PH2 datasets with corresponding Jaccard indices of 77.73%, 85.92% and 85.90%, respectively, for the same datasets. Our experiments on three well-established public benchmark datasets demonstrate that our method is more effective than other state-of-the-art methods for skin lesion segmentation.

Development of a risk predictive scoring system to identify patients at risk of representation to emergency department: a retrospective population-based analysis in Australia

Objective To examine the characteristics of frequent visitors (FVs) to emergency departments (EDs) and develop a predictive model to identify those with high risk of a future representations to ED among younger and general population (aged ≤70 years). Design and setting A retrospective analysis of ED data targeting younger and general patients (aged ≤70 years) were collected between 1 January 2009 and 30 June 2016 from a public hospital in Australia. Participants A total of 343 014 ED presentations were identified from 170 134 individual patients. Main outcome measures Proportion of FVs (those attending four or more times annually), demographic characteristics (age, sex, indigenous and marital status), mode of separation (eg, admitted to ward), triage categories, time of arrival to ED, referral on departure and clinical conditions. Statistical estimates using a mixed-effects model to develop a risk predictive scoring system. Results The FVs were characterised by young adulthood (32.53%) to late-middle (26.07%) aged patients with a higher proportion of indigenous (5.7%) and mental health-related presentations (10.92%). They were also more likely to arrive by ambulance (36.95%) and leave at own risk without completing their treatments (9.8%). They were also highly associated with socially disadvantage groups such as people who have been divorced, widowed or separated (12.81%). These findings were then used for the development of a predictive model to identify potential FVs. The performance of our derived risk predictive model was favourable with an area under the receiver operating characteristic (ie, C-statistic) of 65.7%. Conclusion The development of a demographic and clinical profile of FVs coupled with the use of predictive model can highlight the gaps in interventions and identify new opportunities for better health outcome and planning.