Detection of malignant skin diseases based on the lesion segmentation
03 Apr 2014-pp 382-386
TL;DR: A novel method to automatically detect the malignancy of skin diseases using conventional camera images is proposed, which mainly aims at an early diagnosis of the malignant diseases since they can be cured if detected early.
Abstract: A novel method to automatically detect the malignancy of skin diseases using conventional camera images is proposed. The procedure used would be of great advantage to the dermatologists as a pre-screening system for early diagnosis in situations where the dermoscopes are not accessible. This algorithm mainly aims at an early diagnosis of the malignant diseases since they can be cured if detected early. The proposed method works on color images by taking the HSV component and preprocessing was performed. A robust segmentation procedure is performed for the accurate detection of the lesion. For detection purpose, the morphological features like asymmetry, border irregularity, color variation and diameter are used. These extracted features helps to identify the malignant lesions from the non-malignant ones.
Citations
More filters
28 Dec 2015
TL;DR: A new approach of histogram thresholding for detailed segmentation of skin lesions based on histogram analysis of the saturation color component in the hue-saturation-value (HSV) color space is presented.
Abstract: Detailed segmentation of pigmented skin lesions is an important requirement in computer aided applications for melanoma assessment. In particular, accurate segmentation is necessary for image-guided evaluation of skin lesions characteristics. In this paper, we present a new approach of histogram thresholding for detailed segmentation of skin lesions based on histogram analysis of the saturation color component in the hue-saturation-value (HSV) color space. The proposed technique is specifically developed with the aim to handle the complex variability of features for macroscopic color images taken in uncontrolled environment. A dataset of 30 cases with manual segmentation was used for evaluation. We compare our results with two of most important existing segmentation techniques. For similarity report between automatic and manual segmentation we used dice similarity coefficient (DSC), the true detection rate (TDR), and the false positive rate (FPR). Experimental results show that the proposed method has high precision and low computational complexity.
11 citations
Cites background from "Detection of malignant skin disease..."
...There are a wide range of approaches in the existing literature for skin lesions segmentation for macroscopic images [3], [4], [5], [6], [7], [8], [9], [10], [11] and [12]....
[...]
TL;DR: A diagnosis system based on the techniques of image processing and data mining will be constructed, which will compare the captured image with training dataset using image processing techniques and decides whether a skin suffers from diseases or not using decision tree.
Abstract: Skin diseases are most common form of infections occurring in people of all ages. As the costs of dermatologists to monitor every patient is very high, there is a need for a computerized system to evaluate patient’s risk of skin disease using images of their skin lesions. We will be constructing a diagnosis system based on the techniques of image processing and data mining. The procedure would be of great advantage to the dermatologists as a pre-screening system for early diagnosis in situations where the dermoscopes are not accessible. The proposed system will capture image through smart-phone camera. Preprocessing and segmentation will be performed on each image. Then Feature extraction is done on skin lesion Feature Extraction is very important for Predictive modeling applications. Feature extraction in image Processing is a method of capturing visual content of images for indexing and retrieval. Primitive image features can be either General features, such as extraction of color, texture and shape or Domain specific features. After feature extraction, feature classification can be done. In Feature Extraction, the system will compare the captured image with training dataset using image processing techniques and decides whether a skin suffers from diseases or not using decision tree. If there is disease, then the system will give medical advice through Android application
9 citations
TL;DR: A comprehensive literature survey and review of a total of 594 publications (356 for skin lesion segmentation and 238 for skin lesions classification) published between 2011 and 2022 is provided in this article .
7 citations
TL;DR: Self-learning annotation scheme was proposed in the two-stage deep learning algorithm which consists of U-Net segmentation model with the annotation scheme and CNN classifier model for implementing the proposed self-learning Artificial Intelligence (AI) framework.
Abstract: Image segmentation and classification are the two main fundamental steps in pattern recognition. To perform medical image segmentation or classification with deep learning models, it requires training on large image dataset with annotation. The dermoscopy images (ISIC archive) considered for this work does not have ground truth information for lesion segmentation. Performing manual labelling on this dataset is time-consuming. To overcome this issue, self-learning annotation scheme was proposed in the two-stage deep learning algorithm. The two-stage deep learning algorithm consists of U-Net segmentation model with the annotation scheme and CNN classifier model. The annotation scheme uses a K-means clustering algorithm along with merging conditions to achieve initial labelling information for training the U-Net model. The classifier models namely ResNet-50 and LeNet-5 were trained and tested on the image dataset without segmentation for comparison and with the U-Net segmentation for implementing the proposed self-learning Artificial Intelligence (AI) framework. The classification results of the proposed AI framework achieved training accuracy of 93.8% and testing accuracy of 82.42% when compared with the two classifier models directly trained on the input images.
5 citations
Cites methods from "Detection of malignant skin disease..."
...area, perimeter, minor axis, major axis and colour information for measuring the ABCD parameter [15] of the dermatology such as Asymmetry, Border irregularity, Colour variation and Diameter for proper clinical diagnosis and treatment....
[...]
TL;DR: In this paper, a self-learning annotation scheme was proposed in the two-stage deep learning algorithm consisting of U-Net segmentation model with the annotation scheme and CNN classifier model.
Abstract: Image segmentation and classification are the two main fundamental steps in pattern recognition. To perform medical image segmentation or classification with deep learning models, it requires training on large image dataset with annotation. The dermoscopy images (ISIC archive) considered for this work does not have ground truth information for lesion segmentation. Performing manual labelling on this dataset is time-consuming. To overcome this issue, self-learning annotation scheme was proposed in the two-stage deep learning algorithm. The two-stage deep learning algorithm consists of U-Net segmentation model with the annotation scheme and CNN classifier model. The annotation scheme uses a K-means clustering algorithm along with merging conditions to achieve initial labelling information for training the U-Net model. The classifier models namely ResNet-50 and LeNet-5 were trained and tested on the image dataset without segmentation for comparison and with the U-Net segmentation for implementing the proposed self-learning Artificial Intelligence (AI) framework. The classification results of the proposed AI framework achieved training accuracy of 93.8% and testing accuracy of 82.42% when compared with the two classifier models directly trained on the input images.
2 citations
References
More filters
TL;DR: A systematic overview of the recent border detection methods in the literature paying particular attention to computational issues and evaluation aspects is presented.
425 citations
Patent•
22 Jan 1992TL;DR: In this paper, a look-up table is used to perform color transformation or color retouch operation on multibit digital data defining P color components of image pixels, which is addressed by N most significant bits (MSBS) of each color component value and which contains coarse versions of output pixel color component values.
Abstract: Apparatus for performing a color transformation or color retouch operation on multibit digital data defining P color components of image pixels. The apparatus comprises a look-up table (2) which is addressed by N most significant bits (MSBS) of each color component value and which contains coarse versions of output pixel color component values. An interpolator (3) is responsive to M least significant bits (LSBS) of each color component value to interpolate fine versions of the output pixel values from the coarse output pixel values obtained from the look-up table. A processor loads coarse output pixel data for each color component into the 2NP addresses of the look-up table which may be addressed by the M MSBs of the input pixel data. The processor (8) is adapted to cause the value of N progressively to increase with successive iterations.
202 citations
TL;DR: The preliminary experimental results are promising, and suggest that the method can achieve a classification accuracy of 96.71%, which is significantly better than the accuracy of comparable methods available in the literature.
142 citations
01 Nov 2011
TL;DR: An automatic method for segmenting skin lesions in conventional macroscopic images is presented, where stochastic region merging is initialized first on a pixel level, and subsequently on a region level until convergence until convergence, and overall segmentation error is lower than that achieved by existing methods.
Abstract: An automatic method for segmenting skin lesions in conventional macroscopic images is presented. The images are acquired with conventional cameras, without the use of a dermoscope. Automatic segmentation of skin lesions from macroscopic images is a very challenging problem due to factors such as illumination variations, irregular structural and color variations, the presence of hair, as well as the occurrence of multiple unhealthy skin regions. To address these factors, a novel iterative stochastic region-merging approach is employed to segment the regions corresponding to skin lesions from the macroscopic images, where stochastic region merging is initialized first on a pixel level, and subsequently on a region level until convergence. A region merging likelihood function based on the regional statistics is introduced to determine the merger of regions in a stochastic manner. Experimental results show that the proposed system achieves overall segmentation error of under 10% for skin lesions in macroscopic images, which is lower than that achieved by existing methods.
114 citations
15 Jul 2012
TL;DR: The system presented is a machine intervention in contrast to human arbitration into the conventional medical personnel based ideology of dermatological diagnosis and works on two dependent steps - the first detects skin anomalies and the latter identifies the diseases.
Abstract: This paper presents an automated dermatological diagnostic system. Etymologically, dermatology is the medical discipline of analysis and treatment of skin anomalies. The system presented is a machine intervention in contrast to human arbitration into the conventional medical personnel based ideology of dermatological diagnosis. The system works on two dependent steps - the first detects skin anomalies and the latter identifies the diseases. The system operates on visual input i.e. high resolution color images and patient history. In terms of machine intervention, the system uses color image processing techniques, k-means clustering and color gradient techniques to identify the diseased skin. For disease classification, the system resorts to feedforward backpropagation artificial neural networks. The system exhibits a diseased skin detection accuracy of 95.99% and disease identification accuracy of 94.016% while tested for a total of 2055 diseased areas in 704 skin images for 6 diseases.
75 citations