PDE-based unsupervised repair of hair-occluded information in dermoscopy images of melanoma
Introduction
Malignant melanoma (MM) has consistently had one of the most rapidly increasing incidence of all cancers, with 62,190 new cases and 7700 deaths estimated in the United States in 2006, and the incidence of MM increased 3–8% per year and doubled over the decade in China [1], [2], [3]. Early diagnosis is particularly important since melanoma can be cured with a simple excision if detected early [4], [5].
Dermoscopy is a non-invasive skin imaging technique that uses optical magnification and either liquid immersion and low angle-of-incidence lighting or cross-polarized lighting to make the contact area translucent, making subsurface structures more easily visible when compared to conventional macroscopic (clinical) images [2]. Dermoscopy allows the identification of dozens of morphological features such as pigment networks, dots/globules, streaks, blue-white areas, and blotches [2]. But due to the lack of reproducibility and subjectivity of human interpretation, the development of computerized image analysis techniques is of paramount importance.
Digital image analysis technology about skin melanoma is currently focusing on segmentation, feature extraction and classifier design. Grana et al. [6] used Otsu's threshold to automatically segment the melanoma image, then selected k points for spline interpolating to get smooth lesion border, finally produced statistical parameters about lesion based on ABCD rules. Taouil et al. [7] used top-hat operator as image preprocessor to eliminate noise and preserve lesion border information, then used Snake method to detect lesion border, which got better result than Otsu's threshold. Schmid [8] proposed a technique based color clustering. First, a 2D histogram is calculated from the first two principal components of the CIE L*u*v* color space. Then, the histogram is smoothed and initial cluster centers are determined from the peaks using a perceptron classifier. Finally, the lesion image is segmented using a modified version of the fuzzy c-means (FCM) clustering algorithm. On the classifying aspect, neural network diagnosis of skin tumors has previously been applied by classifying extracted features from digitized dermoscopy images of lesions [9]. The extracted features are based on geometry, colors, and texture of the lesions, involving complex image processing techniques. Kusumoputro and Ariyanto [10] used an artificial neural network to separate the malignant melanoma from benign categories of skin cancers based on cancer shapes and their relative color. Celebi et al. [2], [11] used the Statistical Region Merging (SRM) algorithm to detect the lesion border, and realized the classification of dermoscopy images based on Support vector machine using the shape, color and texture as input features. And he pointed out that the hairs in the image will cause to a bad segmentation result so as to error analysis.
The hair problem has not been fully addressed in the literature of computer image processing techniques for dermatological applications in spite of the rapid growth of this field [12]. It is an important issue, however, especially for designing automatic segmentation and feature extraction algorithms for clinical use. In a similar study conducted in Italy [13], the investigators decided to shave the hairs using a razor before the imaging session. This procedure not only adds extra costs and time to the imaging session, but it also is impractical if we want to apply the technology to total-body nevus imaging [14]. Lee et al. [15] located the hair through morphological closing operator first and then removed the hairs from image by replacing the hair pixels by the nearby non-hair pixels, but the algorithm worked well just for the thick and dark hairs.
In this paper a novel unsupervised hair removal algorithm is proposed through combining hair object extraction with image inpainting technology. Image inpainting is a kind of effectively method for both repairing the lost information or nicks in image and removing some specified object from image. Strictly speaking, current image inpainting methods can hardly implement automatic processing performance because some area information needs to be inputted by user in order to determine which area is to be repaired [16], [17]. In this paper the hairs are extracted based on elongate feature function first and then the information of occluded area is successfully repaired through the image inpainting technology. The experiment results show that the extracted hairs are exact and the repaired texture satisfies the requirement of medical diagnosis.
Section snippets
Segmentation based on closing-based top-hat
In Fig. 1, hairs are a kind of curve objects, and can be detected through detecting curves. Currently there are many methods of detecting line object. For example the road object detection from remote sensing image [18], [19], which includes Hough-transform-based methods, statistics-based methods, morphology-based detecting methods and so on, these kinds of methods treat a section of road as straight line segment generally, they locate the line object first, and then extract the line object
Hair extraction
From Fig. 2(c), with the non-hair noises the hairs are detected, and the connected regions belonging to hair are bigger and longer than the non-hair connected regions which are smaller and shorter. It is simple to take the length or area size of connected region as measure to extract hairs from image, but some short hairs will not be separated from those long non-hair connected regions. In document [15], for each pixel inside the hair region, at first, a check should be proceeded to ensure that
Image inpainting based on PDE
The influence of hair to diagnosis analysis can be eliminated through excluding the hair regions from further analysis. But for the image segmentation, it can improve the veracity of segmentation to repair the melanoma texture occluded by hairs.
Inpainting is the art of rebuilding the basics of visual art and consists of filling-in unknown data in a known region of an image, with the principal objective of restoring harmony to the given damaged picture with parts worn by time, overexposed
Experiment result analysis
The skin melanoma images are segmented by the threshold method with 5% ratio after enhanced with closing-based top-hat and the corresponding binary images are obtained. There are many dermoscopy images without hairs. In such cases, this scheme will force 5% of the image area to be treated as hairs, and these hair noises can be filtered out even entirely by our elongate function defined in (3), as shown Fig. 7(d). We select 40 images containing no hairs to test and calculate the detection error.
Conclusion
Aiming at dermoscopy melanoma image, a novel algorithm for automatically detecting hair and repairing occluded information is proposed in this paper. The hairs in the image are treated as black structures and are processed through morphological closing-based top-hat operator first, and the contrast between hairs and background pixels is enhanced greatly because of the non-sensitivity of the morphology-based enhance method both to the strong and the weak hairs so as to ensure the exactly binary
Acknowledgements
This work was supported by the National Natural Science Foundation of China (grant no. 60672152). The authors are very grateful to the reviewers for their helpful modified comments.
Feng-Ying Xie, female, she is currently a PhD student at the School of Automation Science and Electrical Engineering in BeiHang University. Simultaneously she is also an assistant professor with School of Astronautics in BeiHang University. Her research interests are focused on medical image processing, goal detection and recognition of remotely sensed image, and computer vision.
References (26)
- et al.
Segmentation of skin cancer images
Image Vis Comput
(1999) - et al.
A methodological approach to the classification of dermoscopy images
Comput Med Imaging Graph
(2007) - et al.
Dullrazor: a software approach to hair removal from images
Comput Biol Med
(1997) - et al.
Cancer Statistics 2006
CA: Cancer J Clin
(2006) - et al.
Fast and accurate border detection in dermoscopy images using statistical region merging
Practical skin science
(2005)- et al.
Automated melanoma detection with a novel multispectral imaging system: results of a prospective study
Phys Med Biol
(2005) - et al.
A new algorithm for border description of polarized light surface microscopic images of pigmented skin lesions
IEEE Trans Med Imaging
(2003) - et al.
A new automatic approach for edge detection of skin lesion images
Inf Commun Technol
(2006) Segmentation of digitized dermatoscopic images by two-dimensional color clustering
IEEE Trans Med Imaging
(1999)
A probabilistic neural network framework for detection of malignant melanoma
Neural network diagnosis of malignant skin cancers using principal component analysis as a preprocessor
Nondermatoscopic digital imaging of pigmented lesions
Skin Res Technol
Cited by (79)
Automated hair removal in dermoscopy images using shallow and deep learning neural architectures
2023, Computer Methods and Programs in Biomedicine UpdateResearch on hair removal algorithm of dermatoscopic images based on maximum variance fuzzy clustering and optimization Criminisi algorithm
2022, Biomedical Signal Processing and ControlDigital hair removal by deep learning for skin lesion segmentation
2021, Pattern RecognitionCitation Excerpt :However, this method needs to adjust several parameters that are difficult to determine in general [16]. Xie et al. [11] and Fiorese et al. [15] both apply partial differential equation based (PDE-based) methods for inpainting. However, this is an iterative method which cannot guarantee convergence of the parameters involved [5].
Recent Trends in Computer-Aided Diagnostic Systems for Skin Diseases: Theory, Implementation, and Analysis
2021, Recent Trends in Computer-aided Diagnostic Systems for Skin Diseases: Theory, Implementation, and AnalysisRealistic hair simulator for skin lesion images: A novel benchemarking tool
2020, Artificial Intelligence in MedicineDeep learning-based hair removal for improved diagnostics of skin diseases
2024, Multimedia Tools and Applications
Feng-Ying Xie, female, she is currently a PhD student at the School of Automation Science and Electrical Engineering in BeiHang University. Simultaneously she is also an assistant professor with School of Astronautics in BeiHang University. Her research interests are focused on medical image processing, goal detection and recognition of remotely sensed image, and computer vision.
Shi-Yin Qin, male, he received the Bachelor Degree and the Master Degree of Engineering Science in Automatic Controls and Industrial Systems Engineering from Lanzhou Jiaotong University in 1978 and 1984 respectively, and the PhD Degree in Industrial Control Engineering and Intelligent Automation from Zhejiang University in 1990. He is now a professor at the School of Automation Science and Electrical Engineering in BeiHang University. His current major research topics include image processing and pattern recognition, intelligent optimizing controls of large sale multi-robot hybrid system, complex systems and complexity science, and so on.
Zhi-Guo Jiang, male, he received the Master Degree and PhD Degree in Pattern Recognition and Intelligent Systems from BeiHang University in 1990 and 2005 respectively, and he is now a professor at the Image Processing Center in BeiHang University, his research interests are medical image processing and remotely sensed image processing.
Ru-Song Meng, male, he is now a senior surgery working at General Hospital of the Air Force of PLA, his research interests are morphologic analysis of histiocyte, derma pathology and the clinic application of image analysis technology for dermoscopy image.