Ashish Issac

An adaptive threshold based image processing technique for improved glaucoma detection and classification

Glaucoma is an optic neuropathy which is one of the main causes of permanent blindness worldwide. This paper presents an automatic image processing based method for detection of glaucoma from the digital fundus images. In this proposed work, the discriminatory parameters of glaucoma infection, such as cup to disc ratio (CDR), neuro retinal rim (NRR) area and blood vessels in different regions of the optic disc has been used as features and fed as inputs to learning algorithms for glaucoma diagnosis. These features which have discriminatory changes with the occurrence of glaucoma are strategically used for training the classifiers to improve the accuracy of identification. The segmentation of optic disc and cup based on adaptive threshold of the pixel intensities lying in the optic nerve head region. Unlike existing methods the proposed algorithm is based on an adaptive threshold that uses local features from the fundus image for segmentation of optic cup and optic disc making it invariant to the quality of the image and noise content which may find wider acceptability. The experimental results indicate that such features are more significant in comparison to the statistical or textural features as considered in existing works. The proposed work achieves an accuracy of 94.11% with a sensitivity of 100%. A comparison of the proposed work with the existing methods indicates that the proposed approach has improved accuracy of classification glaucoma from a digital fundus which may be considered clinically significant.

An adaptive threshold based algorithm for optic disc and cup segmentation in fundus images

This paper presents an image processing technique for segmentation of optic disc and cup based on adaptive thresholding using features from the image. The proposed algorithm uses the features obtained from the image, such as mean and standard deviation, to remove information from the red and green channel of a fundus image and obtain an image which contains only the optic nerve head region in both the channels. The optic disc is segmented from the red channel and optic cup from the green channel respectively. The threshold is determined from the smoothed histogram of the preprocessed image. The results of the proposed algorithm are compared with the images that are marked by doctors. The accuracy of the algorithm is good and is computationally very fast. The proposed method can be used for screening purpose.

An automated and robust image processing algorithm for glaucoma diagnosis from fundus images using novel blood vessel tracking and bend point detection

Glaucoma is an ocular disease which can cause irreversible blindness. The disease is currently identified using specialized equipment operated by optometrists manually. The proposed work aims to provide an efficient imaging solution which can help in automating the process of Glaucoma diagnosis using computer vision techniques from digital fundus images. The proposed method segments the optic disc using a geometrical feature based strategic framework which improves the detection accuracy and makes the algorithm invariant to illumination and noise. Corner thresholding and point contour joining based novel methods are proposed to construct smooth contours of Optic Disc. Based on a clinical approach as used by ophthalmologist, the proposed algorithm tracks blood vessels inside the disc region and identifies the points at which first vessel bend from the optic disc boundary and connects them to obtain the contours of Optic Cup. The proposed method has been compared with the ground truth marked by the medical experts and the similarity parameters, used to determine the performance of the proposed method, have yield a high similarity of segmentation. The proposed method has achieved a macro-averaged f-score of 0.9485 and accuracy of 97.01% in correctly classifying fundus images. The proposed method is clinically significant and can be used for Glaucoma screening over a large population which will work in a real time.

Adaptive Thresholding for Skin Lesion Segmentation Using Statistical Parameters

Melanoma is one of those skin cancers whichcan be fatal. So, the segmentation of lesions from the digitalimages becomes a crucial step in analysis and diagnosis of askin cancer using image processing techniques. This workproposes a technique for automatic segmentation of lesionfrom the digital dermoscopic images using adaptive thresholdmaking the process invariant and robust. The statisticalfeatures like standard deviation and mean are used topreprocess and segment the lesions completely in an automaticmanner. The use of image processing techniques, such asaverage filtering for removal of hair and skin scales, mathematical morphology to reject the false positives havebeen successfully able to accurately segment the lesion fromthe images. The results are significant and indicate that themethod has good accuracy. An average correlation of 90% andaverage overlapping score of 83% has been obtained.

An efficient imaging technique for automated macula localization from fundus images

Localization of macula from fundus image plays an important role to design an automated screening tool for detection of retinal diseases. The similar color and texture of red lesions act as a bottleneck in accurate localization of macula in the fundus image. This paper presents a computer vision algorithm for automated and efficient localization of macula from low contrast and diabetic retinopathy affected fundus images. A statistical based model is used to detect macula in a specified region of fundus image which is designed using the geometric features of optic disc. The performance of the proposed algorithm of macula detection was tested on 200 normal/affected fundus images and results are significant. The computational efficiency and accurate localization of macula makes the proposed method competent enough to be used as a part of an automated screening tool for detection of retinal diseases.

An efficient image processing based method for gills segmentation from a digital fish image

The quality and freshness of a fish sample is mainly affected due to the handling and storage conditions during the post harvesting period. The retention time and storage medium are the two main factors affecting the fish quality. This paper presents an image processing based method for automatic and efficient segmentation of gills from the fish sample image which can be used for fish freshness validation and determination of any pesticide from the fish sample under test. The implemented algorithm has produced a maximum correlation of 92.4% with the ground truth results obtained from experts. The method used for gills segmentation is fast and simple.

Automatic imaging method for optic disc segmentation using morphological techniques and active contour fitting

Optic disc segmentation is a crucial step in diagnosis of various ocular diseases like Glaucoma and Diabetic Retinopathy. This work proposes a technique for automatic detection of optic disc from the fundus images using edge based and active contour fitting method. The proposed work has used image processing techniques such as smoothing filters for removal of blood vessels, morphological operations to correctly segment the optic disc and reject the false positives, active contour snake based model for smoothing of optic disc boundary. The results of optic disc segmentation obtained from the proposed work are compared with the ground truth marked by the ophthalmologists. The results are convincing and segmentation results show that the method has good accuracy. An average overlapping score of more than 90% is obtained for the fundus images under test.

Automatic computer vision-based detection and quantitative analysis of indicative parameters for grading of diabetic retinopathy

Abstract Diabetic retinopathy (DR) is one of the complications of diabetes affecting the eyes. If not treated at an early stage, then it can cause permanent blindness. The present work proposes a method for automatic detection of pathologies that are indicative parameters for DR and use them strategically in a framework to grade the severity of the disease. The bright lesions are highlighted using a normalization process followed by anisotropic diffusion and intensity threshold for detection of lesions which makes the algorithm robust to correctly reject false positives. SVM-based classifier is used to reject false positives using 10 distinct feature types. Red lesions are accurately detected from a shade-corrected green channel image, followed by morphological flood filling and regional minima operations. The rejection of false positives using geometrical features makes the system less complex and computationally efficient. A comprehensive quantitative analysis to grade the severity of the disease has resulted in an average sensitivity of 92.85 and 86.03% on DIARETDB1 and MESSIDOR databases, respectively.

Image processing based acrylamide detection from fried potato chip images using continuous wavelet transform

Abstract Fast foods like potato chips and French fries are very common and easily available. The preparation process of such food items is a detrimental factor in deciding if the item is suitable for consumption. Identification of presence of toxic substances like acrylamide in potato chips through conventional methods are time consuming and destructive and needs trained manpower. In the proposed work, an automatic image processing based technique is proposed to detect presence of acrylamide in fried potato chips. The potato chip area is segmented from its background followed by extraction of discriminatory features in the continuous wavelet transform domain using Morlet wavelet. The discriminatory features are analysed strategically and fed to LOOCV based Support Vector Machine classifier to identify presence of acrylamide in the potato chips. The proposed method has an accuracy of 98.33% with 100% specificity. Convincing results and fast computational time indicates that the proposed work can be used for development of non-destructive real-time applications for food quality monitoring.

Automated skin lesion segmentation using K-Means clustering from digital dermoscopic images

Melanoma can prove fatal if not diagnosed at early stage. The accuracy of identification of skin cancer from dermoscopic images is directly proportional to the accuracy of the skin lesion segmentation. This work proposes a skin lesion segmentation method using clustering technique. The use of smoothing filter and area thresholding is competent enough to sufficiently reject the noisy pixels from the finally segmented image. The results of skin lesion segmentation obtained from the proposed algorithm has been compared with the annotated images. The results have been expressed in the form of overlapping score and correlation coefficient. The maximum values of overlapping score and correlation coefficient obtained from the algorithm are 96.75% and 97.66% respectively. The results are convincing and suggests that the proposed work can be used for some real time application.