A Statistical Segmentation Method for Measuring Age-Related Macular Degeneration in Retinal Fundus Images

Abstract

Day by day, huge amount of information is collected in medical databases. These databases include quite interesting information that could be exploited in diagnosis of illnesses and medical treatment of patients. Classification of these data is getting harder as the databases are expanded. On the other hand, automated image analysis and processing is one of the most promising areas of computer vision used in medical diagnosis and treatment. In this context, retinal fundus images, offering very high resolutions that are sufficient for most of the clinical cases, provide many indications that could be exploited in diagnosing and screening retinal degenerations or diseases. Consequently, there is a strong demand in developing automated evaluation systems to utilize the information stored in the medical databases. This study proposes an automatic method for segmentation of ARMD in retinal fundus images. The method used in the automated system extracts lesions of the ARMD by employing a statistical method. In order to do this, the statistical segmentation method is first used to extract the healthy area of the macula that is more familiar and regular than the unhealthy parts. Here, characteristic images of the patterns of the macula are extracted and used to segment the healthy textures of an eye. In addition to this, blood vessels are also extracted and then classified as healthy regions. Finally, the inverse image of the segmented image is generated which determines the unhealthy regions of the macula. The performance of the method is examined on various quality retinal fundus images. Segmented images are also compared with consecutive images of the same patient to follow up the changes in the disease.

Image Classification for Age-related Macular Degeneration Screening Using Hierarchical Image Decompositions and Graph Mining

“…The earliest work [24] used a morphological mechanism to localise drusen. Other image processing techniques that have been applied include: (i) histogram-based adaptive local thresholding [22] , (ii) region growing [19,20]; (iii) wavelet based feature identification coupled with multilevel classification [3]; (iv) anomaly detection based approaches, that employ Support Vector Data Description (SVDD), to segment anomalous pixels [11]; and (v) signal based approaches, namely amplitude-modulation frequency-modulation (AM-FM), to generate multi-scale features for drusen classification [1,2]. Content-Based Image Retrieval (CBIR) techniques have also been applied.…”

Dynamic Time Warping (DTW)

“…Hard drusen have a well defined border, while soft drusen have boundaries that often blend into the retinal background. Figure 2 [4,13,22,23,29] as opposed to AMD classification. The work proposed here however approaches the AMD screening problem without the need for identification of the physical existence of drusen and aims to classify images as either "AMD" or "non-AMD".…”

“…More recent work [4] used a wavelet analysis technique to extract drusen patterns, and multi-level classification (based on various criteria) for drusen categorisation. Other works on the identification of drusen in retina images has focuses on segmentation coupled with image enhancement approaches [22,23,29]. Rapantzikos et al [29] adopted a multilevel histogram equalisation to enhance the image contrast followed by drusen segmentation, in which two types of threshold, global and local, were applied to retinal images.…”

“…Rapantzikos et al [29] adopted a multilevel histogram equalisation to enhance the image contrast followed by drusen segmentation, in which two types of threshold, global and local, were applied to retinal images. Köse et al [22,23] proposed two approaches involving inverse drusen segmentation within the macular area. A region growing technique was used to identify "healthy" pixels by applying a threshold on the colour intensity levels [22].…”

“…Once this was done, the inverse of the segmented image was used to generate the segmentation of the drusen. A similar inverse segmentation approach, supported by statistical information, was adopted in [23]; where healthy Characteristic Images (CIs) were compared to new Sample Images (SIs) and a predetermined threshold is applied to classify SI. In [13] another approach, based on a non-parametric technique for anomaly detection, was described that uses a Support Vector Data Description (SVDD) to segment anomalous pixels.…”