Identification of suitable fundus images using automated quality assessment methods

Abstract

Retinal image quality assessment (IQA) is a crucial process for automated retinal image analysis systems to obtain an accurate and successful diagnosis of retinal diseases. Consequently, the first step in a good retinal image analysis system is measuring the quality of the input image. We present an approach for finding medically suitable retinal images for retinal diagnosis. We used a three-class grading system that consists of good, bad, and outlier classes. We created a retinal image quality dataset with a total of 216 consecutive images called the Diabetic Retinopathy Image Database. We identified the suitable images within the good images for automatic retinal image analysis systems using a novel method. Subsequently, we evaluated our retinal image suitability approach using the Digital Retinal Images for Vessel Extraction and Standard Diabetic Retinopathy Database Calibration level 1 public datasets. The results were measured through the F1 metric, which is a harmonic mean of precision and recall metrics. The highest F1 scores of the IQA tests were 99.60%, 96.50%, and 85.00% for good, bad, and outlier classes, respectively. Additionally, the accuracy of our suitable image detection approach was 98.08%. Our approach can be integrated into any automatic retinal analysis system with sufficient performance scores.

Evaluation of Retinal Image Quality Assessment Networks in Different Color-Spaces

“…There are several publicly available RIQA datasets with manual quality annotations, such as HRF [7], DRIMDB [13], and DR2 [11]. However, they have various drawbacks.…”

Retinal image quality assessment based on image clarity and content

“…Recently, hybrid techniques are being adopted in literature that combine both generic and structural features for RIQA. 7,28,30 Moreover, an interesting emerging approach for sharpness assessment compares retinal images to their blurred versions based on the intuition that unsharp images will be more similar to their blurred versions. 9,16,31…”

“…Then, several distances related to these structures were measured to assess the image's field of view (FOV). Other algorithms relied only on validating the presence of the OD, 43 fovea, 9 or both 30 in their expected locations. As for outlier detection, Giancardo et al 19 and Sevik et al 30 used RGB information for the identification of outliers.…”

“…Other algorithms relied only on validating the presence of the OD, 43 fovea, 9 or both 30 in their expected locations. As for outlier detection, Giancardo et al 19 and Sevik et al 30 used RGB information for the identification of outliers. Yin et al 31 used a two-stage approach to detect outlier images in which a bag of words approach first identified nonretinal images.…”

Retinal Image Enhancement Based on Contrast, Luminosity Adjustment and MSC

“…method though, hardly accounts for the loss of vascular details. [4] Some of the other work related to retinal image processing includes the bowler hat transform a new multiscale vessel enhancement approach which is a mathematical morphology has been proposed. [10] The method combines different structural element to detect the vessels.Another method proposed is Contrast Limited Adaptive Histogram Equalization (CLAHE) for preprocessing and Tandem Pulse Coupled Neural Network (TPCNN) model for automatic feature vectors generation then classification and extraction of the retinal blood vessels based on Deep Learning Based Support Vector Machine (DLBSVM).…”