Low-Resolution Retinal Image Vessel Segmentation

Zengin, Hasan Alp; Camara, José; Coelho, Paulo; Rodrigues, João M. F.; Cunha, António

doi:10.1007/978-3-030-49108-6_44

Cited by 1 publication

(14 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on those class divisions, Table 4 was compiled, showing the number of images, the average MAE, and average IoU for each classification class, being the best results highlighted in bold. Since the dataset is the same (DS1), a direct comparison of these results can be performed to the ones presented by Zengin et al [6]. Considering the overall success as the merge of both Successful and Acceptable classes (IoU > 0.6), for Zengin et al [6] can be observed that 84.3% were successfully classified, although only being 44.7% from Successful class.…”

Section: Retinal Detectionmentioning

confidence: 89%

“…For all the left, one can observe that most images that express good results from the three comparable methods are quite good, presenting BBs quite similar to the ones from the ground truth (in yellow). Considering the middle images for the Acceptable classification ones can observe a visually poor result for Zengin et al [6] method, due to the reduced size of the BB when compared to the ground truth, suggesting a higher MAE value and a lower IoU value. For the rightmost column, visually, the most acceptable misclassification would be the one from YOLO v4, where a good portion of the retinal area is detected.…”

Section: Retinal Detectionmentioning

confidence: 96%

“…Two retinal datasets were used in the development of this work. One is private, the D-Eye dataset [6], and the other is public, the FIRE dataset [5].…”

Section: Datasetsmentioning

confidence: 99%

“…The Proposed method overall results, the trained model of YOLO v4 with DS1, and the Zengin et al [6] retinal detection task results were directly compared. Based on visual analysis and the IoU values, three ranges were defined for image classification: Successful class is defined for images with IoU greater than 0.8.…”

Section: Retinal Detectionmentioning

confidence: 99%

“…Here, two methods were proposed and compared: a classical image processing approach and a YOLO v4 neural network. To explore this task, a private dataset presented by Zengin et al [6] , which contains 26 retina videos around the optic disc, with lower-resolution images, and annotated two subsets, one with the localization of the visible retinal area and other with vessel segmentation was used; •…”

mentioning

confidence: 99%

See 4 more Smart Citations

Detection and Mosaicing Techniques for Low-Quality Retinal Videos

Camara

Silva

Gouveia

et al. 2022

Sensors

Self Cite

View full text Add to dashboard Cite

Ideally, to carry out screening for eye diseases, it is expected to use specialized medical equipment to capture retinal fundus images. However, since this kind of equipment is generally expensive and has low portability, and with the development of technology and the emergence of smartphones, new portable and cheaper screening options have emerged, one of them being the D-Eye device. When compared to specialized equipment, this equipment and other similar devices associated with a smartphone present lower quality and less field-of-view in the retinal video captured, yet with sufficient quality to perform a medical pre-screening. Individuals can be referred for specialized screening to obtain a medical diagnosis if necessary. Two methods were proposed to extract the relevant regions from these lower-quality videos (the retinal zone). The first one is based on classical image processing approaches such as thresholds and Hough Circle transform. The other performs the extraction of the retinal location by applying a neural network, which is one of the methods reported in the literature with good performance for object detection, the YOLO v4, which was demonstrated to be the preferred method to apply. A mosaicing technique was implemented from the relevant retina regions to obtain a more informative single image with a higher field of view. It was divided into two stages: the GLAMpoints neural network was applied to extract relevant points in the first stage. Some homography transformations are carried out to have in the same referential the overlap of common regions of the images. In the second stage, a smoothing process was performed in the transition between images.

show abstract

Section: Retinal Detectionmentioning

confidence: 89%

Section: Retinal Detectionmentioning

confidence: 96%

“…Two retinal datasets were used in the development of this work. One is private, the D-Eye dataset [6], and the other is public, the FIRE dataset [5].…”

Section: Datasetsmentioning

confidence: 99%

Section: Retinal Detectionmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations