Retinal Blood Vessels and Optic Disc Segmentation Using U-Net

Abstract

A color fundus image is a photograph obtained using a fundus camera of the inner wall of the eyeball. In the image, doctors may see changes in the retinal vessels, which can be used to diagnose various dangerous disorders such as arteriosclerosis, some macular degeneration related to age, and glaucoma. To diagnose certain disorders as early as possible, automatic segmentation of retinal arteries is used to help the doctors. Also, it is a challenge for the medical community to analyze the image with the right procedure to diagnose the disorders with high accuracy. Furthermore, this will help the doctor to make the right decision on effective treatment. Hence, the authors have implemented an enhanced architecture called U-Net to segment retinal vessels in this paper. The proposed conventional U-Net permits using all the accessible spatial setting information by adding the multiscale input layer and a thick square to the conventional U-Net in terms of improving the accuracy level of image segmentation. It achieved 95.6% accuracy with a comparatively traditional U-Net model. Moreover, the segmentation results have proved that the proposed approach outperformed in detecting most complex low-contrast blood vessels even when they are very thin. The task of segmenting vessels in retinal images is known as retinal vessel segmentation. Blood vessel density can be assessed using dense pixel values. Data augmentation and analytics play a major role in building the true value of eye blood vessels for medical diagnosis. The proposed method is very promising in the automatic segmentation of retinal arteries.

Grape Leaf Disease Classification Combined with U-Net++ Network and Threshold Segmentation

“…After U-net++ segmentation [13][14][15][16][17][18][19][20][21], the effect is shown in the figure, and the background is set to black and the pixel value is 0.…”

UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias

“…Fundamentally, the bridge network is nothing but a cascade of convolution. We have fundamentally categorized the batch normalization into twelve types of variation, namely, (B1) serial cascades of convolutions [95]; (B2) convolutions with dropouts [102,134]; (B3) dropout in bridge network [80,134,138]; (B4) cascade of convolutions in serial and parallel (DAC and RMP blocks) [108,153]; (B5) bridge normalization [61,95]; (B6) flatten block [69]; (B7) atrous spatial pyramid pooling [135]; (B8) transpose convolution [137]; (B9) patch convolution and transformer [66]; (B10) inception block [97]; (B11) dense layer [114,122]; and (B12) quartent attention [82]. A set of representative examples will be discussed in section VI.…”

“…The different fundamental blocks which were adapted for UNet modification were (Table 4): (1) residual block [75,76,78,84,88,105,129,135,138]; (2) classifier in encoder [145]; (3) Xception block [56,88]; (4) dense layer block [68,100,102,122,142]; (5) recurrent residual block ; (6) attention block [65, 66, 71, 75, 113, 125, 128, 129, 131-135, 139, 161]; (7) dropout layer [70,76,86,95,101,102,134,138]; (8) dilated convolution [67,76]; (9) transpose convolution [66,88,95,137,139]; (10) SE network [92,103,125,133,138], and (11) squeeze and excitation block [92,103,125,133,138].…”

“…2) VASCULAR UNET DESIGNS CHARACTERISTICS David et al [122] designed a UNet system where the authors used different scaled image patches for each contractual layer as input. The idea was to learn more multiscale data.…”

“…Architectures used in both vascular and nonvascular UNet paradigms are the different variations of UNet like UNet with dense block [122], parallel fusion, and serial embedding multiscale feature dense UNet [102], UNet with spatial attention [139], the combination of more than one UNet [186], weighted attention UNet [105], pyramid UNet [106] in the case of retinal vascular, 3D UNet for renal vascular [191] and in case of prostate nonvascular UNet variations like UNet with Xception [88], UNet with residual block [84,189] , 2D bridged UNet [89], UNet with SE block [92], UNet with attention gate [132], UNet with more layers (8) [84], UNet with task consistency learning [83], the combination of UNets (ZNet) [85], and cascaded UNet [93,189], 3D UNet for proximal femur [187]; (ii) It is observed that dropout block [92,103,132,139], residual block [84,105,189] (Figure 19), SE block (Figure 20), attention gate block [103,132,186] (Figure 21), inception block [102], and dense block [102,122] are common in both vascular and non-vascular UNet paradigms; (iii) The augmentation process is also common in both vascular and non-vascular paradigms; and 3D UNet (Figure 22).…”