Joint Optic Disc and Cup Segmentation Based on Multi-Label Deep Network and Polar Transformation

By Huan Fu, Jun Cheng, Yanwu Xu, Damon Wing Kee Wong, Jiang Liu, Xiaochun Cao. IEEE Transactions on Medical Imaging, 2018

836 citations recorded by the source corpus.

Glaucoma is a chronic eye disease that leads to irreversible vision loss. The cup to disc ratio (CDR) plays an important role in the screening and diagnosis of glaucoma. Thus, the accurate and automatic segmentation of optic disc (OD) and optic cup (OC) from fundus images is a fundamental task. Most existing methods segment them separately, and rely on hand-crafted visual feature from fundus images. In this paper, we propose a deep learning architecture, named M-Net, which solves the OD and OC segmentation jointly in a one-stage multi-label system. The proposed M-Net mainly consists of multi-scale input layer, U-shape convolutional network, side-output layer, and multi-label loss function. The multi-scale input layer constructs an image pyramid to achieve multiple level receptive field sizes. The U-shape convolutional network is employed as the main body network structure to learn the rich hierarchical representation, while the side-output layer acts as an early classifier that produces a companion local prediction map for different scale layers. Finally, a multi-label loss function is proposed to generate the final segmentation map. For improving the segmentation performance further, we also introduce the polar transformation, which provides the representation of the original image in the polar coordinate system. The experiments show that our M-Net system achieves state-of-the-art OD and OC segmentation result on ORIGA data set. Simultaneously, the proposed method also obtains the satisfactory glaucoma screening performances with calculated CDR value on both ORIGA and SCES datasets.

Subfields: Ophthalmology, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/83e1eb609e4267251a3bcb7ec47cc0754ae5f54b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Ophthalmology

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

CE-Net: Context Encoder Network for 2D Medical Image Segmentation (2019, methodology): “From left to right: original fundus images, state-of-the-art results obtained by superpixel based method [1], M-Net [2], DeepDisc [65], CE-Net and ground-truth masks.”
MA-Net: A Multi-Scale Attention Network for Liver and Tumor Segmentation (2020, methodology): “Inaddition to applying skip connections, the dilation convolution [8] with different rate and pooling operation [11] is also used to capture Multi-scale information [12], [14], [17], [20].”
Inf-Net: Automatic COVID-19 Lung Infection Segmentation From CT Images (2020, background): “3) Parallel Partial Decoder: Several existing medical image segmentation networks segment interested organs/lesions using all high- and low-level features in the encoder branch [55], [56], [60]–[63].”
REFUGE Challenge: A Unified Framework for Evaluating Automated Methods for Glaucoma Assessment from Fundus Photographs (2019, positive): “Most of existing methods use a surrogate segmentation/detection approach to first localize the ONH area and them crop the images around it (Edupuganti et al., 2018; Fu et al., 2018; Lim et al., 2015; Sevastopolsky, 2017; Zilly et al., 2015).”
IDRiD: Diabetic Retinopathy - Segmentation and Grading Challenge. (2020, background): “…channels to other hand-crafted features or local saliency maps for detection of abnormalities associated with DR ( Chudzik et al., 2018; Orlando et al., 2018; Dai et al., 2018 ), segmentation of OD ( Zilly et al., 2017; Fu et al., 2018 ), and detection of DR ( Rangrej and Sivaswamy, 2017 ).”
Medical Image Segmentation Using Deep Learning: A Survey (2020, background): “and brain-tumor segmentation [3] [4], optic disc segmentation [5] [6], cell segmentation [7] [8], lung segmentation, pulmonary nodules [9] [10], cardiac image segmentation [11] [12], etc.”
Disc-Aware Ensemble Network for Glaucoma Screening From Fundus Image (2018, methodology): “5 seconds to generate the ﬁnal result for one fundus image including disc localization and glaucoma screening, which is faster than the existing measurement-based methods, e.g., superpixel method [10] takes 10 seconds, DeepCDR [12] takes 1 second.”
CABNet: Category Attention Block for Imbalanced Diabetic Retinopathy Grading (2020, background): “Thanks to the powerful capability of high-level feature extraction and representational ability of CNNs, they have also been widely used in medical image analysis tasks such as retinal blood vessel segmentation [7], optic disc segmentation [8] and glaucoma screening [9], [10].”
Diagnose like a Radiologist: Attention Guided Convolutional Neural Network for Thorax Disease Classification (2018, background): “The noticeable progress in deep learning has beneﬁted many trials in medical image analysis, such as lesion segmentation or detection [1], [2], [3], [4], [5], diseases classiﬁcation [6], [7], [8], [9], noise induction [10], image annotation [11], [12], registration [13], regression [14] and so on.”
Medical Image Segmentation Review: The Success of U-Net (2022, background): “To better segment the Optic Disc (OD) and Optic Cup (OC) for accurate diagnosis of glaucoma from fundus images, Fu et al. [51] introduce the polar transformation into the U-shape convolutional network with multi-scale input layers to build the Polar Transformation M-Net, aims to extract the richer…”