scholarly journals Weakly-Supervised Individual Ganglion Cell Segmentation from Adaptive Optics OCT Images for Glaucomatous Damage Assessment

Optica ◽  
2021 ◽  
Author(s):  
Somayyeh Soltanian-Zadeh ◽  
Kazuhiro Kurokawa ◽  
Zhuolin Liu ◽  
Furu Zhang ◽  
Saeedi Osamah ◽  
...  
2015 ◽  
Vol 1 (1) ◽  
Author(s):  
Ameneh Boroomand ◽  
Alexander Wong ◽  
Kostadinka Bizheva

<p>Keratocytes are vital for maintaining the overall health of human<br />cornea as they preserve the corneal transparency and help in healing<br />corneal injuries. Manual segmentation of keratocytes is challenging,<br />time consuming and also needs an expert. Here, we propose<br />a novel semi-automatic segmentation framework, called Conditional<br />Random FieldWeakly Supervised Segmentation (CRF-WSS)<br />to perform the keratocytes cell segmentation. The proposed framework<br />exploits the concept of dictionary learning in a sparse model<br />along with the Conditional Random Field (CRF) modeling to segment<br />keratocytes cells in Ultra High Resolution Optical Coherence<br />Tomography (UHR-OCT) images of human cornea. The results<br />show higher accuracy for the proposed CRF-WSS framework compare<br />to the other tested Supervised Segmentation (SS) andWeakly<br />Supervised Segmentation (WSS) methods.</p>


2016 ◽  
Vol 94 ◽  
Author(s):  
G. Triolo ◽  
P. Monsalve ◽  
W.J. Feuer ◽  
J.C. Mwanza ◽  
S.J. Gedde ◽  
...  

2017 ◽  
Vol 6 (4) ◽  
pp. 6 ◽  
Author(s):  
Donald C. Hood ◽  
Dongwon Lee ◽  
Ravivarn Jarukasetphon ◽  
Jason Nunez ◽  
Maria A. Mavrommatis ◽  
...  

2017 ◽  
Vol 114 (3) ◽  
pp. 586-591 ◽  
Author(s):  
Ethan A. Rossi ◽  
Charles E. Granger ◽  
Robin Sharma ◽  
Qiang Yang ◽  
Kenichi Saito ◽  
...  

Although imaging of the living retina with adaptive optics scanning light ophthalmoscopy (AOSLO) provides microscopic access to individual cells, such as photoreceptors, retinal pigment epithelial cells, and blood cells in the retinal vasculature, other important cell classes, such as retinal ganglion cells, have proven much more challenging to image. The near transparency of inner retinal cells is advantageous for vision, as light must pass through them to reach the photoreceptors, but it has prevented them from being directly imaged in vivo. Here we show that the individual somas of neurons within the retinal ganglion cell (RGC) layer can be imaged with a modification of confocal AOSLO, in both monkeys and humans. Human images of RGC layer neurons did not match the quality of monkey images for several reasons, including safety concerns that limited the light levels permissible for human imaging. We also show that the same technique applied to the photoreceptor layer can resolve ambiguity about cone survival in age-related macular degeneration. The capability to noninvasively image RGC layer neurons in the living eye may one day allow for a better understanding of diseases, such as glaucoma, and accelerate the development of therapeutic strategies that aim to protect these cells. This method may also prove useful for imaging other structures, such as neurons in the brain.


2020 ◽  
Author(s):  
Elena Gofas-Salas ◽  
Yuhua Rui ◽  
Pedro Mecê ◽  
Min Zhang ◽  
Valerie C. Snyder ◽  
...  

AbstractRetinal ganglion cells (RGCs) are the primary output neurons of the retina. RGC dysfunction and death can cause irreversible vision loss in glaucoma and other ocular diseases. However, no methods exist to evaluate RGCs at the level of single cell in the living human eye in the clinic. Our aim is to implement a technique revealing the retinal ganglion cell layer neurons whose contrast, robustness and acquisition time would make it suitable for clinical diagnosis and monitoring of patients. While previously we were able to demonstrate high contrast imaging in monkeys, here we propose a new adaptive optics scanning laser ophthalmoscope configuration that allows us to achieve similar results on humans in vivo. In particular we used a new detection scheme that allowed us to switch from two light sources to one thereby enabling us to increase the light power, eliminate problems caused by chromatic aberration and improve the image registration process. Here we show that this optimized detection scheme and image processing pipeline improve the multi-offset technique for imaging of human RGC layer neurons.


2015 ◽  
Vol 4 (2) ◽  
pp. 12 ◽  
Author(s):  
Donald C. Hood ◽  
Monica F. Chen ◽  
Dongwon Lee ◽  
Benjamin Epstein ◽  
Paula Alhadeff ◽  
...  

2020 ◽  
Vol 12 (2) ◽  
pp. 207 ◽  
Author(s):  
Sherrie Wang ◽  
William Chen ◽  
Sang Michael Xie ◽  
George Azzari ◽  
David B. Lobell

Accurate automated segmentation of remote sensing data could benefit applications from land cover mapping and agricultural monitoring to urban development surveyal and disaster damage assessment. While convolutional neural networks (CNNs) achieve state-of-the-art accuracy when segmenting natural images with huge labeled datasets, their successful translation to remote sensing tasks has been limited by low quantities of ground truth labels, especially fully segmented ones, in the remote sensing domain. In this work, we perform cropland segmentation using two types of labels commonly found in remote sensing datasets that can be considered sources of “weak supervision”: (1) labels comprised of single geotagged points and (2) image-level labels. We demonstrate that (1) a U-Net trained on a single labeled pixel per image and (2) a U-Net image classifier transferred to segmentation can outperform pixel-level algorithms such as logistic regression, support vector machine, and random forest. While the high performance of neural networks is well-established for large datasets, our experiments indicate that U-Nets trained on weak labels outperform baseline methods with as few as 100 labels. Neural networks, therefore, can combine superior classification performance with efficient label usage, and allow pixel-level labels to be obtained from image labels.


Sign in / Sign up

Export Citation Format

Share Document