Determining degree of optic nerve edema from color fundus photography

Agne, Jason; Wang, Jui-Kai; Kardon, Randy H.; Garvin, Mona K.

doi:10.1117/12.2081423

Cited by 12 publications

(10 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…When comparing the volumes computed with from the SD-OCT images and the volumes predicted through this method, initial results were reported at an SPIE conference in 2015 [41], with a mean square error of 2.27 mm 3 and correlation of R = 0.788. Since then, after refining features, a mean square error of 1.93 mm 3 was achieved with a correlation of R = 0.887.…”

Section: Experimental Methods and Resultsmentioning

confidence: 99%

Automated image-based estimation of severity and cause of optic disc edema

Agne¹

Self Cite

View full text Add to dashboard Cite

ACKNOWLEDGEMENTSWhen you have been given the opportunity to spend four years continually improving upon and repurposing prior work, your work becomes prior work and it is difficult to truly deem your role as finished -perhaps more so when you are passing along the torch to someone else. Yet that is what is about to happen here. The work was exciting and challenging, with enough that remains to be done to take decades, but it would appear that my time spent contributing to it is at an end.I first came to the University of Iowa reasonably sure that I wanted to contribute to the field of quantum or optic computation. This ambition was based not on the obvious practical applications, but on the simple fact that the topic seemed mysterious and challenging with relatively few people working on it. However, one of the first lessons I learned when looking for a research advisor to work with was that pretty much all research topics are going to be mysterious and challenging -that is why they need to be researched.On the other hand, one of my favorite things to do is write code. It is so natural and satisfying that I assumed pretty much everyone likes to code, and so any field that requires coding is likely to be more competitive than I wanted to deal with, but it turns out I was wrong about that. When I first met with Professor Garvin to discuss what opportunities she might have available, I was almost expecting a queue. When I explained some version of these assumptions to her, I was expecting confirmation of my suspicions. That is when I learned I was completely wrong about how most of the world felt about coding, and Prof. Garvin became my research advisor.

show abstract

Section: Experimental Methods and Resultsmentioning

confidence: 99%

Automated image-based estimation of severity and cause of optic disc edema

Agne¹

Self Cite

View full text Add to dashboard Cite

show abstract

“…Previous works by Agne et al [15] have sought to extend Echegary's work into the realm of this continuous scale of TRV. His approach uses more widely available 2D fundus photographs to predict the TRV.…”

Section: VIImentioning

confidence: 99%

“…As stated in Chapter 1, Echegaray proposed a semi-automated method of predicting the Frisén grades using fundus photography [12], and Agne extended this work by instead using a monocular fundus photograph to predict the TRV [3,15]. Tang et al [27] has also proven that sterescopic fundus photographs can be used to estimate this volumetric measurement.…”

Section: Graph-based Segmentation Of Retinal Surfaces and Associated mentioning

confidence: 99%

“…Random forest has proven success in many different applications [32], as well as in ophthalmic image analysis. It has been used to generate cost images for layer segmentation [33], predict the total retinal volume of swollen optic nerve heads from fundus photographs [15], estimate papilledema severity using retinal regional volumes and layer shapes [34], and more recently to differentiate between causes of optic disc swelling using retinal layer shape features [1]. A specific deep learning framework, known as convolutional neural networks (CNN), have recently shown encouraging results when applied to image-based tasks for both regression and classification [38].…”

Section: Random Forest Regressionmentioning

confidence: 99%

“…Tang et al [27] demonstrated that the stereoscopic color fundus photographs have the potential to reconstruct the depth information and allow the volumetric estimation for the optic disc swelling, but the requirement of carefully acquired stereo images rather than monocular images limits its applicability. More recently, Agne et al [15] proposed a regression approach, which can directly estimate the total retinal volume (TRV)…”

Section: Motivationmentioning

confidence: 99%

See 2 more Smart Citations