Corpus ID: 237571689

Predicting Visual Improvement after Macular Hole Surgery: a Cautionary Tale on Deep Learning with Very Limited Data

@article{Godbout2021PredictingVI,
  title={Predicting Visual Improvement after Macular Hole Surgery: a Cautionary Tale on Deep Learning with Very Limited Data},
  author={Martin Godbout and Alexandre Lachance and Fares Antaki and Ali Dirani and Anne Durand},
  journal={ArXiv},
  year={2021},
  volume={abs/2109.09463}
}
We investigate the potential of machine learning models for the prediction of visual improvement after macular hole surgery from preoperative data (retinal images and clinical features). Collecting our own data for the task, we end up with only 121 total samples, putting our work in the very limited data regime. We explore a variety of deep learning methods for limited data to train deep computer vision models, finding that all tested deep vision models are outperformed by a simple regression… Expand

Figures and Tables from this paper

References

SHOWING 1-10 OF 26 REFERENCES
Predicting Postoperative Vision for Macular Hole with Automated Image Analysis
A predictive model for the visual outcome after successful macular hole surgery using a prospectively collected dataset and automated image analysis of pre-operative spectral domain optical coherenceExpand
Factors affecting anatomical and visual outcome after macular hole surgery: findings from a large prospective UK cohort
TLDR
The definition of a large hole should be changed to around 500 μm, and patients should be operated on early to help achieve a good post-operative VA, according to a Design Database study of primary macular hole operations. Expand
Area and volume ratios for prediction of visual outcome in idiopathic macular hole.
TLDR
Compared with MHI and HFF, ARF could effectively express three-dimensional characteristics of macular hole and achieve better sensitivity and specificity and could be the most effective parameter to predict the visual outcome in Macular hole surgery. Expand
Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs.
TLDR
An algorithm based on deep machine learning had high sensitivity and specificity for detecting referable diabetic retinopathy and diabetic macular edema in retinal fundus photographs from adults with diabetes. Expand
The Visual Outcomes of Macular Hole Surgery: A Registry-Based Study by the Australian and New Zealand Society of Retinal Specialists.
TLDR
Macular hole surgery substantially improved acuity in approximately two thirds of patients and was seldom associated with loss of vision, and SF6 gas was associated with better visual outcomes, an effect that warrants further study. Expand
Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning
TLDR
A diagnostic tool based on a deep-learning framework for the screening of patients with common treatable blinding retinal diseases, which demonstrates performance comparable to that of human experts in classifying age-related macular degeneration and diabetic macular edema. Expand
Revision Surgery for Idiopathic Macular Hole after Failed Primary Vitrectomy
TLDR
The success rate of revision surgery in eyes with unclosed MH is 85%. Expand
Development of a deep residual learning algorithm to screen for glaucoma from fundus photography
TLDR
A deep residual learning algorithm was developed to screen for glaucoma from fundus photography and measure its diagnostic performance compared to Residents in Ophthalmology, and its diagnostic accuracy was validated in the testing dataset, using the area under the receiver operating characteristic curve (AROC). Expand
Deep learning-based automated detection of retinal diseases using optical coherence tomography images.
TLDR
A 4-class classification problem to automatically detect choroidal neovascularization, diabetic macular edema, DRUSEN, and NORMAL in optical coherence tomography (OCT) images and the effect of the integration of retinal OCT images and medical history data from patients on model performance is explored. Expand
Responsiveness of NEI VFQ-25 to changes in visual acuity in neovascular AMD: validation studies from two phase 3 clinical trials.
TLDR
Using data from the MARINA and ANCHOR trials in neovascular age-related macular degeneration and establishing the change in the NEI VFQ-25 associated with a 15-letter change in best corrected visual acuity (BCVA) supports the use of the NEi VFq-25 as a responsive and sensitive measure of vision-related function in nevascular AMD populations. Expand
...
1
2
3
...