Aaron Lee

This study evaluates five dataset documentation approaches for health datasets and finds none are widely used or fully suitable, recommending a new standard with guidelines and tools.

The paper trains a deep learning model to quantify brain vascular morphology from standard MR images, showing it captures aging and health-related features not found in traditional metrics.

This study analyzes AI-generated text infiltration in scientific preprints using LLM detectors, finding widespread use across domains and positive impacts on citation numbers.

The study examines age-related changes in white matter properties of optic radiations carrying foveal, macular, and peripheral visual field information using diffusion MRI data from the UK Biobank.

This study uses diffusion MRI and a CNN to show glaucoma has a unique non-linear effect on optic radiation tissue properties, unlike linear effects on other brain connections.

The paper trains a machine learning model on human retinal epigenomic data to predict the functional impact of non-coding regulatory variants.

The paper compares Medicare patients treated by ophthalmologists and optometrists, finding ophthalmologists see older, more racially diverse, and more medically complex patients across most states.

This paper uses deep learning to generate perfusion maps from structural OCT images without expert labels, outperforming clinicians in fidelity.

This study trains a deep learning model to predict future 24-2 Humphrey Visual Fields up to 5.5 years ahead using historical data from glaucomatous patients.

This paper develops a neural network for multiclass and multilabel classification of OCT retinal images, incorporating patient age, gender, and visual acuity to improve diagnosis.

MRI2MRI transforms between brain MRI contrasts using a deep convolutional network, improving spatial registration across different imaging modalities.

This study shows that moderate data (1,800 patients) can achieve near-maximal performance in AMD classification using OCT images, and that separate test sets are not always necessary.

The paper develops a CNN to segment macular edema in OCT images, achieving performance comparable to clinicians.

This study uses deep learning to classify OCT images as normal or AMD, achieving high accuracy across different levels of analysis.