Forecasting Disease Progression with Parallel Hyperplanes in Longitudinal Retinal OCT

TL;DR

This paper introduces a novel deep learning survival prediction method using parallel hyperplanes for forecasting late dry AMD progression from retinal OCT scans, addressing domain shifts and patient heterogeneity.

Contribution

It proposes a new DL model with a family of hyperplanes and unsupervised losses for efficient, domain-adaptive disease progression prediction from medical images.

Findings

Achieved AUROCs of 0.82 and 0.83 on two large datasets.

Model effectively predicts disease progression within 6-24 months.

Enables data-efficient fine-tuning on unlabeled datasets.

Abstract

Predicting future disease progression risk from medical images is challenging due to patient heterogeneity, and subtle or unknown imaging biomarkers. Moreover, deep learning (DL) methods for survival analysis are susceptible to image domain shifts across scanners. We tackle these issues in the task of predicting late dry Age-related Macular Degeneration (dAMD) onset from retinal OCT scans. We propose a novel DL method for survival prediction to jointly predict from the current scan a risk score, inversely related to time-to-conversion, and the probability of conversion within a time interval $t$. It uses a family of parallel hyperplanes generated by parameterizing the bias term as a function of $t$. In addition, we develop unsupervised losses based on intra-subject image pairs to ensure that risk scores increase over time and that future conversion predictions are consistent with AMD stage prediction using actual scans of future visits. Such losses enable data-efficient fine-tuning of the trained model on new unlabeled datasets acquired with a different scanner. Extensive evaluation on two large datasets acquired with different scanners resulted in a mean AUROCs of 0.82 for Dataset-1 and 0.83 for Dataset-2, across prediction intervals of 6,12 and 24 months.