Exemplar Diffusion: Improving Medical Object Detection with Opportunistic Labels

TL;DR

This paper introduces Exemplar Diffusion, a training-free method that uses exemplars at inference to enhance medical object detection, improving accuracy and robustness, and enabling uncertainty quantification.

Contribution

It proposes a novel exemplar diffusion framework that leverages existing diffusion methods for improved medical object detection without additional training.

Findings

Increases average precision and recall on medical datasets.

Provides robustness to exemplar quality and supports non-expert annotations.

Enables quantification of predictive uncertainty.

Abstract

We present a framework to take advantage of existing labels at inference, called \textit{exemplars}, in order to improve the performance of object detection in medical images. The method, \textit{exemplar diffusion}, leverages existing diffusion methods for object detection to enable a training-free approach to adding information of known bounding boxes at test time. We demonstrate that for medical image datasets with clear spatial structure, the method yields an across-the-board increase in average precision and recall, and a robustness to exemplar quality, enabling non-expert annotation. Moreover, we demonstrate how our method may also be used to quantify predictive uncertainty in diffusion detection methods. Source code and data splits openly available online: https://github.com/waahlstrand/ExemplarDiffusion