RadEdit: stress-testing biomedical vision models via diffusion image editing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pérez-García, Fernando, Bond-Taylor, Sam, Sanchez, Pedro P., van Breugel, Boris, Castro, Daniel C., Sharma, Harshita, Salvatelli, Valentina, Wetscherek, Maria T. A., Richardson, Hannah, Lungren, Matthew P., Nori, Aditya, Alvarez-Valle, Javier, Oktay, Ozan, Ilse, Maximilian
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929631208144896
author Pérez-García, Fernando
Bond-Taylor, Sam
Sanchez, Pedro P.
van Breugel, Boris
Castro, Daniel C.
Sharma, Harshita
Salvatelli, Valentina
Wetscherek, Maria T. A.
Richardson, Hannah
Lungren, Matthew P.
Nori, Aditya
Alvarez-Valle, Javier
Oktay, Ozan
Ilse, Maximilian
author_facet Pérez-García, Fernando
Bond-Taylor, Sam
Sanchez, Pedro P.
van Breugel, Boris
Castro, Daniel C.
Sharma, Harshita
Salvatelli, Valentina
Wetscherek, Maria T. A.
Richardson, Hannah
Lungren, Matthew P.
Nori, Aditya
Alvarez-Valle, Javier
Oktay, Ozan
Ilse, Maximilian
contents Biomedical imaging datasets are often small and biased, meaning that real-world performance of predictive models can be substantially lower than expected from internal testing. This work proposes using generative image editing to simulate dataset shifts and diagnose failure modes of biomedical vision models; this can be used in advance of deployment to assess readiness, potentially reducing cost and patient harm. Existing editing methods can produce undesirable changes, with spurious correlations learned due to the co-occurrence of disease and treatment interventions, limiting practical applicability. To address this, we train a text-to-image diffusion model on multiple chest X-ray datasets and introduce a new editing method RadEdit that uses multiple masks, if present, to constrain changes and ensure consistency in the edited images. We consider three types of dataset shifts: acquisition shift, manifestation shift, and population shift, and demonstrate that our approach can diagnose failures and quantify model robustness without additional data collection, complementing more qualitative tools for explainable AI.
format Preprint
id arxiv_https___arxiv_org_abs_2312_12865
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle RadEdit: stress-testing biomedical vision models via diffusion image editing
Pérez-García, Fernando
Bond-Taylor, Sam
Sanchez, Pedro P.
van Breugel, Boris
Castro, Daniel C.
Sharma, Harshita
Salvatelli, Valentina
Wetscherek, Maria T. A.
Richardson, Hannah
Lungren, Matthew P.
Nori, Aditya
Alvarez-Valle, Javier
Oktay, Ozan
Ilse, Maximilian
Computer Vision and Pattern Recognition
Artificial Intelligence
Biomedical imaging datasets are often small and biased, meaning that real-world performance of predictive models can be substantially lower than expected from internal testing. This work proposes using generative image editing to simulate dataset shifts and diagnose failure modes of biomedical vision models; this can be used in advance of deployment to assess readiness, potentially reducing cost and patient harm. Existing editing methods can produce undesirable changes, with spurious correlations learned due to the co-occurrence of disease and treatment interventions, limiting practical applicability. To address this, we train a text-to-image diffusion model on multiple chest X-ray datasets and introduce a new editing method RadEdit that uses multiple masks, if present, to constrain changes and ensure consistency in the edited images. We consider three types of dataset shifts: acquisition shift, manifestation shift, and population shift, and demonstrate that our approach can diagnose failures and quantify model robustness without additional data collection, complementing more qualitative tools for explainable AI.
title RadEdit: stress-testing biomedical vision models via diffusion image editing
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2312.12865