Diffusion Models with Implicit Guidance for Medical Anomaly Detection

Cosmin I. Bercea; Benedikt Wiestler; Daniel Rueckert; Julia A. Schnabel

Diffusion Models with Implicit Guidance for Medical Anomaly Detection

Cosmin I. Bercea, Benedikt Wiestler, Daniel Rueckert, Julia A. Schnabel

Research output: Working paper/Preprint › Preprint

11 Downloads (Pure)

Abstract

Diffusion models have advanced unsupervised anomaly detection by improving the transformation of pathological images into pseudo-healthy equivalents. Nonetheless, standard approaches may compromise critical information during pathology removal, leading to restorations that do not align with unaffected regions in the original scans. Such discrepancies can inadvertently increase false positive rates and reduce specificity, complicating radiological evaluations. This paper introduces Temporal Harmonization for Optimal Restoration (THOR), which refines the de-noising process by integrating implicit guidance through temporal anomaly maps. THOR aims to preserve the integrity of healthy tissue in areas unaffected by pathology. Comparative evaluations show that THOR surpasses existing diffusion-based methods in detecting and segmenting anomalies in brain MRIs and wrist X-rays. Code: https://github.com/ci-ber/THOR_DDPM.

Original language	Undefined/Unknown
Publication status	Published - 13 Mar 2024

Keywords

eess.IV
cs.CV
cs.LG

Access to Document

2403.08464v1

Cite this

@techreport{80181956c4de4cf1985b1c19d28b9985,

title = "Diffusion Models with Implicit Guidance for Medical Anomaly Detection",

abstract = "Diffusion models have advanced unsupervised anomaly detection by improving the transformation of pathological images into pseudo-healthy equivalents. Nonetheless, standard approaches may compromise critical information during pathology removal, leading to restorations that do not align with unaffected regions in the original scans. Such discrepancies can inadvertently increase false positive rates and reduce specificity, complicating radiological evaluations. This paper introduces Temporal Harmonization for Optimal Restoration (THOR), which refines the de-noising process by integrating implicit guidance through temporal anomaly maps. THOR aims to preserve the integrity of healthy tissue in areas unaffected by pathology. Comparative evaluations show that THOR surpasses existing diffusion-based methods in detecting and segmenting anomalies in brain MRIs and wrist X-rays. Code: https://github.com/ci-ber/THOR_DDPM.",

keywords = "eess.IV, cs.CV, cs.LG",

author = "Bercea, {Cosmin I.} and Benedikt Wiestler and Daniel Rueckert and Schnabel, {Julia A.}",

year = "2024",

month = mar,

day = "13",

language = "Undefined/Unknown",

type = "WorkingPaper",

}

TY - UNPB

T1 - Diffusion Models with Implicit Guidance for Medical Anomaly Detection

AU - Bercea, Cosmin I.

AU - Wiestler, Benedikt

AU - Rueckert, Daniel

AU - Schnabel, Julia A.

PY - 2024/3/13

Y1 - 2024/3/13

N2 - Diffusion models have advanced unsupervised anomaly detection by improving the transformation of pathological images into pseudo-healthy equivalents. Nonetheless, standard approaches may compromise critical information during pathology removal, leading to restorations that do not align with unaffected regions in the original scans. Such discrepancies can inadvertently increase false positive rates and reduce specificity, complicating radiological evaluations. This paper introduces Temporal Harmonization for Optimal Restoration (THOR), which refines the de-noising process by integrating implicit guidance through temporal anomaly maps. THOR aims to preserve the integrity of healthy tissue in areas unaffected by pathology. Comparative evaluations show that THOR surpasses existing diffusion-based methods in detecting and segmenting anomalies in brain MRIs and wrist X-rays. Code: https://github.com/ci-ber/THOR_DDPM.

AB - Diffusion models have advanced unsupervised anomaly detection by improving the transformation of pathological images into pseudo-healthy equivalents. Nonetheless, standard approaches may compromise critical information during pathology removal, leading to restorations that do not align with unaffected regions in the original scans. Such discrepancies can inadvertently increase false positive rates and reduce specificity, complicating radiological evaluations. This paper introduces Temporal Harmonization for Optimal Restoration (THOR), which refines the de-noising process by integrating implicit guidance through temporal anomaly maps. THOR aims to preserve the integrity of healthy tissue in areas unaffected by pathology. Comparative evaluations show that THOR surpasses existing diffusion-based methods in detecting and segmenting anomalies in brain MRIs and wrist X-rays. Code: https://github.com/ci-ber/THOR_DDPM.

KW - eess.IV

KW - cs.CV

KW - cs.LG

M3 - Preprint

BT - Diffusion Models with Implicit Guidance for Medical Anomaly Detection

ER -