Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images - Takeaways - MDSpire

Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images

  • By

  • Takahiro Nakao

  • Shouhei Hanaoka

  • Yukihiro Nomura

  • Takeharu Yoshikawa

  • Osamu Abe

  • May 25, 2026

  • 0 min

Share

  • 1

    The study focuses on unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT imaging.

  • 2

    This approach reduces false positives compared to traditional subtraction methods in detecting newly appearing lesions.

  • 3

    Unsupervised anomaly detection does not require annotated datasets, allowing for the identification of diverse abnormalities.

  • 4

    The study utilized PET/CT scans from 4,176 subjects, with 3,000 selected for model development.

  • 5

    A total of 261 abnormal cases were evaluated for performance analysis after rigorous inclusion and exclusion criteria.

Original Source(s)

Unsupervised anomaly detection for longitudinal comparison in whole-body PET/CT images

  • by Takahiro Nakao, Shouhei Hanaoka, Yukihiro Nomura, Takeharu Yoshikawa, Osamu Abe

  • May 25, 2026

Related Content

Consensus Guidance May Improve Sinonasal Mass Outcomes

Multidisciplinary recommendations emphasize biopsy technique, imaging, and pathology review in patients with suspected sinonasal malignancy.

Enhanced differentiation of breast lesions through integration of microvascular flow imaging and machine learning algorithms

Diffusion dynamics of technological innovations in health: the case of an innovative medical imaging technology in France