Performance changes in automated lesion detection under federated learning with sequential institution addition - Takeaways - MDSpire
Coming Soon: Introducing MDSpire News. Learn more
Conexiant’s news site is now MDSpire News. Learn more

Impact of Sequential Institution Integration on Automated Lesion Detection Performance in Federated Learning Systems

  • By

  • Yukihiro Nomura

  • Shouhei Hanaoka

  • Aiki Yamada

  • Tomomi Takenaga

  • Takahiro Nakao

  • Toshiya Nakaguchi

  • Takeharu Yoshikawa

  • Osamu Abe

  • September 27, 2026

Share

  • 1

    The study investigates the impact of sequentially adding institutions on lesion detection performance in federated learning systems.

  • 2

    Two types of CAD software were evaluated: one for detecting cerebral aneurysms and another for identifying brain metastases.

  • 3

    The federated learning framework allows institutions to train models collaboratively while keeping data locally stored.

  • 4

    Three integration approaches were assessed: from-scratch training, fine-tuning, and full fine-tuning.

  • 5

    The study compares sequential client addition with an 'At-once' condition for model training.

Original Source(s)

Performance changes in automated lesion detection under federated learning with sequential institution addition

  • by Yukihiro Nomura, Shouhei Hanaoka, Aiki Yamada, Tomomi Takenaga, Takahiro Nakao, Toshiya Nakaguchi, Takeharu Yoshikawa, Osamu Abe

  • September 27, 2026

Related Content

Can brain age sharpen dementia risk?

MRI-based estimates showed differing prognostic value according to baseline diagnosis.

Late enhancement and washout maps for distinguishing tumor progression from radiation necrosis in malignant brain tumors: an exploratory study

Notch-referenced mediolateral verification of tibial tunnel position in anterior cruciate ligament reconstruction: a 3D imaging validation study