Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jun 26, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Automated PROMISE V2 Scoring from PSMA PET/CT Reports Using Large Language Models: A Comparative Evaluation of Prompt

Tilman Speicher1, Isa Ethem Demirkol1, Arne Blickle1

  • 1Department of Nuclear Medicine, Saarland University-Medical Center, 66421 Homburg, Germany.

Current Oncology (Toronto, Ont.)
|June 25, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Outcome and Safety of Radioligand Therapy With [ 161 Tb]Tb-PSMA-617 After Re-Progression in mCRPC Patients Previously Treated With [ 177 Lu]Lu-PSMA-617.

Clinical nuclear medicine·2026
Same author

Zirconium-89-based PSMA PET/CT with delayed imaging for early detection of biochemical recurrence in prostate cancer patients with PSA ≤ 0.2 ng/mL: A feasibility study.

European journal of nuclear medicine and molecular imaging·2026
Same author

FDG PET/CT imaging in recurrent papillary thyroid cancer: does BRAF mutation status matter?

The quarterly journal of nuclear medicine and molecular imaging : official publication of the Italian Association of Nuclear Medicine (AIMN) [and] the International Association of Radiopharmacology (IAR), [and] Section of the Society of...·2026
Same author

Outcome and safety of single-dose docetaxel-augmented targeted radionuclide therapy using [¹⁷⁷Lu]Lu-PSMA-617 radioligand therapy in treatment-refractory mCRPC: preliminary evidence from a pilot experience.

EJNMMI research·2026
Same author

Dosimetric analysis reveals rapid clearance and low absorbed dose of [¹⁷⁷Lu]Lu-PSMA-617 in non-prostate cancers with high PSMA expression.

EJNMMI research·2026
Same author

Auger Emitting [<sup>161</sup>Tb]Tb-PSMA-617 Radioligand Therapy in mCRPC with Peritoneal Carcinomatosis.

Nuclear medicine and molecular imaging·2026
Same journal

Predicting Post-Radiotherapy Lymphocyte Recovery for Individualized Risk Stratification in Locally Advanced Esophageal Squamous Cell Carcinoma.

Current oncology (Toronto, Ont.)·2026
Same journal

From Adjunct to Essential: Integrating Supportive Care into Oncology.

Current oncology (Toronto, Ont.)·2026
Same journal

Next-Generation Sequencing in Differentiated Thyroid Cancer Patients Treated with Lenvatinib: Results and Challenges in Real-Life Practice.

Current oncology (Toronto, Ont.)·2026
Same journal

The Route of Administration Determines the Efficacy of Zinc in Preventing Radiation-Induced Oral Mucositis: A Systematic Review and Meta-Analysis.

Current oncology (Toronto, Ont.)·2026
Same journal

Mechanisms of Progression and Challenges for Intervention in the Natural History of Early Prostate Cancer: A Narrative Review.

Current oncology (Toronto, Ont.)·2026
Same journal

Maintenance Therapy in Acute Myeloid Leukemia: Current Perspectives and Future Directions.

Current oncology (Toronto, Ont.)·2026
See all related articles
This summary is machine-generated.

Large language models (LLMs) show promise for prostate cancer classification. Gemini 3 Flash achieved the highest agreement in PROMISE V2 scoring using detailed prompts.

Area of Science:

  • Medical Imaging and Artificial Intelligence
  • Oncology and Urology

Background:

  • Large language models (LLMs) are being evaluated for clinical applications.
  • Reliability of LLMs in supporting medical reporting, staging, and classification needs further research.

Purpose of the Study:

  • To evaluate and compare multiple LLMs for automated PROMISE V2 classification in prostate cancer.
  • To assess LLM performance on German-language PSMA PET/CT reports.

Main Methods:

  • Retrospective analysis of 126 German PSMA PET/CT reports.
  • Expert consensus established reference standards.
  • Five LLMs (GPT-4, DeepSeek-V3.2, Claude Sonnet 4.6, Gemini 3 Flash, Grok 4) were tested with short and long English prompts.

Main Results:

Keywords:
LLMPET/CTPROMISEPSMAlarge language modelprostate cancer

Related Experiment Videos

Last Updated: Jun 26, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

  • LLM performance improved with longer prompts (74.6-86.5%) compared to short prompts (36.5-79.4%).
  • Gemini 3 Flash demonstrated the highest agreement with the reference standard.
  • High agreement rates were observed across PROMISE V2 subcategories (miT, miN, miM).

Conclusions:

  • Contemporary LLMs show potential for deriving PROMISE V2 scores from clinical reports.
  • Detailed prompting significantly enhances LLM performance in prostate cancer classification.
  • LLMs can reliably support physicians in prostate cancer staging and classification.