PET/CT Imaging Features for 289 Fever of Unknown Origin Patients

A 2026 retrospective cohort study of 289 adult FUO patients with a mean age of 59 years. The dataset, authored by Sule Ceylan, contains PET/CT imaging features used for unsupervised clustering to identify distinct imaging phenotypes, though these clusters showed low concordance with final clinical diagnoses.

Use Cases

• Developing unsupervised clustering models based on PET/CT features like organ involvement and SUVmax.
• Analyzing relationships between imaging phenotypes and clinical etiologies such as infections or rheumatologic diseases.
• Benchmarking consensus clustering approaches using metrics like Adjusted Rand Index and Mutual Information.
• Exploring the complementary role of PET/CT in FUO/IUO evaluation beyond direct diagnosis.

Strengths

• Includes data from 289 patients, providing a substantial cohort for analysis.
• Features were extracted from PET/CT scans, including specific metrics like maximum SUVmax and uptake patterns.
• Applies a consensus clustering approach using four different unsupervised algorithms for robust results.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• The dataset is very small at 14.8 KB, indicating limited scope, likely containing summary results rather than raw patient data.
• Data may reflect the specific biases inherent to the single-center, retrospective study design mentioned in the description.

Provenance

Source

figshare, author Sule Ceylan.

Collection Method

Retrospective cohort study where data were extracted from 18F-FDG PET/CT scans of adult FUO patients.

Freshness

Last updated 2026-05-29 04:50:48; freshness should be verified.