Incremental value of quantitative SPECT/CT integrated parameters in a multimodal machine-learning model for differentiating spinal tuberculosis from pyogenic spondylitis
Frontiers in Cellular and Infection Microbiology·
- DOI
- 10.3389/fcimb.2026.1897479
- PMID
- —
- PMCID
- —
- OpenAlex
- W7203667121
- Study type
- Journal article
- Publisher
- Frontiers Media SA
- Article type
- journal-article
- Integrity
- current
Why this research matters now
Distinguishing spinal tuberculosis from pyogenic spondylitis is clinically important because the two conditions require different treatment approaches. The findings suggest that quantitative SPECT/CT parameters may offer complementary diagnostic information to improve differential diagnosis, though multicenter validation and prospective studies are needed before clinical implementation.
Structured evidence summary
Research question
The study aimed to develop and validate a machine-learning model that integrates clinical, laboratory, imaging, and quantitative SPECT/CT parameters to distinguish spinal tuberculosis from pyogenic spondylitis, and to assess whether quantitative SPECT/CT parameters add diagnostic value beyond conventional information.
Study design
This was a retrospective study of 165 patients (96 with spinal tuberculosis and 69 with pyogenic spondylitis) confirmed by microbiological, molecular, or histopathological testing between September 2024 and May 2026. Machine-learning models (Elastic Net logistic regression and radial basis function support vector machine) were developed and internally validated using repeated stratified nested cross-validation.
Population and setting
The study included 165 patients with confirmed spinal tuberculosis or pyogenic spondylitis, diagnosed through microbiological, molecular, or histopathological methods. The setting and geographic location are not specified in the abstract.
Main findings
All seven integrated quantitative SPECT/CT parameters were significantly higher in pyogenic spondylitis than in spinal tuberculosis (all P < 0.01). The Elastic Net fusion model incorporating LBI SUVmax and LBI SUVmean achieved an AUC of 0.899, representing an increase of 0.110 (95% CI, 0.054–0.172) over the baseline model and a reduction in Brier score of 0.048 (95% CI, 0.024–0.070). The radial basis function SVM fusion model achieved an AUC of 0.957 but showed only marginal incremental value (0.021; 95% CI, −0.003 to 0.050) over its baseline.
Public-health relevance
Distinguishing spinal tuberculosis from pyogenic spondylitis is clinically important because the two conditions require different treatment approaches. The findings suggest that quantitative SPECT/CT parameters may offer complementary diagnostic information to improve differential diagnosis, though multicenter validation and prospective studies are needed before clinical implementation.
Important limitations
The study was retrospective and conducted at a single center with internal validation only. The authors explicitly state that clinical applicability requires confirmation through multicenter external validation and prospective studies.
GIDS interpretation
This article was classified under Tuberculosis and Diagnostics in the surveillance system. It describes a diagnostic differentiation tool rather than epidemiological surveillance data, and therefore provides context on tuberculosis diagnostic methods but does not represent a disease occurrence signal.
Related GIDS surveillance
Literature context does not validate, explain, or change a surveillance signal. Exact and contextual relationships are shown separately.
Evidence relationships
This article has 5 auditable classifier relationships to diseases, places, topics, and study design.