Conventional imaging-derived phenotypes of brain abscess and short-term clinical outcomes: a comparative study in...

To develop a composite imaging phenotype framework for brain abscess (BA) based on conventional imaging features, enabling systematic assessment of imaging heterogeneity, and to explore the distribution of these phenotypes across different age groups and their association with short-term clinical outcomes.

This retrospective study consecutively enrolled patients with surgically confirmed BA admitted to two hospitals between January 2019 and December 2025. Patients were stratified into pediatric and adult groups based on age. All patients underwent surgical treatment and standard anti-infective therapy. Based on preoperative conventional imaging, three-dimensional composite phenotypes were defined: structural complexity, spatial complexity, and aggressive features. These were integrated into an "imaging phenotypic burden" score. Analyses examined age-related differences in phenotype distribution, associations with pathogen profiles, and the relationship between phenotypic burden and in-hospital adverse outcomes.

Significant differences were observed in imaging phenotype distribution between children and adults. Structural complexity was more common in the pediatric group (40.0% vs. 9.8%, p = 0.008), while spatial complexity was higher in adults (58.5% vs. 15.0%, p < 0.001). Microbiological analysis indicated a predominance of Streptococcus infections in children (p = 0.002); however, no stable statistical association was found between pathogen category and any specific imaging phenotype. A significant dose-response relationship existed between phenotypic burden and adverse outcome rates (p = 0.004, trend test). The high-burden group (≥2 phenotypes) had a significantly higher adverse outcome rate compared to the low-burden group (30.0% vs. 6.5%, p = 0.017). Deep-seated lesions were the primary anatomical factor driving high phenotypic burden and spatial complexity.

The developed composite imaging phenotype framework effectively integrates information from conventional imaging to systematically assess BA heterogeneity. The imaging phenotypic burden is independently associated with short-term adverse outcomes. It serves as an intuitive, quantitative tool for early risk stratification, informing surgical timing decisions, and guiding intensive care resource allocation, demonstrating potential for clinical translation.