Assessing Tumor Morphological Complexity Using Fractal Analysis of Contrast-Enhanced CT for Risk Stratification in...

To evaluate whether tumor morphological complexity, quantified via fractal analysis of contrast-enhanced CT images, can support risk stratification in pediatric neuroblastoma.

This retrospective study included 222 pediatric patients with pathologically confirmed neuroblastoma. Tumor regions of interest (ROIs) were manually delineated slice-by-slice on contrast-enhanced CT images at initial diagnosis. Fractal dimension (FD) values were calculated using the box-counting method across multiple spatial scales (ε = 2, 4, 8, 16, and 32) in MATLAB. Extracted metrics included the two-dimensional maximum, mean, minimum, and median FD values, along with the three-dimensional global FD. Associations between FD and clinical/pathological variables, including MYCN amplification, Shimada histology, International Neuroblastoma Risk Group (INRG) stage, Children's Oncology Group (COG) risk classification, and overall survival, were statistically assessed.

FD values were significantly higher in tumors with MYCN amplification and unfavorable Shimada histology (P < 0.05). Significant differences in FD were also observed among INRG stages, especially between L1 vs. L2 and L1 vs. M (P < 0.05), with L2 and M stage tumors exhibiting greater morphological complexity. Moreover, FD metrics increased progressively across low-, intermediate-, and high-risk COG groups (P < 0.05). Multivariate Cox proportional hazards regression analysis revealed that only global FD was independently associated with overall survival (P = 0.021).

FD metrics derived from contrast-enhanced CT images are significantly associated with established clinical/pathological risk factors and overall survival in pediatric neuroblastoma. FD may serve as a non-invasive imaging biomarker to assist in risk stratification and clinical decision-making.