Shape Features of the Lesion Habitat to Differentiate Brain Tumor Progression from Pseudoprogression on Routine...

Area of Science:

• Radiology
• Neuro-oncology
• Medical imaging analysis

Background:

• Distinguishing pseudoprogression from tumor progression in glioblastoma is difficult.
• Current Response Assessment in Neuro-Oncology criteria rely on limited 2D measurements.
• This limitation hinders accurate treatment assessment and patient management.

Purpose of the Study:

• To investigate if quantitative 3D shape features of the lesion habitat can differentiate pseudoprogression from tumor progression.
• To develop a more comprehensive imaging analysis beyond simple diametric measurements.

Main Methods:

• Analysis of 105 glioblastoma studies (59 training, 46 testing).
• Extraction of 30 quantitative 3D shape features from T1WI enhancing lesions and T2WI/FLAIR hyperintensities.
• Utilized a support vector machine classifier with selected discriminative features.

Main Results:

• Identified two key local shape features (enhancing lesion curvature and perilesional region curvedness) as most discriminative.
• Achieved 91.5% accuracy on the training cohort and 90.2% accuracy on the independent test cohort.
• Demonstrated the ability to distinguish pseudoprogression from tumor progression.

Conclusions:

• Preliminary findings suggest 3D shape attributes of the lesion habitat can differentiate pseudoprogression and tumor progression.
• These quantitative shape features offer a potential tool for distinguishing radiographically similar pathologies.
• Further validation may improve glioblastoma treatment response assessment.