Machine Learning Prediction of Pituitary Macroadenoma Consistency: Utilizing Demographic Data and Brain MRI...

Area of Science:

• Neurosurgery
• Medical Imaging
• Machine Learning

Background:

• Pituitary macroadenoma consistency impacts surgical outcomes, with non-soft tumors leading to complications.
• Predicting tumor consistency preoperatively is crucial for effective surgical planning.

Purpose of the Study:

• To develop and validate a machine learning model for predicting pituitary macroadenoma consistency.
• To enhance surgical planning and improve patient outcomes by identifying non-soft tumors.

Main Methods:

• Retrospective analysis of 70 pituitary macroadenoma patients.
• Utilized magnetic resonance imaging (MRI) data (diameter, ADC) and demographics (age, sex).
• Developed and evaluated a support vector machine (SVM) model for consistency prediction.

Main Results:

• The SVM model achieved an ROC AUC of 83.3%, demonstrating significant predictive capability.
• Key predictors for non-soft consistency included male sex and age ≤ 42.25 years.
• An open-access repository was created to facilitate external validation and model improvement.

Conclusions:

• The developed SVM model shows promise in predicting pituitary macroadenoma consistency.
• Accurate prediction can aid surgeons in planning procedures and mitigating risks.
• Further validation and collaboration are encouraged to enhance generalizability and clinical utility.