Multi-task learning for predicting pulmonary nodule growth and follow-up volume

Key Points

• Prediction accuracy for pulmonary nodule growth improves significantly with the multi-task learning model.
• An enhancement of up to 20% in prediction accuracy was achieved, benefiting clinical decision-making.
• Multi-task learning approach utilizes deformation-texture features for predictive modeling of nodule growth.
• This method may enable more effective clinical surveillance, though external validation is needed.

Abstract