Intra- and Inter-Fractional Variation Prediction of Lung Tumors Using Fuzzy Deep Learning.

Area of Science:

• Medical Physics
• Radiotherapy
• Artificial Intelligence

Background:

• Accurate tumor movement prediction is essential for radiotherapy precision.
• Respiration-induced tumor motion presents intra-fractional and inter-fractional variations.
• Existing methods for inter-fractional variation prediction are limited and computationally intensive.

Purpose of the Study:

• To develop a novel predictor for both intra- and inter-fractional tumor movement variations.
• To enhance prediction accuracy and reduce computation time in radiotherapy.
• To address limitations in mathematization and prediction of inconstant inter-fractional variations.

Main Methods:

• Proposed a new model: intra- and inter-fraction fuzzy deep learning (IIFDL).
• IIFDL incorporates fuzzy deep learning (FDL) with breathing clustering.
• Evaluated IIFDL's performance against existing prediction methods.

Main Results:

• IIFDL improved root-mean-square error (RMSE) by 29.98% and prediction overshoot by 70.93% compared to existing methods.
• Further enhanced average RMSE and overshoot by 59.73% and 83.27%, respectively.
• Achieved a significantly reduced average computation time of 1.54 ms for both variation types.

Conclusions:

• The proposed IIFDL model accurately predicts intra- and inter-fractional tumor movements.
• IIFDL offers substantial improvements in prediction accuracy and significant reductions in computation time.
• IIFDL shows potential for real-time estimation and enhanced tracking techniques in radiotherapy.