M Pudasaini¹*, S Pella², T Leventouri¹, W Muhammad¹, (1) Florida Atlantic University, Boca Raton, FL, (2) FAU & 21st Century Oncology (Genesis Care), Boca Raton, FL,

• 
  M Pudasaini

SU-IePD-TRACK 5-4 (Sunday, 7/25/2021) 12:30 PM - 1:00 PM [Eastern Time (GMT-4)]

Purpose: To apply an artificial neural network (ANN) in order to predict equivalent uniform dose and tumor control probability in external beam radiation therapy of lung cancer treatment plans.

Methods: Retroactive lung cancer treatment plans of 100 patients were chosen for this study. The tumor control probability (TCP) of tumor target and equivalent uniform dose (EUD) were calculated for all plans using Niemierko’s EUD based linear-quadratic model. The expected outputs for ANN were calculated EUD and TCP. The inputs for ANN were treatment modality, location of tumor, prescribed dose, fractions, planning target volume (PTV), the maximum dose to the tumor, and mean doses to the OARs. All numeric data were normalized in the range 0 to 1, while non-numeric inputs were categorized into binary numbers (0 or 1). Our ANN is based on a Scaled Conjugate gradient algorithm with two hidden layers having 15-10 nodes with 11 inputs and 2 outputs. The data were employed 70% for training and 30% for validation and testing.

Results: We found EUD and TCP for the target volume with an overall regression value of 0.97. The Mean absolute difference for predicted and expected values for EUD and TCP are found 0.0510 ± 0.0704 and 0.0439 ± 0.0708 respectively. In some cases, we encountered a large error (~0.14), where TCP were 50% or lower in palliative treatment case.

Conclusion: These results indicate that ANN can be employed to predict TCP and EUD. The outcome of this research can be further improved by studying a large cohort of patient’s treatment plans.

ePosters

Keywords

Not Applicable / None Entered.

Taxonomy

TH- External Beam- Photons: Development (new technology and techniques)

Contact Email