MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel
  1. Home
  2. Meeting Program
  3. All
  4. Integrating Gross Tumor Volume and Margin Features to Predict Treatment Response for Locally Advanced Rectal Cancer Patients

Integrating Gross Tumor Volume and Margin Features to Predict Treatment Response for Locally Advanced Rectal Cancer Patients

M diMayorca1*, Y Zhang2, L Shi3, X Sun4, S Jabbour5, Y Zhang6, N Yue7, K Nie8, (1) Rutgers Cancer Institute of New Jersey, New Brunswick, NJ, (2) University of California Irvine, Irvine, CA, (3) Sir Run Run Shaw Hospital, Zhejiang University, Hangzhou, CN, (4) Sir Run Run Shaw Hospital, Zhejiang University, Hangzhou, CN, (5) Rutgers Cancer Institute Of New Jersey, New Brunswick, NJ, (6) Rutgers Cancer Institute of New Jersey, New Brunswick, NJ, (7) Rutgers Cancer Institute of New Jersey, New Brunswick, NJ, (8) Rutgers Cancer Institute of New Jersey, New Brunswick, NJ

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To investigate the effectiveness using combined gross tumor volume (GTV) and environmental margin radiomics features in predicting the treatment response to neoadjuvant chemoradiation therapy (nCRT) in patients with locally advanced rectal cancer (LARC).


Methods: This retrospective study included 102 patients with locally advanced rectal cancer. All patients have pre-treatment MRI with anatomical T1 & T2, diffusion-weighted-images (DWI) with multiple b-values, and dynamic-contrast-enhanced (DCE) MRI. Patients were randomly divided into training and testing sets with a ratio of 3:1. Radiomics features were extracted based on GTV, GTV+3mm and GTV+6mm margins. The radiomics signatures were developed to predict good response (GR) to nCRT based on tumor regression grading (TRG) and pathological complete response (pCR) status. The predictive performance was evaluated by receiver operator characteristic (ROC) curve analysis.


Results: The radiomics signature derived from GTV+3mm margin showed generally better prognostic power than GTV alone for both GR and pCR prediction. The performance with GTV+6mm margin analysis, however, was not necessarily improved. The DCE-MRI showed the best prognostic power in predicting pCR with a 3mm margin, which generated an AUC of 0.93 in the testing dataset.


Conclusion: For both GR and pCR, the optimal models are constructed with image features extracted based on volumes combining both on gross tumor volume and 3mm margin. These suggest the importance of involving both the tumor and environmental information into radiomics analysis. The optimal amount of environmental information to be included in analysis warrants more investigation.

Keywords

MRI, ROC Analysis

Taxonomy

IM- MRI : Radiomics

Contact Email

Additional Links