CT-based radiomics model with machine learning for predicting primary treatment failure in diffuse large B-cell...

Area of Science:

• Radiology
• Oncology
• Artificial Intelligence

Background:

• Identifying Diffuse Large B-Cell Lymphoma (DLBCL) patients likely to be refractory to first-line therapy is essential for personalized treatment strategies.
• Early identification allows for timely alternative therapeutic options to improve patient prognosis.

Purpose of the Study:

• To evaluate the efficacy of a CT-based radiomics approach combined with machine learning in predicting Primary Treatment Failure (PTF)-DLBCL from initial imaging.
• To develop a predictive model for identifying patients with DLBCL who may not respond to standard first-line therapy.

Main Methods:

• A cohort of 26 refractory and 26 non-refractory DLBCL patients was analyzed, with 180 lymph nodes segmented manually for reproducibility.
• 1218 radiomic features were extracted from CT images.
• A Random Forests machine learning classifier was trained and tested on 70% and 30% of the data, respectively.

Main Results:

• The developed model achieved a mean accuracy of 73%, sensitivity of 62%, and specificity of 82% in distinguishing refractory from non-refractory patients.
• The area under the receiver operating characteristic curve (AUC) was 0.83 and 0.79 for the two independent readers.
• The model demonstrated good performance in predicting PTF-DLBCL.

Conclusions:

• CT-based radiomics analysis utilizing machine learning can effectively identify patients with PTF-DLBCL prior to treatment initiation.
• This approach holds promise for improving patient selection and optimizing therapeutic strategies in DLBCL management.