Predicting the Response of Patients Treated with ¹⁷⁷Lu-DOTATATE Using Single-photon Emission Computed...

Area of Science:

• Medical Imaging
• Radiomics
• Oncology

Background:

• Neuroendocrine tumors (NETs) require effective treatment monitoring.
• Lutetium-177 (¹⁷⁷Lu)-DOTATATE is a targeted radionuclide therapy for NETs.
• Predicting treatment response is crucial for patient management.

Purpose of the Study:

• To evaluate the efficacy of radiomics and clinical features from SPECT and SPECT-CT imaging.
• To predict the response to ¹⁷⁷Lu-DOTATATE treatment in NET patients.
• To identify optimal imaging modalities and features for response prediction.

Main Methods:

• Segmentation of tumor volumes from 61 SPECT and 41 SPECT-CT images of 22 NET patients.
• Extraction of 871 radiomics and clinical features.
• Feature selection using Maximum Relevance Minimum Redundancy (mRMR).
• Classification of treatment response using Decision Tree (DT), Random Forest (RF), K-nearest neighbor (KNN), and Support Vector Machine (SVM) models.

Main Results:

• Random Forest (RF) classifier achieved the highest accuracies: 64% for SPECT and 83% for SPECT-CT.
• SPECT-CT image classification accuracies: DT (79%), KNN (74%), SVM (67%).
• SPECT images showed poor accuracy (≈64%), indicating unsuitability for response prediction.

Conclusions:

• SPECT-CT imaging, analyzed with radiomics and machine learning, can predict ¹⁷⁷Lu-DOTATATE treatment response in NET patients.
• Anatomical and gray-level features from SPECT-CT are key predictors.
• This approach supports personalized treatment strategies for NETs.