Breast Tumor Characterization Using [18F]FDG-PET/CT Imaging Combined with Data Preprocessing and Radiomics

Abstract

Background: This study investigated the performance of ensemble learning holomic models for the detection of breast cancer, receptor status, proliferation rate, and molecular subtypes from [18F]FDG-PET/CT images with and without incorporating data pre-processing algorithms. Additionally, machine learning (ML) models were compared with conventional data analysis using standard uptake value lesion classification. Methods: A cohort of 170 patients with 173 breast cancer tumors (132 malignant, 38 benign) was examined with [18F]FDG-PET/CT. Breast tumors were segmented and radiomic features were extracted following the imaging biomarker standardization initiative (IBSI) guidelines combined with optimized feature extraction. Ensemble learning including five supervised ML algorithms was utilized in a 100-fold Monte Carlo (MC) cross-validation scheme. Data pre-processing methods were incorporated prior to machine learning, including outlier and borderline noisy sample detection, feature selection, and class imbalance correction. Feature importance in each model was assessed by calculating feature occurrence by the R-squared method across MC folds. Results: Cross validation demonstrated high performance of the cancer detection model (80% sensitivity, 78% specificity, 80% accuracy, 0.81 area under the curve (AUC)), and of the triple negative tumor identification model (85% sensitivity, 78% specificity, 82% accuracy, 0.82 AUC). The individual receptor status and luminal A/B subtype models yielded low performance (0.46–0.68 AUC). SUVmax model yielded 0.76 AUC in cancer detection and 0.70 AUC in predicting triple negative subtype. Conclusions: Predictive models based on [18F]FDG-PET/CT images in combination with advanced data pre-processing steps aid in breast cancer diagnosis and in ML-based prediction of the aggressive triple negative breast cancer subtype.

Multiparametric Integrated 18F-FDG PET/MRI-Based Radiomics for Breast Cancer Phenotyping and Tumor Decoding

“…Whilst radiomics-based analyses of breast cancer have become a well-investigated research focus over the past few years, the majority of studies were based either on mammographic or MR-based imaging [3,8,34,35,37]. Only a few studies included PET-based data in their analysis, with Krajnc et al and Huang et al demonstrating promising results in recent publications [13,38]. Krajnc et al used 18 F-FDG PET/CT imaging combined with data preprocessing and radiomics analysis to characterize breast tumors.…”

“…Krajnc et al used 18 F-FDG PET/CT imaging combined with data preprocessing and radiomics analysis to characterize breast tumors. Notably, their predictive models achieved good results in breast cancer detection (AUC 0.82) and identification of triple-negative tumors (AUC 0.82), yet determination of luminal A/B subtype and the individual receptor status yielded low performance, with AUCs ranging from 0.46-0.68 [13]. Their results underline the potential of PET-based metabolic data for radiomic signature derivation and indicate that CT-based data may not provide a sufficiently comprehensive platform for breast cancer assessment.…”

“…As recently reported in a rapid review on radiomics and breast cancer, most studies put the focus on breast MRI with the main emphasis on morphology, contrast enhancement kinetics and restricted diffusivity as features. While functional features derived from MRI and 18 F-FDG PET/CT have been shown to be valuable in the detection and characterization of breast cancer and lymph node involvement, only a very limited number of studies investigated the potential of integrated 18 F-FDG PET/MRI for radiomics applications in breast cancer [11][12][13][14]. Hence, the aim of this study is to assess simultaneous 18 F-FDG PET/MRI of the breast as a platform for comprehensive radiomics analysis for breast cancer subtype analysis, hormone receptor status, proliferation rate and lymphonodular and distant metastatic spread.…”

Morpho-Molecular Metabolic Analysis and Classification of Human Pituitary Gland and Adenoma Biopsies Based on Multimodal Optical Imaging

“…In each fold, feature ranking and selection of the highest-ranking 10 features was performed via R-squared ranking in the training set of each MC fold. The selected features were then also selected for the validation subset [37,39]. This step was necessary to minimize the chances of overfitting.…”

Clinical Data Classification With Noisy Intermediate Scale Quantum Computers

“…This study incorporated three open-access clinical datasets that have been presented and evaluated in various contexts [24][25][26] . Each dataset underwent redundancy reduction by correlation matrix analysis 27 followed by a 10-fold cross-validation split with a training-validation ratio of 80%-20% 16 . Training sets of the folds were subjects of feature ranking analysis 28 and the highestranking eight as well as 16 (if available) features were selected for further analysis.…”