Radiomics for detecting prostate cancer bone metastases invisible in CT: a proof-of-concept study

Abstract

Objectives: To investigate, in patients with metastatic prostate cancer, whether radiomics of computed tomography (CT) image data enables the differentiation of bone metastases not visible on CT from unaffected bone using ⁶⁸ Ga-PSMA PET imaging as reference standard.

Methods: In this IRB-approved retrospective study, 67 patients (mean age 71 ± 7 years; range: 55-84 years) showing a total of 205 ⁶⁸ Ga-PSMA-positive prostate cancer bone metastases in the thoraco-lumbar spine and pelvic bone being invisible in CT were included. Metastases and 86 ⁶⁸ Ga-PSMA-negative bone volumes in the same body region were segmented and further post-processed. Intra- and inter-reader reproducibility was assessed, with ICCs < 0.90 being considered non-reproducible. To account for imbalances in the dataset, data augmentation was performed to achieve improved class balance and to avoid model overfitting. The dataset was split into training, test, and validation set. After a multi-step dimension reduction process and feature selection process, the 11 most important and independent features were selected for statistical analyses.

Results: A gradient-boosted tree was trained on the selected 11 radiomic features in order to classify patients' bones into bone metastasis and normal bone using the training dataset. This trained model achieved a classification accuracy of 0.85 (95% confidence interval [CI]: 0.76-0.92, p < .001) with 78% sensitivity and 93% specificity. The tuned model was applied on the original, non-augmented dataset resulting in a classification accuracy of 0.90 (95% CI: 0.82-0.98) with 91% sensitivity and 88% specificity.

Conclusion: Our proof-of-concept study indicates that radiomics may accurately differentiate unaffected bone from metastatic bone, being invisible by the human eye on CT.

Key points: • This proof-of-concept study showed that radiomics applied on CT images may accurately differentiate between bone metastases and metastatic-free bone in patients with prostate cancer. • Future promising applications include automatic bone segmentation, followed by a radiomics classifier, allowing for a screening-like approach in the detection of bone metastases.

Keywords: Bone metastases; Computed tomography; Prostate cancer; Radiomics; Texture analysis.

Fig. 1

Flowchart of the study cohort

Fig. 2

Representative examples of 3D bone volume segmentations of L4 (a), the right iliac bone (b), and the sacrum (c)

Fig. 3

Correlogram illustrating the auto- and cross-correlation of the 105 most important features to classify metastatic and normal bone. Features were recorded after hierarchical clustering for depicting different feature clusters. Eleven clusters of radiomic features were identified (rectangular boxes). Blue points indicate positive correlation, red points negative correlation. The larger the points and the darker the color, the higher the correlation between two variables

Fig. 4

Graph represents receiver operating characteristic (ROC) analysis (a) and the calibration plot (b) for the trained machine learning algorithm in order to differentiate between bone metastases and normal bone. The ROC analyses indicate accuracy, sensitivity, and specificity of the gradient-boosted tree trained on the selected 11 most important radiomic features and applied on the independent test dataset. The calibration plot shows the calibration in terms of agreement between the predicted and the actual probability of bone metastases

Fig. 5

CT and corresponding PET/CT in three representative patients with metastatic bone disease from PCa. Can you identify the bone metastases in the upper (a) and mid (b) thoracic spine, the inferior part of the sacrum (c), and the right iliac bone (d) on CT only, without the additional metabolic information from PET? Corresponding PET/CT images clearly show high ⁶⁸ Ga-PSMA uptake of the bone metastases in the aforementioned skeletal regions (e–h). Note additional ⁶⁸ Ga-PSMA-positive lymph node metastases along the left iliac vessel axis (g, h)