DOI: 10.1097/mnm.0000000000002265 ISSN: 0143-3636

Radiomics analysis of dynamic carbon-11 methionine PET/computed tomography for differentiating benign and malignant musculoskeletal lesions: a preliminary clinical assessment

Yuka Hiroshima, Takayoshi Shinya, Badrakhzul Zayabat, Yoichi Otomi, Shunichi Toki, Yoshiro Abe, Toshihiko Nishisho, Koichi Sairyo, Keisuke Ishizawa, Masafumi Harada

Objective

To evaluate the diagnostic performance of radiomics features and maximum standardized uptake value (SUV max ) extracted from dynamic carbon-11 methionine (C-11 MET) PET/computed tomography (PET/CT) in differentiating malignant from benign musculoskeletal lesions (MSLs) and compare their performance with apparent diffusion coefficient (ADC) values from magnetic resonance (MR) images.

Methods

This retrospective study included 15 patients (malignant: seven and benign: eight) with MSLs who underwent dynamic C-11 MET PET scans [5–15 (phase 1), 15–25 (phase 2), and 25–35 (phase 3) min postinjection of C-11 MET] and conventional acquisition (35 min postinjection). For semiquantitative analysis of MET uptake, we used SUV max . For radiomics analysis, 109 features were extracted from the volume of interest for the entire target tumor in each dynamic phase. ADC values were measured from MR images. Diagnostic capacities were analyzed using the Mann–Whitney U test and receiver operating characteristic curve analyses.

Results

SUV max for each dynamic phase and conventional acquisition was significantly higher in the malignant group than in the benign group ( P  < 0.05). In radiomics analysis, 70, 69, and 69 features in phases 1, 2, and 3, respectively, were significantly different between malignant and benign groups [ P  < 0.05; area under the ROC curves (AUCs) ≥ 0.821]. DeLong’s test showed no significant difference in AUCs between SUV max and any radiomics feature in each dynamic phase. ADC values did not differ significantly between malignant and benign groups.

Conclusion

Dynamic C-11 MET PET/CT has potential for differentiating malignant from benign MSLs and supports the utility of radiomics analyses and SUV max .

More from our Archive