Radiomics in Vulvar Cancer: First Clinical Experience Using ¹⁸F-FDG PET/CT Images

Abstract

This study investigates whether radiomic features derived from preoperative positron emission tomography (PET) images could predict both tumor biology and prognosis in women with invasive squamous cell carcinoma of the vulva. Patients were retrospectively included when they had a unifocal primary cancer of ≥ 2.6 cm in diameter, had received a preoperativeF-fluorodeoxyglucose (F-FDG) PET/computed tomography (CT) scan followed by surgery and had at least six months of follow-up data. F-FDG-PET images were analyzed by semi-automatically drawing on the primary tumor in each PET image, followed by the extraction of 83 radiomic features. Unique radiomic features were identified by principal component analysis (PCA), after which they were compared with histopathology using non-pairwise group comparison and linear regression. Univariate and multivariate Cox regression analyses were used to correlate the identified features with progression-free survival (PFS) and overall survival (OS). Survival curves were estimated using the Kaplan-Meier method. Forty women were included. PCA revealed four unique radiomic features, which were not associated with histopathologic characteristics such as grading, depth of invasion, lymph-vascular space invasion and metastatic lymph nodes. No statistically significant correlation was found between the identified features and PFS. However, Moran's I, a feature that identifies global spatial autocorrelation, was correlated with OS ( = 0.03). Multivariate Cox regression analysis showed that extracapsular invasion of the metastatic lymph nodes and Moran's I were independent prognostic factors for PFS and OS. Our data show that PCA is usable to identify specific radiomic features. Although the identified features did not correlate strongly with tumor biology, Moran's I was found to predict patient prognosis. Larger studies are required to establish the clinical relevance of the observed findings.

Adding the temporal domain to PET radiomic features

“…Parametric features and dynamic features that did not show high correlations with static features were assessed. Unsupervised feature selection of the static feature set was performed following the approach of Collarino et al using redundancy filtering based on Pearson correlation coefficients (r>0.75) and principal component analysis (PCA) [21]. Survival curves were estimated using Kaplan-Meier analysis for the selected features dichotomized at their median and survival curves were compared using log-rank statistics.…”

Fusion of ultrasound and 3D single‐photon‐emission computed tomography/computed tomography to identify sentinel lymph nodes in vulvar cancer: feasibility study

“…Median overall time to perform fusion imaging was 32 (range, [25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40] min. Median duration of Step 1 (acquisition of 3D volumes and target identification on CT and SPECT) was 10 (range, 8-12) min; of Step 2 (coregistration and synchronization of SPECT/CT and ultrasound) was 10 (range, 7-10) min; and of Step 3 (fine tuning and SPECT/CT and ultrasound fusion imaging navigation) was 12 (range, [8][9][10][11][12][13][14][15][16][17][18][19] min. Time spent performing Steps 1 and 2 was similar in all 10 cases, whereas the time spent performing Step 3 reduced significantly from the first to the last examination (19 vs 12 min; Figure 4).…”

“…Standard treatment of early‐stage vulvar cancer consists of radical local excision of the tumor, sometimes with simultaneous plastic reconstruction, combined with sentinel lymph‐node biopsy and/or inguinofemoral lymphadenectomy, according to the findings on preoperative lymph‐node assessment. In particular, according to evidence reported recently in the literature, lymph‐node negative status on imaging could be the most important criterion to indicate a sentinel lymph‐node procedure, overcoming previous restrictive criteria and providing individualized risk assessment through advanced functional imaging tools. Sentinel lymph‐node biopsy is burdened by associated morbidity, while inguinofemoral lymphadenectomy is associated with significant short‐ and long‐term complications such as wound breakdown, wound infection, lymphocele, lymphedema, cellulitis and erysipelas.…”

Artificial intelligence applications for oncological positron emission tomography imaging

“…Principal component analysis (PCA) Prognosis: non-small cell lung cancer [65] Two-way clustering method was used to extract features which was proved more informative than that of PCA method [65]. Two-way clustering method vulvar cancer [81] Prognosis: non-small cell lung cancer [65] Linear discriminant analysis (LDA)…”