No abstract

Background Accurate segmentation of pelvic bones is an initial step to achieve accurate detection and localisation of pelvic bone metastases. This study presents a deep learning-based approach for automated segmentation of normal pelvic bony structures in multiparametric magnetic resonance imaging (mpMRI) using a 3D convolutional neural network (CNN). Methods This retrospective study included 264 pelvic mpMRI data obtained between 2018 and 2019. The manual annotations of pelvic bony structures (which included lumbar vertebra, sacrococcyx, ilium, acetabulum, femoral head, femoral neck, ischium, and pubis) on diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) images were used to create reference standards. A 3D U-Net CNN was employed for automatic pelvic bone segmentation. Additionally, 60 mpMRI data from 2020 were included and used to evaluate the model externally. Results The CNN achieved a high Dice similarity coefficient (DSC) average in both testing (0.80 [DWI images] and 0.85 [ADC images]) and external (0.79 [DWI images] and 0.84 [ADC images]) validation sets. Pelvic bone volumes measured with manual and CNN-predicted segmentations were highly correlated (R2 value of 0.84–0.97) and in close agreement (mean bias of 2.6–4.5 cm3). A SCORE system was designed to qualitatively evaluate the model for which both testing and external validation sets achieved high scores in terms of both qualitative evaluation and concordance between two readers (ICC = 0.904; 95% confidence interval: 0.871–0.929). Conclusions A deep learning-based method can achieve automated pelvic bone segmentation on DWI and ADC images with suitable quantitative and qualitative performance.

Introduction Prostatic stromal tumours of uncertain malignant potential (STUMPs) are rare prostate tumours. The purpose of this study was to investigate the magnetic resonance imaging features of STUMPs. Methods A total of 12 patients with STUMP confirmed with pathology who underwent MRI from 2012 to 2020 were retrospectively reviewed. Pathological characteristics including histopathology and immunohistochemistry were also recorded. Results Among 12 STUMPs, the tumours were detected in the peripheral zone (41.7%[n = 5]) and transitional zone (58.3% [n = 7]) of the prostate. 8 cases (66.7%) were round shape. All lesions were well‐defined and compressed the adjacent structures but without signs of an invasion. Homogeneous T1WI and heterogeneous T2WI signals were observed in the STUMPs. The tumours were mainly composed of solid components, while intratumoral cystic change (58.3%[n = 7]) and haemorrhage (8.3%[n = 1]) were seen. 10 cases(83.3%) were seen as relatively high DWI signal, while 2 cases(16.7%) with no increase in DWI. The mean ADC value was 1.084 ± 0.193 (range: 0.864–1.489 × 10−3 mm2/s). STUMPs had heterogeneous enhancement, with persistent or gradual enhancement. In immunohistochemical staining, Vim, CD34, PR and SMA were positive in the majority of STUMPs. Conclusion MRI features of STUMP are presented as regular, well‐defined and isolated prostatic mass with intact pseudocapsule. The presence of heterogeneous T2WI signal, intratumoral cystic change, slightly low mean ADC value and persistent or gradual enhancement may help predict the STUMPs.

No abstract

No abstract