Automated Detection and Segmentation of Bone Metastases on Spine MRI Using U-Net: A Multicenter Study.

OBJECTIVE: To develop and evaluate a deep learning model for automated segmentation and detection of bone metastasis on spinal MRI. MATERIALS AND METHODS: We included whole spine MRI scans of adult patients with bone metastasis: 662 MRI series from 302 patients (63.5 ± 11.5 years; male:female, 151:151) from three study centers obtained between January 2015 and August 2021 for training and internal testing (random split into 536 and 126 series, respectively) and 49 MRI series from 20 patients (65.9 ± 11.5 years; male:female, 11:9) from another center obtained between January 2018 and August 2020 for external testing. Three sagittal MRI sequences, including non-contrast T1-weighted image (T1), contrast-enhanced T1-weighted Dixon fat-only image (FO), and contrast-enhanced fat-suppressed T1-weighted image (CE), were used. Seven models trained using the 2D and 3D U-Nets were developed with different combinations (T1, FO, CE, T1 + FO, T1 + CE, FO + CE, and T1 + FO + CE). The segmentation performance was evaluated using Dice coefficient, pixel-wise recall, and pixel-wise precision. The detection performance was analyzed using per-lesion sensitivity and a free-response receiver operating characteristic curve. The performance of the model was compared with that of five radiologists using the external test set. RESULTS: The 2D U-Net T1 + CE model exhibited superior segmentation performance in the external test compared to the other models, with a Dice coefficient of 0.699 and pixel-wise recall of 0.653. The T1 + CE model achieved per-lesion sensitivities of 0.828 (497/600) and 0.857 (150/175) for metastases in the internal and external tests, respectively. The radiologists demonstrated a mean per-lesion sensitivity of 0.746 and a mean per-lesion positive predictive value of 0.701 in the external test. CONCLUSION: The deep learning models proposed for automated segmentation and detection of bone metastases on spinal MRI demonstrated high diagnostic performance.

Assessing visibility and bone changes of spinal metastases in CT scans: a comprehensive analysis across diverse cancer types.

OBJECTIVES: To analyze the characteristics of spinal metastasis in CT scans across diverse cancers for effective diagnosis and treatment, using MRI as the gold standard. METHODS: A retrospective study of 309 patients from four centers, who underwent concurrent CT and spinal MRI, revealing spinal metastasis, was conducted. Data on metastasis including total number, volume, visibility on CT (visible, indeterminate, or invisible), and type of bone change were collected. Through chi-square and Mann-Whitney U tests, we characterized the metastasis across diverse cancers and investigated the variation in the intra-individual ratio representing the percentage of lesions within each category for each patient. RESULTS: Out of 3333 spinal metastases from 309 patients, 55% were visible, 21% indeterminate, and 24% invisible. Sclerotic and lytic lesions made up 47% and 43% of the visible and indeterminate categories, respectively. Renal cell carcinoma (RCC), prostate cancer, and hepatocellular carcinoma (HCC) had the highest visibility at 86%, 73%, and 67% (p < 0.0001, p < 0.0001, and p = 0.003), while pancreatic cancer was lowest at 29% (p < 0.0001). RCC and HCC had significantly high lytic metastasis ratios (interquartile range (IQR) 0.96-1.0 and 0.31-1.0, p < 0.001 and p = 0.005). Prostate cancer exhibited a high sclerotic lesion ratio (IQR 0.52-0.97, p < 0.001). About 39% of individuals had invisible or indeterminate lesions, even with a single visible lesion on CT. The intra-individual ratio for indeterminate and invisible metastases surpassed 18%, regardless of the maximal size of the visible metastasis. CONCLUSIONS: This study highlights the variability in characteristics of spinal metastasis based on the primary cancer type through unique lesion-centric analysis.

A new knotless parametrial tissue ligation technique for safe total laparoscopic hysterectomy.

OBJECTIVE: Parametrial tissue ligation during total laparoscopic hysterectomy (TLH) is important in large uteri with large vessels. METHODS: A retrospective study was performed at Asan Medical Center for comparing TLH performed with a new knotless parametrial tissue ligation method and conventional laparoscopic-assisted vaginal hysterectomy (LAVH) from March 2019 to August 2021. For TLH, after anterior colpotomy, the parametrial tissue was ligated by anchoring the suture and making a loop in one direction three times using 1-0 V-LocTM 180 (Covidien, Mansfield, MA, USA) suture. Subsequently, the cranial part of the loop was cut using an endoscopic device. RESULTS: A total of 119 and 178 patients were included in the TLH and LAVH groups, respectively. The maximal diameter of the uterus was larger in the TLH group (106.29±27.16 cm) than in the LAVH group (99.00±18.92 cm, P=0.01). The change in hemoglobin (Hb) level was greater in the LAVH group than in the TLH group (P<0.001). The weight of the removed uterus was greater in the TLH group than in the LAVH group (431.95±394.97 vs. 354.94±209.52 g; P=0.03). However, when the uterine weight was >1,000 g, the operative times and change in Hb levels were similar between the two groups. In both groups, no ureteral complications occurred during or after surgery. CONCLUSION: Knotless parametrial tissue ligation using 1-0 V-LocTM 180 suture in TLH can be safely applied, even in cases with large uteri, without increased risks of ureteral injury or uterine bleeding.