Training Robust T1-Weighted Magnetic Resonance Imaging Liver Segmentation Models Using Ensembles of Datasets with Different Contrast Protocols and Liver Disease Etiologies.

Image segmentation of the liver is an important step in several treatments for liver cancer. However, manual segmentation at a large scale is not practical, leading to increasing reliance on deep learning models to automatically segment the liver. This manuscript develops a deep learning model to segment the liver on T1w MR images. We sought to determine the best architecture by training, validating, and testing three different deep learning architectures using a total of 819 T1w MR images gathered from six different datasets, both publicly and internally available. Our experiments compared each architecture's testing performance when trained on data from the same dataset via 5-fold cross validation to its testing performance when trained on all other datasets. Models trained using nnUNet achieved mean Dice-Sorensen similarity coefficients > 90% when tested on each of the six datasets individually. The performance of these models suggests that an nnUNet liver segmentation model trained on a large and diverse collection of T1w MR images would be robust to potential changes in contrast protocol and disease etiology.

Uterine leiomyosarcoma management, outcome, and associated molecular biomarkers: a single institution's experience.

BACKGROUND: Uterine leiomyosarcoma (ULMS) is an aggressive, rapidly progressive tumor lacking clinical and molecular predictors of outcome. METHODS: ULMS patients (n = 349) were classified by disease status at presentation to MDACC as having intra-abdominal (n = 157) or distant metastatic disease (n = 192). Patient, tumor, treatment, and outcome variables were retrospectively retrieved. Formalin-fixed, paraffin-embedded tumor and control tissues from these patients (n = 109) were assembled in a tissue microarray and evaluated for hormone receptors and markers of angiogenesis, cell-cycle progression and survival. Patient, tumor, and treatment variables were correlatively analyzed. RESULTS: The 5- and 10-year disease-specific survival (DSS) for the cohort was 42 and 27 %, respectively. Patients with primary intra-abdominal tumors had better outcomes than those with recurrent intraperitoneal tumors. Whites had a more favorable prognosis. In patients with intra-abdominal tumors, only mitotic count >10M/10HPF portended poorer prognosis. Patients with pulmonary metastasis had improved outcomes with "curative" metastasectomy. ULMS samples exhibited loss of ER and PR expression, overexpressed Ki-67, and altered p53, Rb, p16, cytoplasmic ß-catenin, EGFR, PDGFR-α, PDGFR-ß, and AXL levels. Metastatic tumors had increased VEGF, Ki-67, and survivin expression versus localized disease. Survivin and ß-catenin expression were associated with intraperitoneal recurrence; high bcl-2 expression predicted longer DSS. CONCLUSIONS: Analysis of both clinicopathologic factors and immunohistochemical biomarkers in ULMS identified several prognostic clinical and molecular factors, suggesting that further study may lead to improved ULMS understanding and treatment.