No abstract

Background. Ciliated hepatic foregut cysts (CHFCs) are uncommon cystic lesions within the liver. CHFCs can undergo a malignant transformation to form a primary squamous cell carcinoma of the liver. The true incidence and natural history of CHFCs is unknown and the risk of malignant transformation is unclear. We present a single centre's experience of CFHC management. Methods. A retrospective review of a departmental database identified all patients with CHFCs over a 4 year time period. Patients with CHFCs confirmed on histological assessment or suspected on radiological imaging were included in this study. Clinical information regarding patient demographics, symptomatic presentation, surgical management and histopathological features were noted. The radiological characteristics of CHFCs were recorded and the malignant transformation rate was calculated. Results. 15 patients with CHFC were identified (7 histologically confirmed and 8 radiologically suspected cases). All patients were asymptomatic and the CHFCs were incidental findings. No CHFC developed an interval change in cyst features or underwent a malignant transformation during follow up. MRI serves as the most sensitive modality to diagnose CHFC. Conclusions. CHFCs may be more prevalent than previously reported. Definitive management should encompass a patient centred discussion regarding the merits of long term follow up with serial imaging versus resection on an individual basis once CHFC is diagnosed.

We propose to take on the problem of Word Sense Disambiguation (WSD). In language, words of the same form can take different meanings depending on context. While humans easily infer the meaning or gloss of such words by their context, machines stumble on this task. As such, we intend to replicated and expand upon the results of Huang et al.'s GlossBERT, a model which they design to disambiguate these words (Huang et al., 2019). Specifically, we propose the following augmentations: data-set tweaking (α hyper-parameter), ensemble methods, and replacement of BERT with BART and ALBERT. The following GitHub repository contains all code used in this report, which extends on the code made available by Huang et al.: https://github.com/nkhlp/glossBERT. Additionally, the following links to a short presentation: https://youtu.be/X2OxgcF7lsM.