Prediction Of Liver Function Research Articles

Applied machine learning for liver surgery

Abstract Background and objectives: Both hepatic functional reserve and the underlying histology are important determinants in the preoperative risk evaluation before major hepatectomies. In this project we developed a new approach that implements cutting-edge research in machine learning and nevertheless is cheap and easily applicable in a routine clinical setting is needed. Methods: After splitting the study population into a training and test set we trained a convolutional neural network to predict the liver function as determined by hepatobiliary mebrofenin scintigraphy and single photon emission computer tomography (SPECT) imaging. Results: We developed a workflow for predicting liver function from routine CT imaging data using convolutional neural networks. We also evaluated in how far transfer learning and data augmentation can help to solve remaining manual data pre-processing steps and implemented the developed workflow in a clinical routine setting. Conclusion: We propose a robust semiautomatic end-to-end classification workflow for abdominal CT scans for the prediction of liver function based on a deep convolutional neural network model that shows reliable and accurate results even with limited computational resources.

Viability Assessment in Liver Transplantation-What Is the Impact of Dynamic Organ Preservation?

Based on the continuous increase of donor risk, with a majority of organs classified as marginal, quality assessment and prediction of liver function is of utmost importance. This is also caused by the notoriously lack of effective replacement of a failing liver by a device or intensive care treatment. While various parameters of liver function and injury are well-known from clinical practice, the majority of specific tests require prolonged diagnostic time and are more difficult to assess ex situ. In addition, viability assessment of procured organs needs time, because the development of the full picture of cellular injury and the initiation of repair processes depends on metabolic active tissue and reoxygenation with full blood over several hours or days. Measuring injury during cold storage preservation is therefore unlikely to predict the viability after transplantation. In contrast, dynamic organ preservation strategies offer a great opportunity to assess organs before implantation through analysis of recirculating perfusates, bile and perfused liver tissue. Accordingly, several parameters targeting hepatocyte or cholangiocyte function or metabolism have been recently suggested as potential viability tests before organ transplantation. We summarize here a current status of respective machine perfusion tests, and report their clinical relevance.

Prediction of Liver Function by Using Magnetic Resonance-based Portal Venous Perfusion Imaging

To evaluate whether liver function can be assessed globally and spatially by using volumetric dynamic contrast-enhanced magnetic resonance imaging MRI (DCE-MRI) to potentially aid in adaptive treatment planning. Seventeen patients with intrahepatic cancer undergoing focal radiation therapy (RT) were enrolled in institution review board-approved prospective studies to obtain DCE-MRI (to measure regional perfusion) and indocyanine green (ICG) clearance rates (to measure overall liver function) prior to, during, and at 1 and 2 months after treatment. The volumetric distribution of portal venous perfusion in the whole liver was estimated for each scan. We assessed the correlation between mean portal venous perfusion in the nontumor volume of the liver and overall liver function measured by ICG before, during, and after RT. The dose response for regional portal venous perfusion to RT was determined using a linear mixed effects model. There was a significant correlation between the ICG clearance rate and mean portal venous perfusion in the functioning liver parenchyma, suggesting that portal venous perfusion could be used as a surrogate for function. Reduction in regional venous perfusion 1 month after RT was predicted by the locally accumulated biologically corrected dose at the end of RT (P<.0007). Regional portal venous perfusion measured during RT was a significant predictor for regional venous perfusion assessed 1 month after RT (P<.00001). Global hypovenous perfusion pre-RT was observed in 4 patients (3 patients with hepatocellular carcinoma and cirrhosis), 3 of whom had recovered from hypoperfusion, except in the highest dose regions, post-RT. In addition, 3 patients who had normal perfusion pre-RT had marked hypervenous perfusion or reperfusion in low-dose regions post-RT. This study suggests that MR-based volumetric hepatic perfusion imaging may be a biomarker for spatial distribution of liver function, which could aid in individualizing therapy, particularly for patients at risk for liver injury after RT.