Automated Analysis of Microscopic Images of Isolated Pancreatic Islets

IsletSwipe, a mobile platform for expert opinion exchange on islet graft images

We previously developed a deep learning-based web service (IsletNet) for an automated counting of isolated pancreatic islets. The neural network training is limited by the absent consensus on the ground truth annotations. Here, we present a platform (IsletSwipe) for an exchange of graphical opinions among experts to facilitate the consensus formation. The platform consists of a web interface and a mobile application. In a small pilot study, we demonstrate the functionalities and the use case scenarios of the platform. Nine experts from three centers validated the drawing tools, tested precision and consistency of the expert contour drawing, and evaluated user experience. Eight experts from two centers proceeded to evaluate additional images to demonstrate the following two use case scenarios. The Validation scenario involves an automated selection of images and islets for the expert scrutiny. It is scalable (more experts, images, and islets may readily be added) and can be applied to independent validation of islet contours from various sources. The Inquiry scenario serves the ground truth generating expert in seeking assistance from peers to achieve consensus on challenging cases during the preparation for IsletNet training. This scenario is limited to a small number of manually selected images and islets. The experts gained an opportunity to influence IsletNet training and to compare other experts' opinions with their own. The ground truth-generating expert obtained feedback for future IsletNet training. IsletSwipe is a suitable tool for the consensus finding. Experts from additional centers are welcome to participate.

A Multiparametric Assessment of Human Islets Predicts Transplant Outcomes in Diabetic Mice

Prior to transplantation into individuals with type 1 diabetes, in vitro assays are used to evaluate the quality, function and survival of isolated human islets. In addition to the assessments of these parameters in islet, they can be evaluated by multiparametric morphological scoring (0–10 points) and grading (A, B, C, D, and F) based on islet characteristics (shape, border, integrity, single cells, and diameter). However, correlation between the multiparametric assessment and transplantation outcome has not been fully elucidated. In this study, 55 human islet isolations were scored using this multiparametric assessment. The results were correlated with outcomes after transplantation into immunodeficient diabetic mice. In addition, the multiparametric assessment was compared with oxygen consumption rate of isolated islets as a potential prediction factor for successful transplantations. All islet batches were assessed and found to score: 9 points (n = 18, Grade A), 8 points (n = 19, Grade B), and 7 points (n = 18, Grade B). Islets that scored 9 (Grade A), scored 8 (Grade B) and scored 7 (Grade B) were transplanted into NOD/SCID mice and reversed diabetes in 81.2%, 59.4%, and 33.3% of animals, respectively (P < 0.0001). Islet scoring and grading correlated well with glycemic control post-transplantation (P < 0.0001) and reversal rate of diabetes (P < 0.05). Notably, islet scoring and grading showed stronger correlation with transplantation outcome compared to oxygen consumption rate. Taken together, a multiparametric assessment of isolated human islets was highly predictive of transplantation outcome in diabetic mice.

Semi-Automated Assessment of Human Islet Viability Predicts Transplantation Outcomes in a Diabetic Mouse Model

In clinical and experimental human pancreatic islet transplantations, establishing pretransplant assessments that accurately predict transplantation outcomes is crucial. Conventional in vitro viability assessment that relies on manual counting of viable islets is a routine pretransplant assessment. However, this method does not correlate with transplantation outcomes; to improve the method, we recently introduced a semi-automated method using imaging software to objectively determine area-based viability. The goal of the present study was to correlate semi-automated viability assessment with posttransplantation outcomes of human islet transplantations in diabetic immunodeficient mice, the gold standard for in vivo functional assessment of isolated human islets. We collected data from 61 human islet isolations and 188 subsequent in vivo mouse transplantations. We assessed islet viability by fluorescein diacetate and propidium iodide staining using both the conventional and semi-automated method. Transplantations of 1,200 islet equivalents under the kidney capsule were performed in streptozotocin-induced diabetic immunodeficient mice. Among the pretransplant variables, including donor factors and post-isolation assessments, viability measured using the semi-automated method demonstrated a strong influence on in vivo islet transplantation outcomes in multivariate analysis. We calculated an optimized cutoff value (96.1%) for viability measured using the semi-automated method and showed a significant difference in diabetes reversal rate for islets with viability above this cutoff (77% reversal) vs. below this cutoff (49% reversal). We performed a detailed analysis to show that both the objective measurement and the improved area-based scoring system, which distinguished between small and large islets, were key features of the semi-automated method that allowed for precise evaluation of viability. Taken together, our results suggest that semi-automated viability assessment offers a promising alternative pretransplant assessment over conventional manual assessment to predict human islet transplantation outcomes.

The Flaws and Future of Islet Volume Measurements

When working with isolated islet preparations, measuring the volume of tissue is not a trivial matter. Islets come in a large range of sizes and are often contaminated with exocrine tissue. Many factors complicate the procedure, and yet knowledge of the islet volume is essential for predicting the success of an islet transplant or comparing experimental groups in the laboratory. In 1990, Ricordi presented the islet equivalency (IEQ), defined as one IEQ equaling a single spherical islet of 150 μm in diameter. The method for estimating IEQ was developed by visualizing islets in a microscope, estimating their diameter in 50 μm categories and calculating a total volume for the preparation. Shortly after its introduction, the IEQ was adopted as the standard method for islet volume measurements. It has helped to advance research in the field by providing a useful tool improving the reproducibility of islet research and eventually the success of clinical islet transplants. However, the accuracy of the IEQ method has been questioned for years and many alternatives have been proposed, but none have been able to replace the widespread use of the IEQ. This article reviews the history of the IEQ, and discusses the benefits and failings of the measurement. A thorough evaluation of alternatives for estimating islet volume is provided along with the steps needed to uniformly move to an improved method of islet volume estimation. The lessons learned from islet researchers may serve as a guide for other fields of regenerative medicine as cell clusters become a more attractive therapeutic option.