Imagine you’re trying to build a puzzle, but the pieces are so tiny and intricate that it’s nearly impossible to see where they fit. That’s the challenge scientists face in connectomics, where they try to understand how neurons connect in our brains at the tiniest scale. To tackle this puzzle, computer vision technology has been a game-changer, especially deep learning methods used in image processing. But here’s the twist – the current best methods are still falling short of what scientists need. So, inspired by the success of ImageNet, scientists have created an incredible dataset called U-RISC. It’s like getting a magnifying glass to help us solve that puzzle! This dataset has ultra-high-resolution images of cell membranes taken with an electron microscope, with a mind-boggling resolution of 2.18 nanometers per pixel. And not just that – the dataset comes with multiple annotations to ensure its quality. When scientists put their deep learning algorithms to the test on U-RISC, they found that there’s still a big gap between the algorithms and human performance. In fact, the algorithms struggle to predict whether a pixel belongs to a cell membrane or not – it’s like needing a wider view to make a decision. But don’t worry, they’re not giving up. They’ve integrated existing methods and even improved upon them to create a new benchmark for cell membrane segmentation on U-RISC. And if you’re curious, you can explore the dataset and codes they used to develop these algorithms! Trust me, it’s mind-blowing stuff!

Connectomics is a developing field aiming at reconstructing the connection of the neural system at the nanometer scale. Computer vision technology, especially deep learning methods used in image processing, has promoted connectomic data analysis to a new era. However, the performance of the state-of-the-art (SOTA) methods still falls behind the demand of scientific research. Inspired by the success of ImageNet, we present an annotated ultra-high resolution image segmentation dataset for cell membrane (U-RISC), which is the largest cell membrane-annotated electron microscopy (EM) dataset with a resolution of 2.18 nm/pixel. Multiple iterative annotations ensured the quality of the dataset. Through an open competition, we reveal that the performance of current deep learning methods still has a considerable gap from the human level, different from ISBI 2012, on which the performance of deep learning is closer to the human level. To explore the causes of this discrepancy, we analyze the neural networks with a visualization method, which is an attribution analysis. We find that the U-RISC requires a larger area around a pixel to predict whether the pixel belongs to the cell membrane or not. Finally, we integrate the currently available methods to provide a new benchmark (0.67, 10% higher than the leader of the competition, 0.61) for cell membrane segmentation on the U-RISC and propose some suggestions in developing deep learning algorithms. The U-RISC dataset and the deep learning codes used in this study are publicly available.

Read Full Article (External Site)