With the addition of new staff dedicated to the development of Artificial Intelligence (AI) tools for MBI and the support of a strong IT core with high-performance resources, the imaging core facility has putted in place an ‘AI initiative’ aiming to link researchers interested to develop AI algorithm and make them available for the researcher community.
Together with our IT team, we currently have created 4 environment (R, Keras, PyTorch or Matlab algorithm) and put available detailed easy-to-use Jupyter notebooks for several popular network architectures dedicated to tasks like denoising, image restoration, image segmentation, or object recognition.
U-Nets Family
U-Net is a convolutional neural network that was developed for biomedical image segmentation at the Computer Science Department of the University of Freiburg, Germany. The network consists of a contracting path and an expansive path, which gives it the u-shaped architecture. The contracting path is a typical convolutional network that consists of repeated application of convolutions, each followed by a rectified linear unit (ReLU) and a max pooling operation. During the contraction, the spatial information is reduced while feature information is increased. The expansive pathway combines the feature and spatial information through a sequence of up-convolutions and concatenations with high-resolution features from the contracting path.
(Source : https://arxiv.org/abs/1505.04597, image courtesy of Vidhya Acharya)
Noise2Void
The field of image denoising is currently dominated by discriminative deep learning methods that are trained on pairs of noisy input and clean target images. Here, we introduce Noise2Void (N2V), a training scheme that takes this idea one step further. It does not require noisy image pairs, nor clean target images. Consequently, N2V allows us to train directly on the body of data to be denoised and can therefore be applied when other methods cannot. Especially interesting is the application to biomedical image data, where the acquisition of training targets, clean or noisy, is frequently not possible.Intuitively, N2V cannot be expected to outperform methods that have more information available during training. Still, we observe that the denoising performance of Noise2Void drops in moderation and compares favorably to training-free denoising methods.
(Source : https://ieeexplore.ieee.org/document/8954066, image courtesy of Mario Serrano)
YOLO (v2, v5)
You only look once (YOLO) is a state-of-the-art, real-time object detection system. It improves upon YOLOv1 in several ways, including the use of Darknet-19 as a backbone, batch normalization, use of a high-resolution classifier, and the use of anchor boxes to predict bounding boxes, and more. YOLO v2 and YOLO 9000 was proposed by J. Redmon and A. Farhadi in 2016 in the paper titled YOLO 9000: Better, Faster, Stronger. In terms of speed, YOLO is one of the best models in object recognition, able to recognize objects and process frames at the rate up to 150 FPS for small networks.
(Source : https://arxiv.org/abs/1612.08242, https://www.biorxiv.org/content/10.1101/2020.03.20.000133v4, film courtesy of Anne Beghin)
CARE
CARE network developed by CSBDeep showed how deep learning enables biological observations beyond the physical limitations of microscopes. On seven concrete examples, they illustrated how microscopy images can be restored even if 60-fold fewer photons are used during acquisition, how isotropic resolution can be achieved, and how diffraction-limited structures can be resolved.
(Source : https://www.nature.com/articles/s41592-018-0216-7)