Abstract:Computer aid auto-focus system is necessary for accurate microscope diagnosis, especially for the high precision microscope, which leaves little physical distance for focus adjusting manually. We proposed an image-driven microscope fast auto-focus system with a deep neural network. There are two main contributions. First, combining the high-level feature learning ability advantages of convolution neural network (CNN) and the handcraft feature selection ability of statistical learning, we proposed a High-level-Handcraft Neural Network (HH-Net) to accurately determine the distance index between microscope lens and cell smear by evaluating the image focus quality. It deployed 13 layers CNN for the high-level feature extraction from image patches. While the handcraft features which provide global information from the raw image were extracted by statistical algorithms and merged into CNN features. Finally, the combined features are utilized by the fully connected layers in the network to obtain the final distance index by classifying the biomedical image focus quality. Second, cooperated with the HH-Net, we propose an end to end image driven microscope fast auto-focus system, which can learn auto-focus policies from visual input and finish at a clear spot automatically. The accuracy of our patch level focus quality prediction is 92.4% with HH-Net, while the real-time image level focus quality predication can be 99.99% with 0.025s cost time by certainty voting strategy. Our auto-focus system can also cooperate with the X-Y Micro platform to automatically scan the whole cell smear and get the real-time best-in-focus image of a microscope with fast response, accuracy, and robustness.

HH-Net: Image Driven Microscope Fast Auto-Focus with Deep Neural Network

Single-shot Autofocusing in Light Sheet Fluorescence Microscopy with Multiplexed Structured Illumination and Deep Learning

Machine learning guided rapid focusing with sensor-less aberration corrections.

Design and Optimization for Microscope Autofocus Systems Based on the Analysis of Video Signals

A Sensorless Adaptive Optics Control System for Microscopy Based on Extreme Learning Machine

Deep Learning-Based Single-Shot Autofocus Method for Digital Microscopy

Lensfree Auto-Focusing Imaging with Coarse-to-fine Tuning Method

A Deep-Learning-based Workflow to Deal with the Defocusing Problem in High-Throughput Experiments

Single-shot autofocusing of microscopy images using deep learning

Learning to High-Performance Autofocus Microscopy with Laser Illumination

Rapid Whole Slide Imaging via Learning-based Two-shot Virtual Autofocusing

Low-cost whole slide imaging system with single-shot autofocusing based on color-multiplexed illumination and deep learning

Deep Learning for Camera Autofocus

DeepFocus: A Few-Shot Microscope Slide Auto-Focus Using a Sample Invariant CNN-Based Sharpness Function

Single-Shot Focus Estimation for Microscopy Imaging with Kernel Distillation

An Auto-Focus Method of Microscope for the Surface Structure of Transparent Materials under Transmission Illumination

Towards smart optical focusing: Deep learning-empowered wavefront shaping in nonstationary scattering media

Robust deep learning optical autofocus system applied to automated multiwell plate single molecule localization microscopy

Deep Learning Microscopy

Fast Autofocusing using Tiny Transformer Networks for Digital Holographic Microscopy

Intelligent Autofocus