Microscopy Learning Research Articles

Machine learning in microscopy - insights, opportunities and challenges.

Machine learning (ML) is transforming the field of image processing and analysis, from automation of laborious tasks to open-ended exploration of visual patterns. This has striking implications for image-driven life science research, particularly microscopy. In this Review, we focus on the opportunities and challenges associated with applying ML-based pipelines for microscopy datasets from a user point of view. We investigate the significance of different data characteristics - quantity, transferability and content - and how this determines which ML model(s) to use, as well as their output(s). Within the context of cell biological questions and applications, we further discuss ML utility range, namely data curation, exploration, prediction and explanation, and what they entail and translate to in the context of microscopy. Finally, we explore the challenges, common artefacts and risks associated with ML in microscopy. Building on insights from other fields, we propose how these pitfalls might be mitigated for in microscopy.

Combining unsupervised and supervised learning in microscopy enables defect analysis of a full 4H-SiC wafer

Detecting and analyzing various defect types in semiconductor materials is an important prerequisite for understanding the underlying mechanisms and tailoring the production processes. Analysis of microscopy images that reveal defects typically requires image analysis tasks such as segmentation and object detection. With the permanently increasing amount of data from experiments, handling these tasks manually becomes more and more impossible. In this work, we combine various image analysis and data mining techniques to create a robust and accurate, automated image analysis pipeline for extracting the type and position of all defects in a microscopy image of a KOH-etched 4H-SiC wafer.Graphical abstract

DeepImageJ: A user-friendly environment to run deep learning models in ImageJ.

DeepImageJ is a user-friendly solution that enables the generic use of pre-trained deep learning models for biomedical image analysis in ImageJ. The deepImageJ environment gives access to the largest bioimage repository of pre-trained deep learning models (BioImage Model Zoo). Hence, nonexperts can easily perform common image processing tasks in life-science research with deep learning-based tools including pixel and object classification, instance segmentation, denoising or virtual staining. DeepImageJ is compatible with existing state of the art solutions and it is equipped with utility tools for developers to include new models. Very recently, several training frameworks have adopted the deepImageJ format to deploy their work in one of the most used softwares in the field (ImageJ). Beyond its direct use, we expect deepImageJ to contribute to the broader dissemination and reuse of deep learning models in life sciences applications and bioimage informatics.

The promise and peril of deep learning in microscopy.

The promise and peril of deep learning in microscopy.

Free annotated data for deep learning in microscopy? A hitchhiker’s guide

In microscopy, the time burden and cost of acquiring and annotating large datasets that many deep learning models take as a prerequisite, often appears to make these methods impractical. Can this requirement for annotated data be relaxed? Is it possible to borrow the knowledge gathered from datasets in other application fields and leverage it for microscopy? Here, we aim to provide an overview of methods that have recently emerged to successfully train learning-based methods in bio-microscopy.

Volume 95A, Number 4, April 2019 Cover Image

AbstractThe special section “Image Cytometry” presents 4 articles on the use of automation, and machine learning in microscopy applied to the acquisition, processing and analysis of digital images. See editorial by Bengtsson and Tárnok in this issue. Cover design by Bärbel Beran [www.beran-design.de]

A Virtual Microscopy Learning Platform - a High Quality, Innovative and Interactive Tool for Training Haematologists of the Future: A UK Pilot Study

Abstract Introduction: Virtual microscopy (VM) allows a whole slide once scanned, to be visualised, navigated and annotated at different magnifications on a digital viewing platform. Advantages for trainees are numerous; slides can be viewed simultaneously by large numbers of students or accessed remotely at the student’s time and place of preference. In the UK all Haematology trainees are required to be competent in diagnostic evaluation of blood films, bone marrow aspirates and trephines. The drive towards laboratory centralisation, increasing automation and full shift working patterns has decreased access to traditional small group teaching using a multi-headed microscope. VM has the potential to realise new modes of training delivery and to prepare haematologists for the advent of VM as a mainstream diagnostic tool. Aim: To review the utility and acceptability of incorporating virtual microscopy into a virtual learning environment (VLE) for haematology trainees. Methods : Over a 6 month period a VLE has been developed centred on the use of VM comprising of: basic morphology tutorials including interactive learning modules and instructional videosannotated image bankclinical case scenario simulationmock examination practice Curriculum mapped content was created to develop acquisition of morphological skills from first principles. 4 interactive modules take users through a step-wise approach to examining a blood film, recognising normal appearances, identifying morphological abnormalities and interpreting their findings in the film report. Case scenarios, requiring trainees to identify morphological abnormalities and put these into clinical context were developed using structured questioning, providing instant feedback on performance at each stage of the case. Additional resources include an annotated image bank of >250 slides and mock examination cases. The use of VM to support large group teaching was piloted at an afternoon for haematology trainees in London. Haematopathology slides were made available on-line prior to the session with videos introducing an approach to examining trephines and lymph nodes. Questions to facilitate examination of the slides were built in to promote learning. The interactive session was facilitated by trainees using the VLE site and case discussions made available on-line for future reference. Informal feedback on the site was encouraged alongside its development and formal surveys were undertaken to evaluate the quality and utility of the VLE both for self-directed and large group teaching. Results: The VLE was piloted to 40 trainee haematology physicians and 10 biomedical scientists only 1 of whom had previous experience of VM. 100% of participants viewed the experience as ‘useful’ or ‘very useful’ and would recommend to other colleagues. Junior trainees particularly valued the structured tutorials on blood film reporting and annotated practice cases. The novel application of the VLE for simulating haematological emergencies presenting with laboratory abnormalities (e.g. APML) was found to be invaluable by new trainees in preparation for out of hours work. 88 trainees attended the large group teaching session. Slides were viewed on-line by 53/88 (60%) trainees in preparation. 78% of trainees rated the training experience as 5/5 (very good). Free text comments included: “pre-course material very useful; really liked the digital microscopy; the most useful day this year”. The use of VLE in this context has been incorporated into future training events. When comparing to light microscopy the key benefits of VM identified by users were: AccessibilityAbility to return to review material as desiredAnnotation of digital imagesAbility to compare different slides side by side within the same screen windowAvailability of slides of rare diagnoses. Overall, image quality was rated as equivalent to viewing glass slides. Technicalities of viewing the slides (panning, changing magnification and adjusting focus) were easier using a light microscope compared to the digital microscope interface. Conclusions: Incorporation of VM into a VLE is an innovative and effective platform for transforming haematology training for the 21st century. Its versatility, quality and accessibility facilitate its implementation into a wide range of learning settings. Minor technical limitations can be easily addressed to improve the user experience. Disclosures No relevant conflicts of interest to declare.