Training to Improve Precision and Accuracy in the Measurement of Fiber Morphology

Abstract

An estimated $7.1 billion dollars a year is spent due to irreproducibility in pre-clinical data from errors in data analysis and reporting. Therefore, developing tools to improve measurement comparability is paramount. Recently, an open source tool, DiameterJ, has been deployed for the automated analysis of scanning electron micrographs of fibrous scaffolds designed for tissue engineering applications. DiameterJ performs hundreds to thousands of scaffold fiber diameter measurements from a single micrograph within a few seconds, along with a variety of other scaffold morphological features, which enables a more rigorous and thorough assessment of scaffold properties. Herein, an online, publicly available training module is introduced for educating DiameterJ users on how to effectively analyze scanning electron micrographs of fibers and the large volume of data that a DiameterJ analysis yields. The end goal of this training was to improve user data analysis and reporting to enhance reproducibility of analysis of nanofiber scaffolds. User performance was assessed before and after training to evaluate the effectiveness of the training modules. Users were asked to use DiameterJ to analyze reference micrographs of fibers that had known diameters. The results showed that training improved the accuracy and precision of measurements of fiber diameter in scanning electron micrographs. Training also improved the precision of measurements of pore area, porosity, intersection density, and characteristic fiber length between fiber intersections. These results demonstrate that the DiameterJ training module improves precision and accuracy in fiber morphology measurements, which will lead to enhanced data comparability.

Automated cell segmentation in FIJI® using the DRAQ5 nuclear dye

“…1a. To evaluate the segmentation from the leaky DRAQ5 signal, we used an established approach that relies on manually monitoring the automated detection results on a set of test images [10]; similar human-comparison approaches have been used elsewhere [11, 12]. The evaluation was performed by applying the selected frames on the corresponding bright-field image (Fig.…”

Resorbable vascular grafts show rapid cellularization and degradation in the ovine carotid

“…Wall thickness measurements were taken from 150× images of five graft cross‐sections ( n = 5). Fiber orientation and diameter were measured from 1,000× en face images using the DiameterJ v1.018 plugin for FIJI running ImageJ 1.51w (Hotaling et al, 2016; Hotaling, Bharti, Kriel, & Simon, 2015). Three 1,000× images were analyzed from each of five graft segments for a total of 15 images.…”

“…(b) The spinneret and focusing electrodes viewed from the top right. The tip of the needle and the vertical portions of the focusing electrodes are coplanar running ImageJ 1.51w (Hotaling et al, 2016;Hotaling, Bharti, Kriel, & Simon, 2015). Three 1,000× images were analyzed from each of five graft segments for a total of 15 images.…”

Mining Artifacts in Mycelium SEM Micrographs

“…The images analyzed were prepared with magnification 3000X. The systematic error observed for nontrained DiameterJ operators has been considerably reduced after completion of the DiameterJ online training [8], showing that the human factor can greatly affect the quality of the characterization. This fact suggests an opportunity for machine learning improvement of the segmentation choice process.…”