Search

The image processing of muscular fibers has seen significant improvements over the last decades, from light microscopy acquisition to virtual microscopy and from manual segmentation and fiber typing to automatic. The present study aims to discuss the main tools employed in different histo-enzymological image processing phases applied to muscle fibers, which are respectively: the acquisition, segmentation and fiber typing as well as their efficiency in the determination of morphometric parameters. First, the acquisition: to count the cell numbers within the muscle fiber sections, optical microscopic images with different magnifications (x100, x200, x250, and x400) were compared with virtual slides digitized by Slide Scanner. Second, the segmentation: Three softwares (Fiji "Digitizing pen, Mouse", Image Pro Plus 10 [semi-automatic], and Cytoinformatics LLC [automatic]) were compared for image segmentation quality and the determining of fiber numbers. Third, manual fiber typing: employing Fiji of segmented images using the 3 previously mentioned softwares were performed to calculate the accuracy of the morphometric parameter (Cross sectional Area [CSA], perimeter and Minimal Feret's Diameter [MFD]). The results of the acquisition showed that the scanner slides have a better resolution and detected a higher number of cells than when optical microscopes with different magnifications were used; with the latter, the processing can be performed only with an acceptable resolution, where the number of cells is lesser, which unfortunately requires several repetitions and exhausting work. Our findings regarding segmentation indicate that the Cytoinformatics LLC showed the best processing time and the highest quality followed by IP and Fiji. Finally and for the ultimate step, the morphometric parameter calculation: showed that the best accuracy was attained by using Fiji followed by Cytoinformatics and finally by IP. The findings of this study suggest that Fiji (semi-automatic) showed the best quality/price ratio (open access software) for segmentation and fiber typing, but it proved to be time consuming when compared to Image Pro Plus 10 (semi-automatic) and Cytoinformatics LLC (automatic). [ABSTRACT FROM AUTHOR]

• Deep learning models often require large volumes of labelled data. • The labelling process is usually very labour intensive. • We demonstrate a method of generating and labelling B-mode ultrasound images automatically. • This approach may have applications in image processing and data sharing. Deep learning approaches are common in image processing, but often rely on supervised learning, which requires a large volume of training images, usually accompanied by hand-crafted labels. As labelled data are often not available, it would be desirable to develop methods that allow such data to be compiled automatically. In this study, we used a Generative Adversarial Network (GAN) to generate realistic B-mode musculoskeletal ultrasound images, and tested the suitability of two automated labelling approaches. We used a model including two GANs each trained to transfer an image from one domain to another. The two inputs were a set of 100 longitudinal images of the gastrocnemius medialis muscle, and a set of 100 synthetic segmented masks that featured two aponeuroses and a random number of 'fascicles'. The model output a set of synthetic ultrasound images and an automated segmentation of each real input image. This automated segmentation process was one of the two approaches we assessed. The second approach involved synthesising ultrasound images and then feeding these images into an ImageJ/Fiji-based automated algorithm, to determine whether it could detect the aponeuroses and muscle fascicles. Histogram distributions were similar between real and synthetic images, but synthetic images displayed less variation between samples and a narrower range. Mean entropy values were statistically similar (real: 6.97, synthetic: 7.03; p = 0.218), but the range was much narrower for synthetic images (6.91 – 7.11 versus 6.30 – 7.62). When comparing GAN-derived and manually labelled segmentations, intersection-over-union values- denoting the degree of overlap between aponeurosis labels- varied between 0.0280 – 0.612 (mean ± SD: 0.312 ± 0.159), and pennation angles were higher for the GAN-derived segmentations (25.1° vs. 19.3°; p < 0.001). For the second segmentation approach, the algorithm generally performed equally well on synthetic and real images, yielding pennation angles within the physiological range (13.8–20°). We used a GAN to generate realistic B-mode ultrasound images, and extracted muscle architectural parameters from these images automatically. This approach could enable generation of large labelled datasets for image segmentation tasks, and may also be useful for data sharing. Automatic generation and labelling of ultrasound images minimises user input and overcomes several limitations associated with manual analysis. [ABSTRACT FROM AUTHOR]