Tissue analysis is pivotal research to determine the pathological properties that occur after the wound healing process. Several staining techniques to understand the morphology of scar tissue are widely used, such as staining with HE (Hematoxylin and Eosin), picrosirius red, and Masson's Trichome. Tissue staining using hematoxylin and eosin has several limitations: labor-intensive, time-consuming, high memory, and cost. Besides that, used a whole slide image to analyze the scar lesion can be more challenging. Hence, we used deep learning to automatically classify and localize scar lesions in the whole slide image based on object instance segmentation. Deep learning trained the patterns from the data representation through a neural network and convolution equations. Deep learning recognized 384 images in less than a minute with 99.89% accuracy. Therefore, the proposed deep learning method can be time- and cost-effective to characterize the pathological feature of scar tissue for the objective histological analysis. In addition to confirming the scar's recognition in the qualitative analysis, the authors also performed a quantitative analysis to obtain information from the scar tissue, such as collagen density from color extraction and collagen directional variance. Segmentation analysis is also used to determine the morphological structure in scar tissue compared to normal tissue. The analysis results can determine various further therapeutic methods to reduce or even eliminate scars on urological tissues in future works.
Tissue analysis needs to determine the pathological properties that occur after wound healing process. Several staining techniques are widely used to understand the morphology of scar tissue, such as staining with hematoxylin and eosin (HE), picrosirius red, and Masson’s Trichome. In spite of the common staining technique, the tissue staining using HE has several limitations, such as labor-intensive, time-consuming, high memory and cost. Due to the limited of view, using the whole slide image is quite challenging to analyze the scar lesion. Hence, we developed a deep learning technique to simultaneously classify and characterize a scar lesion in the whole slide image, based on object instance segmentation. The deep learning trained the patterns from the data representation through neural network and convolution equations. The proposed technique recognized 384 images in less than a minute with 99.9% accuracy. Based on classification, quantitative analysis was performed to confirm the recognition of the scar based on the important features, such as collagen density and directional variance of collagen in scar area. After created the density map and directional variance map of collagen, the differences were almost 50% in normal and scar tissue. Therefore, the proposed deep learning method can be time- and cost-effective to characterize the pathological feature of scar tissue for the objective histological analysis. The analyses are expected to optimize various therapeutic methods to reduce or even eliminate scars on the skin.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.