Efficiency and Quality Improvement In Pathology Diagnostic by Using Computational Pathology: Software-Based Analysis of Perineural Invasion in Colon Carcinoma
Abstract
Dora Demirdag, Mariam Khacheishvili and Alexi Baidoshvili
Introduction: Many pathology laboratories are transitioning from diagnostics with glass slides to diagnostics with whole-slide images (WSI). One of the advantages of digital pathology is the possibility of using the software. Annotations made by software could increase the efficiency and the quality of pathologists ́ work. In this research, annotations for nerves were used in WSI ́s of colon carcinoma cases to help the detection of perineural invasion to improve the efficiency (time needed to find perineural invasion) and quality (the possibility of finding more perineural invasion in colon carcinoma cases and standardization) of the diagnostics.
Materials & Methods: 4 pathologists got 148 colon carcinoma cases without annotations made by software to screen the slides for perineural invasion. After a minimum of 4 weeks of washing out, they got the same instances with annotations for nerves made by software to find perineural invasion. We compared the time they needed for the cases with and without annotations and the percentage of perineural invasion found with and without annotations.
Results: All 4 pathologists took advantage of the annotations made by software regarding the time they needed to conclude the perineural invasion, and the difference between the average time of all 4 pathologists without and with the use of software was statistically significant. All 4 pathologists found individual more perineural invasion. However, the average % of found perineural invasion without and with the help of software needed to be more substantial.