Pilot Study on the Performance of a New System for Image Based Analysis of Peripheral Blood Smears on Normal Samples.

Abstract

Image analysis based automated systems aiming to automate the manual microscopic review of peripheral blood smears have gained popularity in recent times. In this paper, we evaluate a new blood smear analysis system based on artificial intelligence, Shonit™ by SigTuple Technologies Private Limited. One hundred normal samples with no flags from an automated haematology analyser were taken. Peripheral blood smear slides were prepared using the autostainer integrated with an automated haematology analyser and stained using May-Grunwald-Giemsa stain. These slides were analysed with Shonit™. The metrics for evaluation included (1) accuracy of white blood cell classification for the five normal white blood cell types, and (2) comparison of white blood cell differential count with the automated haematology analyser. In addition, we also explored the possibility of estimating the value of red blood cell and platelet indices via image analysis. Overall white blood cell classification specificity was greater than 97.90% and the precision was greater than 93.90% for all the five white blood cell classes. The correlation of the white blood cell differential count between the automated haematology analyser and Shonit™ was found to be within the known inter cell-counter variability. Shonit™ was found to show promise in terms of its ability to analyse peripheral blood smear images to derive quantifiable metrics useful for clinicians. Future enhancement should include the ability to analyse abnormal blood samples.