Automated Assessment of Genotoxicity Using Mutant Drosophila Hair Images

Abstract

In this work, the SMART assay, a genotoxicity test using mutant Drosophila hairs, was automated. The SMART assay assesses the genotoxicity of a chemical compound by counting mutant hairs on the Drosophila's wings. Even though the Drosophila has many advantages in cost and ethnic problems, the speed and accuracy are limited, since the counting is manual. So far, no research has been given to the automation of SMART assay. For the first time, an automated image analysis system that counts the mutant hairs automatically was developed in this work. The automation consists of four parts: image acquisition, image preprocessing, hair detection, and mutant classification. In each part, new automation methods are proposed. In the image acquisition, a wing detection method using ellipse detection and an optimizing method for image acquisition using the wing detection are proposed. In the image preprocessing, the hair image separation into upper and lower using a wind surface reconstruction is proposed, and the hair area segmentation is proposed. In hair detection, a line fitting method in 3D is proposed. In mutant classification, a upper and lower classification and mutant classification using the wing surface are proposed. The proposed system is validated using the proposed automatic matching system. The genotoxicity of the automated SMART assay coincides with that of the original manual SMART assay.