Optical Image Processing In The Drosophila Compound Eye Lesions

To speed up the development of biomedicine has been the goal of the scientific researchers for many years. Drosophila, as a kind of excellent material of experimental animal models, is used extensively in the research of biomedicine and pharmacy. In medical studies, they are usually treated with reagents of interest and the changes in their organs, particularly their compound eyes, are then collected and analyzed. At present, these series of examine processes are still done manually, which makes the research very laborious and inefficient and sometimes affects the accuracy of the results. This paper studies on the optical image processing in the system of drosophila's compound eyes lesion diagnosis. It is a novel attempt to make use of the technologies of optical image capturing, digital image processing and embedded computer to study the kind of medical micro-image domain such as drosophila and mini-type samples. This work will be worth researching and developing merchandise.It has made some effective improvements in the following two main aspects: the algorithm of drosophila's compound eyes lesion and its application. In that, the design of the software part of the intelligent diagnosis system of drosophila's compound eyes lesion is realized based on embedded Linux system. Here, a novel algorithm is presented for extracting the region of drosophila's compound eyes based on HSI color space. In the algorithm, the color images of drosophila's compound eyes are resolved into three monochrome components (R, G and B), followed by replacing B with R. After mapping from RGB to HSI, the target regions are accurately extracted using OTSU. Two preparing algorithms for the target detection of the drosophila's compound eyes are proposed, which are Fast Fuzzy C-Means (FFCM) clustering algorithm for monochrome image segmentation and maximum class variance (OTSU) thresholding algorithm. After comparing their performance in practice, the results indicate that OTSU thresholding algorithm is better than the other for extracting drosophila's compound eyes on the saturation space. Since of the massive compute of the OTSU thresholding algorithm, improved Genetic Algorithm (IGA), which is robustness, parallelizable and novel, is used to search the best threshold for segmentation. The paper expatiates on IGA principle, its experiment realization and result. In this paper, it uses the characters of area and intensity to diagnose the two different pathological forms, defined as A and B. The environment of software development is established on embedded Linux system, and then the functions are realized, which are the drosophila's compound eyes micro-image collection based on optical image capturing, the multi-format image compression and storage, the compound eyes pathological diagnosis, the diagnosis pictorial information management and the large batch of compound eyes images of intelligent diagnostic, are contained. Finally, based on the QT software as the platform, GUI for automatic detection of pathological changes of drosophila's compound eye is designed.The text results show that the detection precision of this system has been achieved over 95%, the response time below 0.1s, the processing time below 3s/sheet and a friendly interacting interface. Compared with manual inspection, the intelligent diagnosis system supplies a new effective and better method to test the drug reaction of the subject (such as drosophila) participated the experiment.