Designing a preanalytical reader for determining hemolysis, lipemia and icterus in clinical laboratory samples

The main purpose in clinical laboratories is to determine the true value of a concentration or activity. However, results are often affected by the presence of hemolysis, icterus and lipemia interferans, as evidenced by the color change or turbidity in the appearance of the sample. Hemolysis, icterus and lipemia serum indices are determined by two different methods today, namely spectral analysis and visual evaluation. Not every biochemistry laboratory has high-cost, high-tech auto analyzer systems. For this reason, measurements are often made using a visual evaluation procedure. Hemolysis, icterus and lipemia levels are graded by visually comparing the color change in serum samples with calibration samples. In visual evaluation, the results depend on subjective decisions, the experience of the staff, the fatigue depending on the workload, the laboratory environment and light conditions, etc. It is often affected for reasons. The mechanical parts of the system are designed with SolidWorks and produced with a three-dimensional printer. Esp32 was used as the electronic controller of the reader system and Ov2640 camera was used for image recordings. Reader system embedded software studies have been done using Arduino IDE platform C and C ++ programming languages. For image processing, the color data of the image area that constitutes the serum region was obtained by using the Python programming language and artificial intelligence algorithms. Images of 788 samples with different index levels were recorded under light with a wavelength of 519 - 628 nm for hemolysis, 467 - 519 nm for icterus, and 467 - 519 - 628 nm for lipemia. A negative and strong correlation was found between R, G and B color values and serum index values of hemolysis, icterus and lipemia at a rate of 0.92, 0.90 and 0.86, respectively. The coefficient of determination of the obtained regression equations was calculated as 0.86 for hemolysis, 0.82 for icterus, and 0.88 for lipemia. By recording the images of 197 test samples, the success of the regression equations and the manual algorithm to predict serum index levels were calculated. The accuracy of the regression equations for detecting the presence/absence of hemolysis, icterus, lipemia serum indices (> H+, I+, L+) was found to be 89%, 72%, and 99%, respectively. In addition, success calculations were made on hemolysis, icterus and lipemia samples graded with regression equations. Accuracy rate of 86% for H+ grade hemolysis, 86% for H++ grade icterus, 82% for I+ grade icterus, 89% for I++ grade icterus, 96% for I+++ grade icterus, 98% for I++++ grade icterus, 99% for L+ grade lipemia found. With the developed manual algorithm, the accuracy of detecting the presence/absence of hemolysis, icterus, lipemia serum indices (> H+, I+, L+) was found to be 78%, 81%, and 99%, respectively. As a result of this study, a new serum index calculation method was developed in which the results are more objective, precise, stable and accurate compared to the visual assessment method. As a result of the study, a user-oriented preanalytical reader system has been developed that is cost-effective, compatible with laboratory conditions, compared to commercial auto analyzers.