Advances in lensless imaging based on machine learning using ray tracing

Sickle cell disease (SCD), an inherited red blood cell disorder characterized by the cell’s sickle shape, affects nearly 100 million people worldwide. Microscopy of red blood cells is crucial for disease detection. However, conventional methods often rely on bulky, expensive devices and require skilled human analysts to interpret Red Blood Cell (RBC) images. To address these challenges, we replaced the lens of the microscope with a phase modulator to capture the opto-biological signature(OBS) of blood samples. Our approach bypasses the need for image reconstruction, enabling direct classification using Machine Learning (ML) algorithms based on statistical features extracted from OBS data. In this project, we focus on optimizing the experimental setup using ray tracing software simulations to enhance the efficiency and accuracy of our classification system, rather than optimizing the ML algorithm itself. In this paper, we have shown how optical design can effectively help enhance the performance of lensless imaging systems. We utilized Zemax OpticStudio for ray tracing simulations to model the lensless microscope and subsequently replicated these simulations using Fourier optics in MATLAB. The advantages of using ray tracing software like Zemax include its specialized environment for optical simulations and the efficiency in handling optical components. Additionally, Zemax provides a well-established framework for ensuring accuracy, whereas MATLAB simulations may vary based on the quality of the code and algorithms used. In conclusion, depending on the specific application of the lensless microscope, parameters such as the distances between elements and the type of modulator can be effectively optimized using a ray tracing software like Zemax.