Real-Time Control and Automation Framework for Acousto-Holographic Microscopy

Hasan Berkay Abdioğlu 1 , Yağmur Işık 1 , Mustafa İsmail İnal 1 , Nehir Serin 2 Kerem Bayer 1 , Muhammed Furkan Koşar 1 , Taha Ünal 3 , Hüseyin Uvet 1 · 2025

Read Real-Time Control and Automation Framework for Acousto-Holographic Microscopy by Hasan Berkay Abdioğlu 1 , Yağmur Işık 1 , Mustafa İsmail İnal 1 , Nehir Serin 2 Kerem Bayer 1 , Muhammed Furkan Koşar 1 , Taha Ünal 3 , Hüseyin Uvet 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-03539.txt.

Introduction

Holographic microscopes can image cells without labels, but focusing is hard. Hasan Berkay Abdioğlu and co-authors build an automation framework in this 2025 work. Manual operation for repetitive cell tasks wastes expert time and invites human error.

Digital holographic images are noisy and low in contrast. That makes autofocus and object detection tough for standard tools. The authors add real-time control and automation for acousto-holographic microscopy. Automated microscopes could speed cell biology labs and make results more consistent across users. Acousto-holographic control adds sound-based focusing to the imaging, widening what automation must handle in live cell experiments and assays.

Catalog record

CREATORHasan Berkay Abdioğlu 1 , Yağmur Işık 1 , Mustafa İsmail İnal 1 , Nehir Serin 2 Kerem Bayer 1 , Muhammed Furkan Koşar 1 , Taha Ünal 3 , Hüseyin Uvet 1
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-03539

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Real-Time Control and Automation Framework for Acousto-Holographic Microscopy Hasan Berkay Abdioğlu 1 , Yağmur Işık 1 , Mustafa İsmail İnal 1 , Nehir Serin 2 Kerem Bayer 1 , Muhammed Furkan Koşar 1 , Taha Ünal 3 , Hüseyin Uvet 1 Abstract Manual operation of microscopes for repetitive tasks in cell biology is a significant bottleneck, consuming invaluable expert time, and introducing human error. Automation is essential, and while Digital Holographic Microscopy (DHM) offers powerful, label-free quantitative phase imaging (QPI), its inherently noisy and low-contrast holograms make robust autofocus and object detection challenging. We present the design, integration, and validation of a fully automated closed-loop DHM system engineered for high-throughput mechanical characterization of biological cells. The system integrates automated serpentine scanning, real-time YOLO-based object detection, and a high-performance, multi-threaded software architecture using pinned memory and SPSC queues. This design enables the GPU-accelerated reconstruction pipeline to run fully in parallel with the 50 fps data acquisition, adding no sequential overhead. A key contribution is the validation of a robust, multi-stage holographic autofocus strategy; we demonstrate that a selected metric (based on a low-pass filter and standard deviation) provides reliable focusing for noisy holograms where conventional methods (e.g., Tenengrad, Laplacian) fail entirely. Performance analysis of the complete system identifies the 2.23-second autofocus operation-not reconstruction-as the primary throughput bottleneck, resulting in a 9.62-second analysis time per object. This work delivers a complete functional platform for autonomous DHM screening and provides a clear, data-driven path for future optimization, proposing a hybrid brightfield imaging modality to address current bottlenecks. I Introduction Scientific progress, particularly in cell biology and materials science, is often built upon complex, repetitive experimental analysis. A significant portion of these tasks, such as sample screening, imaging, and focusing, is still performed manually by highly-skilled scientists and technicians. This manual operation is not only a source of significant human error and inter-operator bias, but it also consumes thousands of invaluable man-hours. This bottleneck ties high-knowledge personnel to routine labor, slowing the pace of discovery and representing an inefficient use of critical research funding. The automation of these experiments is therefore an essential step, as it promises to eliminate human bias, ensure experimental consistency, and, most importantly, free expert researchers to focus on data analysis and hypothesis-driven science [ 3 ] . Digital Holographic Microscopy (DHM) is a prime candidate for such automation. As a powerful quantitative phase imaging (QPI) technique [ 6 ] , it provides rich, label-free, quantitative data, enabling the measurement of cell morphology and mechanical properties from a single acquisition. However, the operation of DHM systems and the analysis of the resulting off-axis holograms are often complex, requiring manual focusing and intervention. While automated microscopy systems are common, many rely on clean, high-contrast brightfield or fluorescence imaging. The off-axis holograms used in QPI, however, are inherently noisy and present low-contrast features. This makes the robust, unmanned application of standard computer vision algorithms for object detection and autofocus extremely challenging. Most existing systems fail to provide a complete, closed-loop solution that integrates autonomous scanning, intelligent holographic focusing, and a high-performance acquisition and reconstruction pipeline for mechanical analysis. In this work, we present the design and validation of a fully automated system built around a bespoke, custom-built DHM (off-axis interferometer) [ 11 ] , engineered for high-throughput mechanical characterization. Our system integrates (A) a robust multi-threaded software architecture built on pinned memory, SPSC queues, and GPU-accelerated processing; (B) a serpentine scanning stage with a novel adaptive logic for boundary handling and duplicate rejection; (C) a real-time YOLO-based object detector; and (D) a multi-stage holographic autofocus strategy that overcomes the limitations of noisy interferograms. This paper details the complete methodology for this integrated system, from its opto-mechanical components to its parallel software architecture. We then provide a rigorous validation of its key components, presenting a data-driven analysis of the autofocus performance, detection speed, and overall system throughput. We conclude with a critical discussion of the system’s performance, identifying the primary bottlenecks and proposing a clear, data-driven path for future optimization. II Methodology The methodology details the design and integration of an automated Digital Holographic Microscopy (DHM) system engineered for real-time, high-throughput detection, and topographical mapping and mechanical characterization (Young’s Modulus) of biological cells. The system is partitioned into four primary subsystems: II-A the interferometric imaging subsystem for hologram acquisition; II-C the automated scanning and positioning subsystem for sample manipulation; II-D the real-time image processing and control subsystem for object detection and automated focus; and II-E the real-time reconstruction subsystem for quantitative phase retrieval. The system architecture integrates optical, mechanical, and computational components into a closed-loop platform, managed by a central host computer. The operational workflow is designed for automated, multi-stage analysis at each scanning position. First, the host computer directs the automated scanning subsystem to position the sample, which consists of a PDMS chip with an integrated piezoelectric (PZT) transducer. The imaging subsystem acquires a single survey hologram, which is immediately processed by the control subsystem to perform object detection and per-object autofocus. Once a target is identified and focused, the system proceeds to the mechanical analysis phase. While the integrated PZT provides continuous 10 Hz excitation, the imaging subsystem is triggered to acquire a rapid burst of 250 holograms at 50 frames per second (fps), capturing the sample’s dynamic response. This high-frequency holographic sequence is then made available to the real-time reconstruction subsystem, which processes the acquired holograms to compute the quantitative phase (height) map. This phase data is subsequently used as an input to the Hertzian elasticity model [ 7 ] to derive the object’s mechanical stiffness (Young’s Modulus). II-A Interferometric Imaging Subsystem The optical core is a custom-built, off-axis DHM in a Mach-Zehnder interferometer configuration. A coherent, solid-state green laser (532 nm) serves as the illumination source. The beam is split into an object arm, which illuminates the sample, and a reference arm. The two beams are recombined at a slight angle to generate the off-axis hologram, which is recorded directly by a Basler BOA monochrome CMOS sensor in a lensless configuration. Fig. 1: Side-view of the Mach-Zehnder Interferometer Setup II-B System Architecture II-C Automated Scanning and Positioning Subsystem Automated sample traversal is achieved using a high-precision, three-axis (XYZ) stepper motor stage assembly, managed by a PI (Physik Instrumente) Mikromove C-884 controller. The system is configured to scan a user-defined rectangular area, delimited by x_{min} , x_{max} , y_{min} , and y_{max} coordinates, allowing for dynamic region selection. To maximize throughput, the stage executes a serpentine scanning pattern. The X-axis first traverses from x_{min} to x_{max} . Upon completion of this sweep, the Y-axis increments by a single step. The X-axis then traverses in the reverse direction ( x_{max} to x_{min} ). This path is repeated until the Y-axis reaches y_{max} , minimizing the mechanical overhead associated with stage acceleration and deceleration. The incremental step size in X and Y is calculated based on the optical Field of View (FOV), which is determined by the CMOS sensor’s active area and the system’s optical magnification. The step size is configured to be slightly less than the FOV to ensure sufficient overlap between adjacent fields to minimize object misses between the boundaries. II-D Real-Time Image Processing and Control Subsystem This subsystem is

Read the complete text