Single Exposure Quantitative Phase Imaging with a Conventional Microscope Using Diffusion Models

Authors: Gabriel della Maggiora, Luis Alberto Croquevielle, Harry Horsley, Thomas Heinis, Artur Yakimovich

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To validate our approach, we employ a widespread brightfield microscope equipped with a commercially available color camera. We apply our model to clinical microscopy of patients urine, obtaining accurate phase measurements. To evaluate our model, we conduct three distinct experiments. First, we generate a set of synthetic samples analogous to those in the training set. Second, we acquire four through-focus stacks (which allow us to compute a ground truth) and subsequently test our method on these images. Finally, we use an uncurated clinical dataset comprised of real-world clinical images. Table 1 shows the results, and Figure 2 shows a qualitative comparison of the methods.
Researcher Affiliation Academia 1 Center for Advanced Systems Understanding (CASUS), G orlitz, Germany 2 Helmholtz-Zentrum Dresden-Rossendorf e. V. (HZDR), Dresden, Germany 3 Department of Computing, Imperial College London, London, United Kingdom 4 Bladder Infection and Immunity Group (BIIG), UCL Centre for Kidney and Bladder Health, Division of Medicine, University College London, Royal Free Hospital Campus, London, United Kingdom 5 Institute of Computer Science, University of Wrocław, Wrocław, Poland 6 School of Computation, Information and Technology, Technical University of Munich, Germany
Pseudocode No No specific pseudocode or algorithm blocks are present in the paper. The methodology is described using mathematical equations and textual explanations in Section 3.
Open Source Code No The paper does not provide concrete access to source code or explicitly state that the code will be made open-source. No repository links are included.
Open Datasets Yes To simulate polychromatic acquisitions, we use the Image Net dataset (Deng et al. 2009). To validate the model s performance, we use the HCOCO dataset (Cong et al. 2019) to simulate synthetic examples. We conduct experiments using the Urinary Tract Infection (UTI) dataset as outlined in (Liou et al. 2024).
Dataset Splits No The paper uses the ImageNet, HCOCO, and Urinary Tract Infection (UTI) datasets but does not provide specific training/validation/test dataset splits, percentages, or methodology for partitioning these datasets.
Hardware Specification No The paper describes the optical hardware used for image acquisition (e.g., 'Olympus BX41F microscope', 'Infinity 3S-1UR, Teledyne Lumenera camera'). However, it does not provide any specific hardware details such as GPU or CPU models used for training or running the diffusion models.
Software Dependencies No The paper does not provide specific software dependencies with version numbers for reproducing the methodology (e.g., programming languages, libraries, or frameworks with their versions).
Experiment Setup No The paper describes the components of the loss function used in training the Zero-Mean Diffusion model, including `LCVDM` and `Lmean`, and mentions a weighting parameter `ω`. However, it does not provide concrete values for hyperparameters such as learning rate, batch size, number of epochs, or specific optimizer settings.