Reconstructing Interpretable Features in Computational Super-Resolution microscopy via Regularized Latent Search

TL;DR

A robust super-resolution (SR) method based on regularized latent search (RLS) that offers an actionable balance between fidelity to the ground truth (GT) and realism of the recovered image given a distribution prior is proposed.

Abstract

Supervised deep learning approaches can artificially increase the resolution of microscopy images by learning a mapping between two image resolutions or modalities. However, such methods often require a large set of hard-to-get low-res/high-res image pairs and produce synthetic images with a moderate increase in resolution. Conversely, recent methods based on GAN latent search offered a drastic increase in resolution without the need of paired images. However, they offer limited reconstruction of the high-resolution image interpretable features. Here, we propose a robust super-resolution method based on regularized latent search~(RLS) that offers an actionable balance between fidelity to the ground-truth and realism of the recovered image given a distribution prior. The latter allows to split the analysis of a low-resolution image into a computational super-resolution task performed by deep learning followed by a quantification task performed by a handcrafted algorithm and based on interpretable biological features. This two-step process holds potential for various applications such as diagnostics on mobile devices, where the main aim is not to recover the high-resolution details of a specific sample but rather to obtain high-resolution images that preserve explainable and quantifiable differences between conditions.

Figures (8)

• Figure 1: Distribution analysis of squared $L_2$ norms demonstrating the gaussianization of latent style vectors. The graph compares the density of squared norms from the original $\mathcal{Z}$ space (blue), untransformed $\mathcal{W}$ space (orange), and the distributions resulting from the PULSE (green) and our method (red).
• Figure 2: Qualitative comparison of super-resolution reconstructions on the BBBC021 dataset: visualizing the performance of RLS against baseline methods in reconstructing cellular structures and phenotypes under negative control (DMSO) and various treatment conditions at a 16x upscaling factor.
• Figure 3: Visual examples of super-resolving Translocation assay low-resolution images at a 32x upscaling factor under negative control (DMSO) and TNF$\alpha$ treatment conditions: left: low-resolution image, middle: super-resolution reconstruction, right: ground truth.
• Figure 4: Super-resolution of images from a Golgi assay at a 16x upscaling factor under negative control (DMSO) and Nocodazole treatment conditions. The left column shows the low-resolution images, the middle column shows the super-resolution reconstructions and the right column shows the ground truth images.
• Figure 5: Making interpretable measurements from low-resolution images: first increasing the resolution by a super-resolution method, then measuring a handcrafted interpretable feature. Each pair in the boxplots displays the distribution of handcrafted interpretable measurements, where the solid box represents the negative control (DMSO) and the dotted box signifies the positive controls (TNF$\alpha$ for translocation and Nocodazole for Golgi), across various super-resolution methods including RLS, BRGM, PULSE and "w/o regu" alongside with the high-resolution (HR) images for benchmarking. (a) Translocation ratio measurement: The y-axis quantifies the translocation ratio, an interpretable metric indicating TNF-induced NF$\kappa$B translocation (green). The translocation ratio can be differentiated between two conditions from real high-resolution images (HR), but also from super-resolution images (SR). (b) Mean spot area measurement: The y-axis quantifies the mean spot area, an interpretable metric indicating nocodazole-induced Golgi spreading (green), distinguishable between two conditions from real high-resolution images (HR), but also from super-resolution images (SR) reconstructed by our method.