Self-supervised Deep Learning Research Articles

IGWAS: Image-based genome-wide association of self-supervised deep phenotyping of retina fundus images.

Existing imaging genetics studies have been mostly limited in scope by using imaging-derived phenotypes defined by human experts. Here, leveraging new breakthroughs in self-supervised deep representation learning, we propose a new approach, image-based genome-wide association study (iGWAS), for identifying genetic factors associated with phenotypes discovered from medical images using contrastive learning. Using retinal fundus photos, our model extracts a 128-dimensional vector representing features of the retina as phenotypes. After training the model on 40,000 images from the EyePACS dataset, we generated phenotypes from 130,329 images of 65,629 British White participants in the UK Biobank. We conducted GWAS on these phenotypes and identified 14 loci with genome-wide significance (p<5×10-8 and intersection of hits from left and right eyes). We also did GWAS on the retina color, the average color of the center region of the retinal fundus photos. The GWAS of retina colors identified 34 loci, 7 are overlapping with GWAS of raw image phenotype. Our results establish the feasibility of this new framework of genomic study based on self-supervised phenotyping of medical images.

Depth estimation and 3D reconstruction from UAV-borne imagery: Evaluation on the UseGeo dataset

Depth estimation and 3D model reconstruction from aerial imagery is an important task in photogrammetry, remote sensing, and computer vision. To compare the performance of different image-based approaches, this study presents a benchmark for UAV-based aerial imagery using the UseGeo dataset. The contributions include the release of various evaluation routines on GitHub, as well as a comprehensive comparison of baseline approaches, such as methods for offline multi-view 3D reconstruction resulting in point clouds and triangle meshes, online multi-view depth estimation, as well as single-image depth estimation using self-supervised deep learning. With the release of our evaluation routines, we aim to provide a universal protocol for the evaluation of depth estimation and 3D reconstruction methods on the UseGeo dataset. The conducted experiments and analyses show that each method excels in a different category: the depth estimation from COLMAP outperforms that of the other approaches, ACMMP achieves the lowest error and highest completeness for point clouds, while OpenMVS produces triangle meshes with the lowest error. Among the online methods for depth estimation, the approach from the Plane-Sweep Library outperforms the FaSS-MVS approach, while the latter achieves the lowest processing time. And even though the particularly challenging nature of the dataset and the small amount of training data leads to a significantly higher error in the results of the self-supervised single-image depth estimation approach, it outperforms all other approaches in terms of processing time and frame rate. In our evaluation, we have also considered modern learning-based approaches that can be used for image-based 3D reconstruction, such as NeRFs. However, due to the significantly lower quality of the resulting 3D models, we have only included a qualitative comparison of such methods against conventional approaches in the scope of this work.

Bridging imaging and clinical scores in parkinson's progression via multimodal self-supervised deep learning

Bridging imaging and clinical scores in parkinson's progression via multimodal self-supervised deep learning

Self-Supervised Shear Wave Noise Adaptive Subtraction in Ocean Bottom Node Data

Ocean Bottom Node (OBN) acquisition is a technique for marine seismic survey that has gained increased attention in recent years. The removal of shear wave noise from the vertical component of receivers plays a crucial role in the subsequent processing and interpretation of OBN data. Previous solutions suffer from noise residue or signal impairment for complex noise and signal overlap scenarios. In this work, we present and explore a self-supervised deep learning approach to attenuate shear wave noise in OBN data. It applies a deep neural network (DNN) to perform adaptive subtraction and comprises two steps to remove the noise associated with the two horizontal components of receivers, respectively. The two horizontal components are considered as noise reference and are sequentially fed into the DNN, and the DNN predicts the actual leaked noise from the contaminated vertical components data. The self-supervised method achieves improvements in the signal-to-noise ratio (SNR) on a set of synthetic data. The implementation of our method on field data demonstrates that it effectively attenuates the shear wave noise and preserves the valid signal.

DIMOND: DIffusion Model OptimizatioN with Deep Learning.

Diffusion magnetic resonance imaging is an important tool for mapping tissue microstructure and structural connectivity non-invasively in the in vivo human brain. Numerous diffusion signal models are proposed to quantify microstructural properties. Nonetheless, accurate estimation of model parameters is computationally expensive and impeded by image noise. Supervised deep learning-based estimation approaches exhibit efficiency and superior performance but require additional training data and may be not generalizable. A new DIffusion Model OptimizatioN framework using physics-informed and self-supervised Deep learning entitled "DIMOND" is proposed to address this problem. DIMOND employs a neural network to map input image data to model parameters and optimizes the network by minimizing the difference between the input acquired data and synthetic data generated via the diffusion model parametrized by network outputs. DIMOND produces accurate diffusion tensor imaging results and is generalizable across subjects and datasets. Moreover, DIMOND outperforms conventional methods for fitting sophisticated microstructural models including the kurtosis and NODDI model. Importantly, DIMOND reduces NODDI model fitting time from hours to minutes, or seconds by leveraging transfer learning. In summary, the self-supervised manner, high efficacy, and efficiency of DIMOND increase the practical feasibility and adoption of microstructure and connectivity mapping in clinical and neuroscientific applications.

Self-supervised learning of shedding droplet dynamics during steam condensation

Knowledge of condensate shedding droplet dynamics provides important information for the characterization of two-phase heat and mass transfer phenomena. Detecting and segmenting the droplets during shedding requires considerable time and effort if performed manually. Here, we developed a self-supervised deep learning model for segmenting shedding droplets from a variety of dropwise and filmwise condensing surfaces. The model eliminates the need for image annotation by humans in the training step and, therefore, reduces labor significantly. The trained model achieved an average accuracy greater than 0.9 on a new unseen test dataset. After extracting the shedding droplet size and speed, we developed a data-driven model for shedding droplet dynamics based on condensation heat flux and surface properties such as wettability and tube diameter. Our results demonstrate that condensate droplet departure size is both heat flux and tube size dependent and follows different trends based on the condensation mode. The results of this work provide an annotation-free methodology for falling droplet segmentation as well as a statistical understanding of droplet dynamics during condensation.

Enabling unsupervised discovery in astronomical images through self-supervised representations

ABSTRACT Unsupervised learning, a branch of machine learning that can operate on unlabelled data, has proven to be a powerful tool for data exploration and discovery in astronomy. As large surveys and new telescopes drive a rapid increase in data size and richness, these techniques offer the promise of discovering new classes of objects and of efficient sorting of data into similar types. However, unsupervised learning techniques generally require feature extraction to derive simple but informative representations of images. In this paper, we explore the use of self-supervised deep learning as a method of automated representation learning. We apply the algorithm Bootstrap Your Own Latent to Galaxy Zoo DECaLS images to obtain a lower dimensional representation of each galaxy, known as features. We briefly validate these features using a small supervised classification problem. We then move on to apply an automated clustering algorithm, demonstrating that this fully unsupervised approach is able to successfully group together galaxies with similar morphology. The same features prove useful for anomaly detection, where we use the framework astronomaly to search for merger candidates. While the focus of this work is on optical images, we also explore the versatility of this technique by applying the exact same approach to a small radio galaxy data set. This work aims to demonstrate that applying deep representation learning is key to unlocking the potential of unsupervised discovery in future data sets from telescopes such as the Vera C. Rubin Observatory and the Square Kilometre Array.

Clustering Method for Signals in the Wideband RF Spectrum Using Semi-Supervised Deep Contrastive Learning

This paper presents the application of self-supervised deep contrastive learning in clustering signals detected in the wideband RF spectrum, presented in the form of spectrograms. Radio clustering is a method of searching for similar signals within the analyzed part of the radio spectrum. Typically, it is based on one or several specific parameters processed from the signal in a given channel. The authors propose a slightly different, innovative approach; thanks to the self-supervised learning of neural networks, there is no need to define specific parameters, and the feature vector, enabling comparison of Euclidean distances between signals, is generated by a deep neural network trained using a contrastive loss function on a dataset containing different radio modulations. The authors describe self-supervised solutions based on contrastive learning and the methods of signal segmentation and augmentation. The training process utilizes a custom database and the Resnet-50 network with a contrastive cost function. Radio clustering is used for autonomous spectrum analysis across wide frequency ranges and enables, among other things, the detection of tactical radio stations operating with widely dispersed frequency-hopping or a significant reduction in computational power required for real-time analysis of a large number of radio signals.

Abstract 1073: AI-assisted high-dimensional cytological single cell analysis distinguishes normal and malignant urine samples using cell morphology images with high accuracy at the single cell level

Abstract Background: Urothelial cancer (UCC) is the 4th most common malignancy in men. The current clinical problem in UCC management is the high interobserver variability of tissue UCC grading and the low sensitivity of urine cytology for early detection of recurrences and progression. We hypothesize that assessing urine samples and UCC tissues with detailed quantitative image analysis based on artificial intelligence has strong potential in both urine-based early detection of tumor recurrence and progression as well as tissue-based pathological grading. Methods: Here, we utilized the Deepcell® Human Foundation Model (HFM), a self-supervised deep learning model, to assess a cohort of urine samples from 54 normal healthy cases, 19 atypical cases, and 4 UCC cases. High resolution images of label-free single cells were captured and analyzed using the HFM to identify distinct morphological characteristics of cells from different urine origins. The urine samples were loaded onto a single-use microfluidic cartridge as part of the Deepcell® REM-I platform and brightfield images of single cells were captured in real-time. For data analysis, images of all cells run through the REM-I instrument are automatically sent to the Deepcell® Axon Suite, which provides an analysis platform to visualize cell images from specific runs, generate UMAPs of the embedding space, and run additional deep neural network models on samples in silico. Results: High-dimensional analysis of 115 deep learning and morphometric features reveals 12 morphotypes among the pooled urine samples, as represented by morphology UMAP plots. Morphology analysis reveals distinct morphotypes in normal compared to pathological urine samples. Confusion matrix showed deep learning model performance in classifying atypical versus UCC with 89.1% (Atypical (n = 8642), UCC (n = 12637)). This indicates high model performance accuracy for AI prediction of atypical versus UCC samples at the single cell level. Conclusions: The Deepcell® REM-I platform performs well with urine samples and identified numerous cell clusters among normal and pathological urines. AI-assisted high-dimensional morphology analysis distinguishes atypical urine samples from UCC cases with 89.1% accuracy at the single cell level. Future applications of this study include early detection or monitoring of UCC cases using non-invasive patient samples such as urine. Citation Format: Thierry P. van den Bosch, Geert J.L.H. van Leenders, Siamala Sinnadurai, Zhouyang Lian, Ryan Carelli, Matthew Nakaki, Vivian Lu, Manisha Ray, Janifer Cruz, Kelly R. Miller, Mahyar Salek, Maddison Masaeli, Peter J. van der Spek. AI-assisted high-dimensional cytological single cell analysis distinguishes normal and malignant urine samples using cell morphology images with high accuracy at the single cell level [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1073.

Abstract 3523: Self-supervised foundation model captures high-dimensional morphology data from single cell brightfield images

Abstract Our knowledge of cancer biology has advanced with the characterization of distinct cell types and cell states within heterogeneous tumor environments. Many methods for imaging and sorting tumor cells require biomarker labels that alter cell characteristics and create a selection bias. The use of label-free single-cell methods should further improve cancer studies involving viable, unperturbed cells for downstream assays such as RNA-Seq, cell culture expansion, and functional assays. To eliminate biomarkers and enable broader assessment of cells, we developed REM-I, a platform that characterizes and sorts unlabeled single cells based on high-dimensional morphology. Cells are captured with brightfield imaging and processed in real-time by self-supervised deep learning models to generate quantitative AI embeddings representative of cell morphology. A significant technical challenge in building REM-I was developing an AI model based on extracted features from cell images without prior knowledge of cell types, cell preparation, or other application-specific knowledge. Accordingly, we developed the Human Foundation Model (HFM), a hybrid architecture that combines self-supervised learning (SSL) and morphometrics (computer vision) to extract 115 dimensional embeddings representing cell morphology from high-resolution REM-I cell images. SSL produces a foundation model with high generalization capacity that enables hypothesis-free sample exploration and efficient generation of application-specific models. Meanwhile, computer vision extracts features that represent measurable and interpretable concepts (e.g., cell size, shape, texture, intensity). The training process for the HFM self-supervised backbone model utilizes the discriminatory power of supervised tasks. Using synthetic cells and three cancer cell lines, we trained and then validated the reproducibility and generalization capabilities of the resulting model. Results show combining deep learning and morphometrics improve interpretability of data and enable rapid characterization and classification of tumor cells with high accuracy. We also report on the Deepcell® Axon data suite, a tool to analyze data and customize reusable workflows. This includes the ability to store and manage data, visualize high dimensional data as low-dimensional projections, and train classifiers to identify and sort cell populations. To enable compatibility with user-preferred downstream analysis pipelines, Axon provides data export options for images, plots, and embeddings. Our approach allows users of all skill levels to access and interpret AI-enabled morphological profiling. Applications of REM-I include hypothesis-free evaluation of heterogeneous tumor samples, label-free cancer cell enrichment, characterization of distinct cell states, and multi-omic integration. Citation Format: Kiran Saini, Senzeyu Zhang, Ryan Carelli, Kevin B. Jacobs, Amy Wong-Thai, Cris Luengo, Vivian Lu, Anastasia Mavropoulos, Andreja Jovic, Jeanette Mei, Thomas Vollbrecht, Stephane C. Boutet, Mahyar Salek, Maddison Masaeli. Self-supervised foundation model captures high-dimensional morphology data from single cell brightfield images [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3523.

Abstract 6934: Integrating deep learning morphology profiling, cell sorting, and scRNASeq reveals sample diversity and enriched sub-populations

Abstract Advances in single cell technologies have ushered in a new multi-omics era that has provided more granular information on cell types and functions at unprecedented resolution. Morpholomics (the information encoded by cell morphology) has traditionally been used as a readout of cell identity, state, and function; but scalability, interpretation, and sorting based on morphology has remained challenging. The Deepcell® platform, REM-I, performs high-dimensional morphology analysis of unlabeled, single cells using deep-learning on high-resolution bright-field images captured in microfluidic flow to profile and sort target cells in real-time and at scale. Morphology analysis of single cells is achieved by a self-supervised deep learning foundation model, ‘Human Foundation Model’ (HFM), which combines deep learning features with morphometrics. HFM determines cell identities using morphological features associated with distinct processes characterized by granules, vesicles, cell size, pigmentation. and others. To demonstrate the ability of HFM modeling to distinguish and characterize a diversity of cell types representative of tumor microenvironments, we processed various sample types on the REM-I platform. Human melanoma, immune cells, stromal cells, and dissociated biopsies were combined to recapitulate heterogeneous tumor samples. HFM analysis enabled morphology-based exploration of the resulting bright-field images, revealing the presence of various morphologically distinct populations within each sample. Tumor cells were morphologically distinct and clustered separately from non-tumor cells. Immune cells exhibited subtle but separable morphologies, including T cells of distinct activation states. To demonstrate the morphology-based sorting capabilities of REM-I, we applied it to a ‘reference sample’ consisting of several fixed cell lines, at proportions ranging from 1% to &amp;gt;90%. Using only brightfield images of single cells analyzed through HFM, we used the platform to enrich each cell line from the ‘reference sample’ with high yield and purity, and identified cell types constituting a small percentage (~1%) of the total population. We next showed that unlabeled cells enriched on REM-I that are characterized as unperturbed and viable are amenable to downstream molecular analyses. These cell populations were sorted and the retrieved live cells were analyzed by scRNA-Seq (10x Chromium), demonstrating enrichment of various cell types of the tumor microenvironment in each cell population relative to the pre-sorted sample. These results illustrate: 1) the ability to perform downstream molecular analyses on sorted cells from REM-I, 2) the platform can be used to parse the tumor microenvironment, and 3) the link between morphology, cell identity, and molecular signatures. Citation Format: Andreja Jovic, Christian Corona, Kiran Saini, Tiffine Pham, Ryan Carelli, Vivian Lu, Senzeyu Zhang, Stephane C. Boutet, Maddison Masaeli. Integrating deep learning morphology profiling, cell sorting, and scRNASeq reveals sample diversity and enriched sub-populations [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6934.

Self-supervised component separation for the extragalactic submillimetre sky

We use a new approach based on self-supervised deep learning networks originally applied to transparency separation in order to simultaneously extract the components of the extragalactic submillimeter sky, namely the cosmic microwave background (CMB), the cosmic infrared background (CIB), and the Sunyaev-Zel'dovich (SZ) effect. In this proof-of-concept paper, we test our approach on the WebSky extragalactic simulation maps in a range of frequencies from 93 to 545 GHz, and compare with one of the state-of-the-art traditional methods, MILCA, for the case of SZ. We first visually compare the images, and then statistically analyse the full-sky reconstructed high-resolution maps with power spectra. We study the contamination from other components with cross spectra, and particularly emphasise the correlation between the CIB and the SZ effect and compute SZ fluxes around positions of galaxy clusters. The independent networks learn how to reconstruct the different components with less contamination than MILCA. Although this is tested here in an ideal case (without noise, beams, or foregrounds), this method shows significant potential for application in future experiments such as the Simons Observatory (SO) in combination with the Planck satellite.

How to track and segment fish without human annotations: a self-supervised deep learning approach

Tracking fish movements and sizes of fish is crucial to understanding their ecology and behaviour. Knowing where fish migrate, how they interact with their environment, and how their size affects their behaviour can help ecologists develop more effective conservation and management strategies to protect fish populations and their habitats. Deep learning is a promising tool to analyse fish ecology from underwater videos. However, training deep neural networks (DNNs) for fish tracking and segmentation requires high-quality labels, which are expensive to obtain. We propose an alternative unsupervised approach that relies on spatial and temporal variations in video data to generate noisy pseudo-ground-truth labels. We train a multi-task DNN using these pseudo-labels. Our framework consists of three stages: (1) an optical flow model generates the pseudo-labels using spatial and temporal consistency between frames, (2) a self-supervised model refines the pseudo-labels incrementally, and (3) a segmentation network uses the refined labels for training. Consequently, we perform extensive experiments to validate our method on three public underwater video datasets and demonstrate its effectiveness for video annotation and segmentation. We also evaluate its robustness to different imaging conditions and discuss its limitations.

A Novel Fully Convolutional Auto-Encoder Based on Dual Clustering and Latent Feature Adversarial Consistency for Hyperspectral Anomaly Detection

With the development of artificial intelligence, the ability to capture the background characteristics of hyperspectral imagery (HSI) has improved, showing promising performance in hyperspectral anomaly detection (HAD) tasks. However, existing methods proposed in recent years still suffer from certain limitations: (1) Constraints are lacking in the deep feature learning process in terms of the issue of the absence of prior background and anomaly information. (2) Hyperspectral anomaly detectors with traditional self-supervised deep learning methods fail to ensure prioritized reconstruction of the background. (3) The architecture of fully connected deep networks in hyperspectral anomaly detectors leads to low utilization of spatial information and the destruction of the original spatial relationship in hyperspectral imagery and disregards the spectral correlation between adjacent pixels. (4) Hypotheses or assumptions for background and anomaly distributions restrict the performance of many hyperspectral anomaly detectors because the distributions of background land covers are usually complex and not assumable in real-world hyperspectral imagery. In consideration of the above problems, in this paper, we propose a novel fully convolutional auto-encoder based on dual clustering and latent feature adversarial consistency (FCAE-DCAC) for HAD, which is carried out with self-supervised learning-based processing. Firstly, density-based spatial clustering of applications with a noise algorithm and connected component analysis are utilized for successive spectral and spatial clustering to obtain more precise prior background and anomaly information, which facilitates the separation between background and anomaly samples during the training of our method. Subsequently, a novel fully convolutional auto-encoder (FCAE) integrated with a spatial–spectral joint attention (SSJA) mechanism is proposed to enhance the utilization of spatial information and augment feature expression. In addition, a latent feature adversarial consistency network with the ability to learn actual background distribution in hyperspectral imagery is proposed to achieve pure background reconstruction. Finally, a triplet loss is introduced to enhance the separability between background and anomaly, and the reconstruction residual serves as the anomaly detection result. We evaluate the proposed method based on seven groups of real-world hyperspectral datasets, and the experimental results confirm the effectiveness and superior performance of the proposed method versus nine state-of-the-art methods.

Improving quantitative MRI using self-supervised deep learning with model reinforcement: Demonstration for rapid T1 mapping.

This paper proposes a novel self-supervised learning framework that uses model reinforcement, REference-free LAtent map eXtraction with MOdel REinforcement (RELAX-MORE), for accelerated quantitative MRI (qMRI) reconstruction. The proposed method uses an optimization algorithm to unroll an iterative model-based qMRI reconstruction into a deep learning framework, enabling accelerated MR parameter maps that are highly accurate and robust. Unlike conventional deep learning methods which require large amounts of training data, RELAX-MORE is a subject-specific method that can be trained on single-subject data through self-supervised learning, making it accessible and practically applicable to many qMRI studies. Using quantitative mapping as an example, the proposed method was applied to the brain, knee and phantom data. The proposed method generates high-quality MR parameter maps that correct for image artifacts, removes noise, and recovers image features in regions of imperfect image conditions. Compared with other state-of-the-art conventional and deep learning methods, RELAX-MORE significantly improves efficiency, accuracy, robustness, and generalizability for rapid MR parameter mapping. This work demonstrates the feasibility of a new self-supervised learning method for rapid MR parameter mapping, that is readily adaptable to the clinical translation of qMRI.

Self-Supervised Deep Learning—The Next Frontier

Self-Supervised Deep Learning—The Next Frontier

Fast Real-Time Brain Tumor Detection Based on Stimulated Raman Histology and Self-Supervised Deep Learning Model.

In intraoperative brain cancer procedures, real-time diagnosis is essential for ensuring safe and effective care. The prevailing workflow, which relies on histological staining with hematoxylin and eosin (H&E) for tissue processing, is resource-intensive, time-consuming, and requires considerable labor. Recently, an innovative approach combining stimulated Raman histology (SRH) and deep convolutional neural networks (CNN) has emerged, creating a new avenue for real-time cancer diagnosis during surgery. While this approach exhibits potential, there exists an opportunity for refinement in the domain of feature extraction. In this study, we employ coherent Raman scattering imaging method and a self-supervised deep learning model (VQVAE2) to enhance the speed of SRH image acquisition and feature representation, thereby enhancing the capability of automated real-time bedside diagnosis. Specifically, we propose the VQSRS network, which integrates vector quantization with a proxy task based on patch annotation for analysis of brain tumor subtypes. Training on images collected from the SRS microscopy system, our VQSRS demonstrates a significant speed enhancement over traditional techniques (e.g., 20-30min). Comparative studies in dimensionality reduction clustering confirm the diagnostic capacity of VQSRS rivals that of CNN. By learning a hierarchical structure of recognizable histological features, VQSRS classifies major tissue pathological categories in brain tumors. Additionally, an external semantic segmentation method is applied for identifying tumor-infiltrated regions in SRH images. Collectively, these findings indicate that this automated real-time prediction technique holds the potential to streamline intraoperative cancer diagnosis, providing assistance to pathologists in simplifying the process.

Detecting Intentional AIS Shutdown in Open Sea Maritime Surveillance Using Self-Supervised Deep Learning

Detecting Intentional AIS Shutdown in Open Sea Maritime Surveillance Using Self-Supervised Deep Learning

Self-supervised deep learning uncovers the semantic landscape of drug-induced latent mitochondrial phenotypes

Self-supervised deep learning uncovers the semantic landscape of drug-induced latent mitochondrial phenotypes

Self-supervised deep learning of gene-gene interactions for improved gene expression recovery.

Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful tool to gain biological insights at the cellular level. However, due to technical limitations of the existing sequencing technologies, low gene expression values are often omitted, leading to inaccurate gene counts. Existing methods, including advanced deep learning techniques, struggle to reliably impute gene expressions due to a lack of mechanisms that explicitly consider the underlying biological knowledge of the system. In reality, it has long been recognized that gene-gene interactions may serve as reflective indicators of underlying biology processes, presenting discriminative signatures of the cells. A genomic data analysis framework that is capable of leveraging the underlying gene-gene interactions is thus highly desirable and could allow for more reliable identification of distinctive patterns of the genomic data through extraction and integration of intricate biological characteristics of the genomic data. Here we tackle the problem in two steps to exploit the gene-gene interactions of the system. We first reposition the genes into a 2D grid such that their spatial configuration reflects their interactive relationships. To alleviate the need for labeled ground truth gene expression datasets, a self-supervised 2D convolutional neural network is employed to extract the contextual features of the interactions from the spatially configured genes and impute the omitted values. Extensive experiments with both simulated and experimental scRNA-seq datasets are carried out to demonstrate the superior performance of the proposed strategy against the existing imputation methods.