Supervised deep learning techniques have demonstrated remarkable performance in segmentation tasks in various domains including medical, biomaterial, and bioengineering fields. The advancement in imaging technologies has enabled the gathering of large amounts of raw data. Nevertheless, acquiring fully annotated ground-truth datasets for supervised deep learning-based segmentation tasks has remained a challenge attributed to the availability of expert resources and the laborious task of pixel-level annotation. Self-supervised learning(SSL) techniques have gained the momentum to learn representations from unlabeled datasets while also preserving the domain-specific information over transfer learning techniques thereby overcoming the challenge of annotating huge datasets. Weakly supervised learning(WSL) techniques have gained attention to address the annotation challenge by using weak annotations such as scribbles, dot annotations, and bounding boxes instead of pixel-wise annotations. In this work, we leveraged the advantages of both SSL and WSL to perform segmentation of single cells in optical images of biofilms attached on single-layer graphene-coated copper substrates guided by scribble annotations. Conditional random fields are applied as a post-processing to further improve segmentation performance. The method showed promising performance in segmenting single cells on biofilm images with scribble annotations that make up only 40-50% of pixel-level annotations.

Biofilms are ubiquitous in aqueous environments, exerting significant influence on diverse surfaces, including metals prone to microbiologically influenced corrosion (MIC). This multifaceted phenomenon demands interdisciplinary collaborations to combat its far-reaching implications. In this context, our research delves into the intricate characterization of twodimensional (2D) materials, particularly hexagonal boron nitride (hBN), which is crucial for advancing corrosion prevention coatings. The nanoscale dimensions of 2D materials pose challenges in microstructural analysis and defect identification, necessitating labor-intensive traditional techniques. To address these complexities, we utilized two unsupervised machine learning models, namely, (a) K-means clustering, and (b) Gaussian Mixture Model (GMM), which enabled clear differentiation between multilayer hBN (MLhBN) and cracks. Our approach will streamline the characterization process and facilitate the extraction of thin layers with enhanced accuracy.

Serious games have been widely used as medium of instruction in various domains. A serious game is composed of various exercises a player has to accomplish in order to achieve the final goal. Each exercise is designed using a set of concepts that a player has to achieve proficiency in. The design of problems is of greater importance in such scenarios than the physical game design. However, traversal through the game is challenging for users with no gaming experience. Spatial distribution of these exercises in the game may influence the learning potential of a player, especially when the exercises have a conceptual dependency associated with them. Our approach in this project is to map the problem graph on to a selected environment in the game for efficient traversal through the game space. Towards this goal, we plan to explore existing environments from Unreal Engine with possible locations for exercises. We will study the theory of graph embedding used to map multi-threaded programs to high performance architecture topologies and explore their adaptation to the problem of exercise distribution in serious game environments.

Scanning electron microscopy (SEM) images play a crucial role in the quantitative analyses of biofilms on materials by providing detailed information about biofilm formation, ultrastructure, cells and their interaction with materials. In addition to some intrinsic limitations of SEM imaging, some of the characteristics of images in SEM volumes often vary in terms of the wide range of magnifications, high resolutions, depth of the field, and the SEM protocols. Quantitative characterization of biofilm morphologies such as the cell size, geometry, and density from these SEM volumes is a challenging problem. This paper presents deep learning based super-resolution (DLSR) as a step towards addressing this problem. Three DLSR approaches based on generative adversarial learning techniques are applied to an SEM biofilm dataset and compared in terms of their a) preservation of the morphological features, and b) their impact on the performance of a deep learning based image segmentation task. Our results show that the DLSR approaches vary in preservation of morphological features. We also show that DLSR can considerably improve the performance of image segmentation outputs of deep learning networks and hence be incorporated in any deep learning pipeline used for quantitative analyses of biofilms based on SEM images.

With the growth of technology capturing huge amounts of image data has become possible including electron microscopic images. Deep learning techniques have been thus drastically improved for analysing images due to their huge availability. Deep learning has been applied in various domains including medical, biomaterial, engineering fields t o analyze complex images. However, supervised deep learning techniques require huge amounts of annotated images. Annotating the images, specifically pixel wise annotations for segmentation tasks could be overwhelming and requires expert resources. Weakly-supervised learning has been popularly employed in such scenarios where weak labels are used for segmentation purposes. Also, self-supervised learning techniques have greatly reduced the amount of labeled data required to train a model for any downstream task. In this study, we would employ self-supervised learning technique followed by scribble supervision for performing biofilm segmentation on optical images. Our initial classification results and the proposed method for segmentation using scribble annotation are provided in this paper.

Detection of microbially influenced corrosion (MIC) products in biofilm images is an important problem in material science and engineering. Neural network models that can accurately detect biofilm images having corrosion products are likely to have significant and broad implications in the study and development of materials. However, generating and annotating biofilm image datasets in sufficient volumes is a major impediment in developing such networks. A self-supervised learning approach is presented for automatically detecting MIC products on metal surfaces based on analyses of Scanning Electron Microscope (SEM) biofilm images. The proposed approach uses a Simple Siamese (SimSiam) architecture to learn visual image representations from an unlabeled set of biofilm images, which is then fine-tuned using a scarce set of labeled images to build a model to detect biofilms containing corrosion products. The architecture generates two different augmented versions of the input images and learns the representations using an encoder that uses ResNet backbone. The architecture aims to minimize the negative cosine similarity of the outputs from the encoders and hence learns the representations of the images as both augmented versions belong to the same image. In order to improve the dataset quality and volume, input images are contrast enhanced, scaled, and overlapping image patches are generated and used to learn representations and fine-tuning. The performance of the models are analyzed using precision and recall metrics for patches of varying sizes. An overall accuracy of 62% was obtained for the classification of corrosion in the images after finetuning the model with scarce labeled dataset. Our results show that the models built using the proposed self-supervised learning approach can successfully detect corrosion products in biofilm images and that the performance of the models successively improves with increase in the patch size.