Bayesian Deep Learning for Image Reconstruction: From structured sparsity to uncertainty estimation

Metasurface has drawn extensive attention due to its capability of modulating light with a high degree of freedom through ultrathin and sub-wavelength optical elements, and metalens, as one of its important applications, promises to replace the bulky refractive optics, facilitating the imaging system light-weight and compact characteristics. Besides, computer-generated holography (CGH) is of substantial interest for three-dimensional (3D) imaging technology by virtue of its ability of restoring the whole optical wave field and re-constructing the true 3D scene. Consequently, the combination of metalens and CGH holds transformative potential in enabling the miniaturization of 3D imaging systems. However, its imaging performance is subject to the aberrations and speckle noises originating from the metalens and CGH. Inspired by recent progress that computational imaging can be applied to close the gap, a novel full-color imaging system, adopting end-to-end joint optimization of metalens and CGH for high imaging quality, is proposed in this paper. The U-net based network as the pre-processing adjusts weights to make the holographic reconstruction offset imaging defects, incorporating the imaging processing into the step of generating hologram. Optimized by deep learning, the proposed imaging system is capable of full-color imaging with high fidelity in a compact form factor, envisioned to take an essential step towards the high-performance miniaturized imaging system.

Imaging systems using aspheric imaging lenses with complementary computation can deliver performance unobtainable in conventional imaging systems. These new imaging systems, termed Wavefront coded imaging systems, use specialized optics to capture a coded image of the scene. Decoding the intermediate image provides the "human-usable" image expected of an imaging system. Computation for the decoding step can be made completely transparent to the user with today's technology. Real-time Wavefront coded systems are feasible and cost-effective. This "computational imaging" technology can be adapted to solve a wide range of imaging problems. Solutions include the ability to provide focus-free imaging, to increase the field of view, to increase the depth of read, to correct for aberrations (even in single lens systems), and to account for assembly and temperature induced misalignment. Wavefront coded imaging has been demonstrated across a wide range of applications, including microscopy, miniature cameras, machine vision systems, infrared imaging systems and telescopes.

In the modern tactical imaging environment, new computational imaging (CI) systems and algorithms are being used to improve speed and accuracy for detection tasks. Therefore, a measurement technique is needed to predict the performance of complex non-shift invariant EO/IR imaging systems, including CI systems. Detection performance of traditional imaging systems can be modeled using current system metrics and measurements such as Modulation Transfer Function (MTF), Signal to Noise (SNR), and instantaneous Field of View (iFOV). In this correspondence, we propose a technique to experimentally measure a detection sensitivity metric for non-traditional CI systems. The detection sensitivity metric predicts the upper bound of linear algorithm performance though evaluation of a matched filter. The experimental results are compared with theoretical expected values though the Night Vision Integrated Performance Model (NV-IPM). Additionally, we demonstrate the experimental results for a variety of imaging systems (IR, visible, and color), target sizes and orientations, as well as SNR values. Our results demonstrate how this detection sensitivity metric can be measured to provide additional insight into the final system performance.

Medical images are acquired for the purpose of gathering information regarding the disease state of a patient. The quality of the image should be a measure of how a specified task may be performed given the resulting image. Image quality is task dependent; one imaging system may provide images that allow better task performance than another system's for one task, but worse for another task. Determining the quality of an imaging system involves specifying the task of interest, the range of objects that will be imaged, and the observer (human or machine reader) who will perform the task. In radiography it has become common to use Fourier measures to describe image quality through a combination of Fourier‐based measures of system resolution and noise. The Line Spread Function, when Fourier transformed, yields the Modulation Transfer Function (MTF) as a measure of system resolution. Use of the MTF to describe a system's resolution is appropriate when the imaging system is shift‐invariant, that is, the line spread function is independent of position. Image noise is described in the Fourier domain by the Noise Power Spectrum (NPS). We can use the NPS to describe the noise in an imaging system whenever the noise properties are stationary, meaning that the variance and correlations between fluctuations in different locations are not dependent on their absolute location. These Fourier measures of system resolution and noise are commonly combined to give an overall figure of merit known as the Detective Quantum Efficiency (DQE). The DQE describes the signal‐to‐noise transfer characteristics of an imaging system as long as those assumptions of shift‐invariance and noise stationarity are valid. In digital systems the image of an object is affected by the location of the object with respect to the pixel grid. In other words, the detector pixels cause the system to be shift‐variant. In this case the MTF does not describe the resolution properties completely. This is particularly the case for tomographic systems, where the point‐response function is quite position‐dependent. Moreover, in real imaging systems the noise properties will depend on position, most notably because real patients always have structure. So, the Fourier‐domain noise concept of Noise Power Spectrum does not capture all the properties of noise for real imaging systems and tasks. In this talk we will consider a general treatment of the imaging system as a mapping of information from the patient to the data, without the assumptions inherent in Fourier analysis. We will discuss approaches to the characterization of system resolution and noise in terms of position‐dependent parameters. Once such position‐dependent imaging properties of a system are determined, we can use them to calculate measures of image quality that summarize the usefulness of the images for the performance of various visual tasks.

Emotion detection, hence, is an important step toward making human-computer interaction a more enhanced process, where systems are made capable of identifying and responding to the emotional state of users. In fact, multimodal emotion detection systems in which both auditory and visual information are fused are emerging, and these approaches toward expressive emotional states are complementary and robust. Multimodal systems enhance the quality of interacting and, through many applications, can diagnose emotional disorders, monitor automotive safety, and improve human-robot interactions. In nature, the high-dimensional space and dynamic threats have resulted in obtaining low accuracy and high computational cost in applying the traditional models based on single-modality data. On the other hand, multimodal systems explore the synergy between audio and visual data, giving better performance and higher accuracy in inferring subtle emotional expressions. The latest improvement was done on these systems using recent advancements in transfer learning and deep learning techniques.That being said, this research Proposal devises a multimodal emotion recognition system integrating speech and face information through transfer learning for improved accuracy and robustness. Serving this purpose, the objectives of this research entail the effective comparison among different transfer-learning strategies, including the impact of pre-trained models in speech-based emotion recognition, and to introduce the role of voice activity detection in the process. Advanced neural network architectures like Spatial Transformer Networks and bidirectional LSTM in facial emotion recognition will also be tested. Early and late fusion strategies will also be used to find the best strategy for combining speech and facial data.This research will target several challenges that involve the complexity of data, balancing of the model performance-robustness balance, computational limitations, and standardization of evaluations in developing a working and robust emotion recognition system to enhance digital interaction and apply in practical areas. The goal is to create a system that oversteps the limitation of single-modality models through state-of-the-art advances in deep learning, as well as front-line improvements in transfer learning, in the manner of emotion detection performance.

Recent advancements in deep learning techniques tailored for autonomous driving in Indian road conditions are crucial for revolutionizing transportation systems. Indian roads present unique challenges, including unpredictable traffic patterns, diverse road infrastructures, and challenging weather conditions. Deep learning is pivotal in addressing these challenges, focusing on perception, decision-making, and control. Various architectures, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Deep Reinforcement Learning (DRL), are analyzed for their efficacy in autonomous driving. Sensor fusion techniques enhance perception capabilities, such as lidar, radar, and camera data integration. Additionally, advancements in semantic segmentation and object detection algorithms improve scene understanding and obstacle recognition. Collaborative efforts between academia, industry, and government agencies are essential for accelerating the deployment of autonomous driving technology in Indian road conditions. In conclusion, leveraging cutting-edge deep learning methodologies enables safe and efficient navigation of autonomous vehicles, ushering in a transformative future in transportation systems.

Quantization-aware training for low precision photonic neural networks

Computational imaging modalities support a simplification of the active architectures required in an imaging system and these approaches have been validated across the electromagnetic spectrum. Recent implementations have utilized pseudo-orthogonal radiation patterns to illuminate an object of interest-notably, frequency-diverse metasurfaces have been exploited as fast and low-cost alternative to conventional coherent imaging systems. However, accurately measuring the complex-valued signals in the frequency domain can be burdensome, particularly for sub-centimeter wavelengths. Here, computational imaging is studied under the relaxed constraint of intensity-only measurements. A novel 3D imaging system is conceived based on 'phaseless' and compressed measurements, with benefits from recent advances in the field of phase retrieval. In this paper, the methodology associated with this novel principle is described, studied, and experimentally demonstrated in the microwave range. A comparison of the estimated images from both complex valued and phaseless measurements are presented, verifying the fidelity of phaseless computational imaging.

Computational imaging is a proven strategy for obtaining high-quality images with fast acquisition rates and simpler hardware. Metasurfaces provide exquisite control over electromagnetic fields, enabling the radiated field to be molded into unique patterns. The fusion of these two concepts can bring about revolutionary advances in the design of imaging systems for security screening. In the context of computational imaging, each field pattern serves as a single measurement of a scene; imaging a scene can then be interpreted as estimating the reflectivity distribution of a target from a set of measurements. As with any computational imaging system, the key challenge is to arrive at a minimal set of measurements from which a diffraction-limited image can be resolved. Here, we show that the information content of a frequency-diverse metasurface aperture can be maximized by design, and used to construct a complete millimeter-wave imaging system spanning a 2 m by 2 m area, consisting of 96 metasurfaces, capable of producing diffraction-limited images of human-scale targets. The metasurfacebased frequency-diverse system presented in this work represents an inexpensive, but tremendously flexible alternative to traditional hardware paradigms, offering the possibility of low-cost, real-time, and ubiquitous screening platforms.

Brain imaging is essential to the clinical care of patients with stroke, a leading cause of disability and death worldwide. Whereas advanced neuroimaging techniques offer opportunities for aiding acute stroke management, several factors, including time delays, inter-clinician variability, and lack of systemic conglomeration of clinical information, hinder their maximal utility. Recent advances in deep machine learning (DL) offer new strategies for harnessing computational medical image analysis to inform decision making in acute stroke. We examine the current state of the field for DL models in stroke triage. First, we provide a brief, clinical practice-focused primer on DL. Next, we examine real-world examples of DL applications in pixel-wise labeling, volumetric lesion segmentation, stroke detection, and prediction of tissue fate postintervention. We evaluate recent deployments of deep neural networks and their ability to automatically select relevant clinical features for acute decision making, reduce inter-rater variability, and boost reliability in rapid neuroimaging assessments, and integrate neuroimaging with electronic medical record (EMR) data in order to support clinicians in routine and triage stroke management. Ultimately, we aim to provide a framework for critically evaluating existing automated approaches, thus equipping clinicians with the ability to understand and potentially apply DL approaches in order to address challenges in clinical practice. ANN NEUROL 2022;92:574-587.

Breast cancer is the most prevalent cancer globally and presents a significant threat to human health. The low survival rate for advanced stages highlights the importance of early and accurate diagnosis. Pathology images, providing insights into cancer's cellular aggressiveness, are crucial for breast cancer diagnosis. Advances in deep learning have transformed automated medical image diagnosis. However, traditional deep learning methods, reliant on large labeled datasets, face challenges with the typically small size of medical image datasets. To overcome this, our study combines deep transfer learning (DTL) with convolutional neural networks(CNN). We utilize and enhance a ResNet50 model pretrained on a large dataset. Additionally, we apply four different deep transfer learning methods to feature extraction and classification of the BreakHis dataset, aiming to determine if breast cancer is benign or malignant. Ultimately, our system achieved the best average accuracy rates under four different magnifications: 98.57%, 96.83%, 97.76%, and 97.16%. These results demonstrate the accuracy and potential of deep transfer learning in classifying breast cancer histopathological images and its promising application in tumor prediction.

The rapid growth of genomic data presents new challenges and opportunities in the field of bioinformatics. Deep learning has emerged as a powerful tool for analyzing large and complex genomic datasets, offering unprecedented capabilities for identifying patterns, predicting gene functions, and understanding genetic variations. This article explores the latest advancements in deep learning techniques applied to genomic data analysis. It discusses key approaches such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and autoencoders, and their applications in various aspects of genomics, including gene expression analysis, variant calling, and drug discovery. The article also highlights the integration of deep learning with other emerging technologies, such as CRISPR, and its potential to revolutionize personalized medicine and genetic research..

The underwater domain presents unique challenges and opportunities for scientific exploration, resource extraction, and environmental monitoring. Autonomous underwater vehicles (AUVs) rely on simultaneous localization and mapping (SLAM) for real-time navigation and mapping in these complex environments. However, traditional SLAM techniques face significant obstacles, including poor visibility, dynamic lighting conditions, sensor noise, and water-induced distortions, all of which degrade the accuracy and robustness of underwater navigation systems. Recent advances in deep learning (DL) have introduced powerful solutions to overcome these challenges. DL techniques enhance underwater SLAM by improving feature extraction, image denoising, distortion correction, and sensor fusion. This survey provides a comprehensive analysis of the latest developments in DL-enhanced SLAM for underwater applications, categorizing approaches based on their methodologies, sensor dependencies, and integration with deep learning models. We critically evaluate the benefits and limitations of existing techniques, highlighting key innovations and unresolved challenges. In addition, we introduce a novel classification framework for underwater SLAM based on its integration with underwater wireless sensor networks (UWSNs). UWSNs offer a collaborative framework that enhances localization, mapping, and real-time data sharing among AUVs by leveraging acoustic communication and distributed sensing. Our proposed taxonomy provides new insights into how communication-aware SLAM methodologies can improve navigation accuracy and operational efficiency in underwater environments. Furthermore, we discuss emerging research trends, including the use of transformer-based architectures, multi-modal sensor fusion, lightweight neural networks for real-time deployment, and self-supervised learning techniques. By identifying gaps in current research and outlining potential directions for future work, this survey serves as a valuable reference for researchers and engineers striving to develop robust and adaptive underwater SLAM solutions. Our findings aim to inspire further advancements in autonomous underwater exploration, supporting critical applications in marine science, deep-sea resource management, and environmental conservation.

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper OTC 35821, “A Mobile Platform for AI-Powered Underwater Image Processing,” by Chenyu Zhou, University of Houston; Lily Walker, William & Mary; and Xin Fu, University of Houston, et al. The paper has not been peer reviewed. Copyright 2025 Offshore Technology Conference. _ Timely inspection of subsea infrastructure—particularly pipelines—is crucial for preventing oil spills and protecting the environment. This paper introduces a technology for offshore pipeline inspection centered on an autonomous robotic system equipped with underwater computer vision and edge-computing capabilities. The aim is to create a fast, cost-effective solution for inspecting underwater structures, enabling early detection of environmental hazards. Introduction Autonomous underwater vehicles (AUVs) have revolutionized the ability to explore and monitor marine environments. The cornerstone of AUV capabilities lies in their imaging systems, although the underwater environment poses unique and complex challenges for these systems. Additional image-processing and correction ability remains a goal for AUVs, creating computational demands that must be balanced against limited onboard power and processing capabilities (Fig. 1). Recently, advancements in deep learning have enabled deep neural networks (DNNs) to achieve remarkable success in solving a wide range of image-processing challenges, significantly surpassing the performance of traditional methods. In this research, the authors build upon foundational work in the literature to present a comprehensive workflow for adapting and deploying pretrained DNN models onto mobile computing platforms suitable for integration with AUVs. The described approach examines and documents the transition from powerful desktop systems to resource-constrained mobile platforms. Underwater Image Enhancement Motion blur is a common challenge in underwater image processing, especially when the camera is mounted on a moving platform such as an AUV. The lack of adequate lighting in underwater environments often requires longer exposure times, further exacerbating motion blur. The blurring process can be mathematically described as a convolution of the sharp image with a blur kernel, plus noise. If the blur kernel is known beforehand, the problem becomes one of nonblind deconvolution. However, when the blur kernel is unknown—common in dynamic underwater settings—it becomes a blind deconvolution problem. This problem is inherently ill-posed because multiple combinations of sharp images and blur kernels can explain the same blurred image. To resolve this ambiguity, assumptions or previous knowledge about the blur kernel or image structure typically are introduced. Traditional blind deconvolution methods often use iterative algorithms, refining the blur kernel and sharp image estimates over multiple steps. These classical approaches, however, often suffer from artifacts such as ringing and struggling with large motions or noise. To address these limitations, researchers have explored regularization techniques that incorporate priors to constrain the solution space. For example, sparsity priors in the temporal domain encourage smooth results, while Bayesian frameworks provide probabilistic formulations of the problem. One promising direction involves leveraging domain-specific priors for underwater imagery. By incorporating this property into optimization frameworks, researchers have developed more-effective and -robust blind deblurring methods tailored to underwater scenarios.

This research seeks to investigate the possibility of using deep learning strategies in the process of diagnosing malaria, a virus that affects billions of people all over the world. Standard lab tests for malaria require the services of a qualified laboratory technician as well as an in-depth analysis of blood samples. This process can be expensive, time-consuming, and prone to errors caused by humans. This work attempts to enhance the accuracy of malaria diagnosis while also increasing the rate at which it can be performed by utilizing the capabilities of deep learning. We evaluate the performance of various methods for identifying the Plasmodium parasite in thin blood smear images by using deep learning models such as CNN, ResNet50, and VGG19 in accordance with noise reduction techniques and image segmentation methods. This allows us to compare the accuracy of the various methods. According to the findings of our research, the VGG19 model had the greatest overall performance. It had an accuracy of 0.9286 as well as a low false-positive and losing rate. The model is also tiny, making it easy to transport and use in a variety of contexts due to its portability. This study gives an overview of the current advancements in deep learning for malaria diagnosis. It also illustrates the potential for AI to increase both the accuracy and speed of malaria diagnosis.