-
EndoUIC: Promptable Diffusion Transformer for Unified Illumination Correction in Capsule Endoscopy
Authors:
Long Bai,
Tong Chen,
Qiaozhi Tan,
Wan Jun Nah,
Yanheng Li,
Zhicheng He,
Sishen Yuan,
Zhen Chen,
Jinlin Wu,
Mobarakol Islam,
Zhen Li,
Hongbin Liu,
Hongliang Ren
Abstract:
Wireless Capsule Endoscopy (WCE) is highly valued for its non-invasive and painless approach, though its effectiveness is compromised by uneven illumination from hardware constraints and complex internal dynamics, leading to overexposed or underexposed images. While researchers have discussed the challenges of low-light enhancement in WCE, the issue of correcting for different exposure levels rema…
▽ More
Wireless Capsule Endoscopy (WCE) is highly valued for its non-invasive and painless approach, though its effectiveness is compromised by uneven illumination from hardware constraints and complex internal dynamics, leading to overexposed or underexposed images. While researchers have discussed the challenges of low-light enhancement in WCE, the issue of correcting for different exposure levels remains underexplored. To tackle this, we introduce EndoUIC, a WCE unified illumination correction solution using an end-to-end promptable diffusion transformer (DiT) model. In our work, the illumination prompt module shall navigate the model to adapt to different exposure levels and perform targeted image enhancement, in which the Adaptive Prompt Integration (API) and Global Prompt Scanner (GPS) modules shall further boost the concurrent representation learning between the prompt parameters and features. Besides, the U-shaped restoration DiT model shall capture the long-range dependencies and contextual information for unified illumination restoration. Moreover, we present a novel Capsule-endoscopy Exposure Correction (CEC) dataset, including ground-truth and corrupted image pairs annotated by expert photographers. Extensive experiments against a variety of state-of-the-art (SOTA) methods on four datasets showcase the effectiveness of our proposed method and components in WCE illumination restoration, and the additional downstream experiments further demonstrate its utility for clinical diagnosis and surgical assistance.
△ Less
Submitted 8 July, 2024; v1 submitted 19 June, 2024;
originally announced June 2024.
-
Ultrafast 3-D Super Resolution Ultrasound using Row-Column Array specific Coherence-based Beamforming and Rolling Acoustic Sub-aperture Processing: In Vitro, In Vivo and Clinical Study
Authors:
Joseph Hansen-Shearer,
Jipeng Yan,
Marcelo Lerendegui,
Biao Huang,
Matthieu Toulemonde,
Kai Riemer,
Qingyuan Tan,
Johanna Tonko,
Peter D. Weinberg,
Chris Dunsby,
Meng-Xing Tang
Abstract:
The row-column addressed array is an emerging probe for ultrafast 3-D ultrasound imaging. It achieves this with far fewer independent electronic channels and a wider field of view than traditional 2-D matrix arrays, of the same channel count, making it a good candidate for clinical translation. However, the image quality of row-column arrays is generally poor, particularly when investigating tissu…
▽ More
The row-column addressed array is an emerging probe for ultrafast 3-D ultrasound imaging. It achieves this with far fewer independent electronic channels and a wider field of view than traditional 2-D matrix arrays, of the same channel count, making it a good candidate for clinical translation. However, the image quality of row-column arrays is generally poor, particularly when investigating tissue. Ultrasound localisation microscopy allows for the production of super-resolution images even when the initial image resolution is not high. Unfortunately, the row-column probe can suffer from imaging artefacts that can degrade the quality of super-resolution images as `secondary' lobes from bright microbubbles can be mistaken as microbubble events, particularly when operated using plane wave imaging. These false events move through the image in a physiologically realistic way so can be challenging to remove via tracking, leading to the production of 'false vessels'. Here, a new type of rolling window image reconstruction procedure was developed, which integrated a row-column array-specific coherence-based beamforming technique with acoustic sub-aperture processing for the purposes of reducing `secondary' lobe artefacts, noise and increasing the effective frame rate. Using an {\it{in vitro}} cross tube, it was found that the procedure reduced the percentage of `false' locations from $\sim$26\% to $\sim$15\% compared to traditional orthogonal plane wave compounding. Additionally, it was found that the noise could be reduced by $\sim$7 dB and that the effective frame rate could be increased to over 4000 fps. Subsequently, {\it{in vivo}} ultrasound localisation microscopy was used to produce images non-invasively of a rabbit kidney and a human thyroid.
△ Less
Submitted 15 November, 2023;
originally announced November 2023.
-
Large-scale environment mapping and immersive human-robot interaction for agricultural mobile robot teleoperation
Authors:
Tao Liu,
Baohua Zhang,
Qianqiu Tan
Abstract:
Remote operation is a crucial solution to problems encountered in agricultural machinery operations. However, traditional video streaming control methods fall short in overcoming the challenges of single perspective views and the inability to obtain 3D information. In light of these issues, our research proposes a large-scale digital map reconstruction and immersive human-machine remote control fr…
▽ More
Remote operation is a crucial solution to problems encountered in agricultural machinery operations. However, traditional video streaming control methods fall short in overcoming the challenges of single perspective views and the inability to obtain 3D information. In light of these issues, our research proposes a large-scale digital map reconstruction and immersive human-machine remote control framework for agricultural scenarios. In our methodology, a DJI unmanned aerial vehicle(UAV) was utilized for data collection, and a novel video segmentation approach based on feature points was introduced. To tackle texture richness variability, an enhanced Structure from Motion (SfM) using superpixel segmentation was implemented. This method integrates the open Multiple View Geometry (openMVG) framework along with Local Features from Transformers (LoFTR). The enhanced SfM results in a point cloud map, which is further processed through Multi-View Stereo (MVS) to generate a complete map model. For control, a closed-loop system utilizing TCP for VR control and positioning of agricultural machinery was introduced. Our system offers a fully visual-based immersive control method, where upon connection to the local area network, operators can utilize VR for immersive remote control. The proposed method enhances both the robustness and convenience of the reconstruction process, thereby significantly facilitating operators in acquiring more comprehensive on-site information and engaging in immersive remote control operations. The code is available at: https://github.com/LiuTao1126/Enhance-SFM
△ Less
Submitted 1 March, 2024; v1 submitted 14 August, 2023;
originally announced August 2023.
-
Transthoracic super-resolution ultrasound localisation microscopy of myocardial vasculature in patients
Authors:
Jipeng Yan,
Biao Huang,
Johanna Tonko,
Matthieu Toulemonde,
Joseph Hansen-Shearer,
Qingyuan Tan,
Kai Riemer,
Konstantinos Ntagiantas,
Rasheda A Chowdhury,
Pier Lambiase,
Roxy Senior,
Meng-Xing Tang
Abstract:
Micro-vascular flow in the myocardium is of significant importance clinically but remains poorly understood. Up to 25% of patients with symptoms of coronary heart diseases have no obstructive coronary arteries and have suspected microvascular diseases. However, such microvasculature is difficult to image in vivo with existing modalities due to the lack of resolution and sensitivity. Here, we demon…
▽ More
Micro-vascular flow in the myocardium is of significant importance clinically but remains poorly understood. Up to 25% of patients with symptoms of coronary heart diseases have no obstructive coronary arteries and have suspected microvascular diseases. However, such microvasculature is difficult to image in vivo with existing modalities due to the lack of resolution and sensitivity. Here, we demonstrate the feasibility of transthoracic super-resolution ultrasound localisation microscopy (SRUS/ULM) of myocardial microvasculature and hemodynamics in a large animal model and in patients, using a cardiac phased array probe with a customised data acquisition and processing pipeline. A multi-level motion correction strategy was proposed. A tracking framework incorporating multiple features and automatic parameter initialisations was developed to reconstruct microcirculation. In two patients with impaired myocardial function, we have generated SRUS images of myocardial vascular structure and flow with a resolution that is beyond the wave-diffraction limit (half a wavelength), using data acquired within a breath hold. Myocardial SRUS/ULM has potential to improve the understanding of myocardial microcirculation and the management of patients with cardiac microvascular diseases.
△ Less
Submitted 28 March, 2023; v1 submitted 24 March, 2023;
originally announced March 2023.
-
Linearized Integrated Microwave Photonic Circuit for Filtering and Phase Shifting
Authors:
Gaojian Liu,
Kaixuan Ye,
Okky Daulay,
Qinggui Tan,
Hongxi Yu,
David Marpaung
Abstract:
Photonic integration, advanced functionality, reconfigurability, and high RF performance are key features in integrated microwave photonic systems that are still difficult to achieve simultaneously. In this work, we demonstrate an integrated microwave photonic circuit that can be reconfigured for two distinct RF functions, namely, a tunable notch filter and a phase shifter. We achieved $>$50dB hig…
▽ More
Photonic integration, advanced functionality, reconfigurability, and high RF performance are key features in integrated microwave photonic systems that are still difficult to achieve simultaneously. In this work, we demonstrate an integrated microwave photonic circuit that can be reconfigured for two distinct RF functions, namely, a tunable notch filter and a phase shifter. We achieved $>$50dB high-extinction notch filtering over 6-16 GHz and 2$π$ continuously tunable phase shifting over 12-20 GHz frequencies. At the same time, we implemented an on-chip linearization technique to achieve a spurious-free dynamic range of more than 120$\rm{dB}\cdot \rm{Hz}^{4/5}$ for both functions. Our work combines multi-functionality and linearization in one photonic integrated circuit, and paves the way to reconfigurable RF photonic front-ends with very high performance.
△ Less
Submitted 26 February, 2023;
originally announced February 2023.
-
Computation Resource Allocation Solution in Recommender Systems
Authors:
Xun Yang,
Yunli Wang,
Cheng Chen,
Qing Tan,
Chuan Yu,
Jian Xu,
Xiaoqiang Zhu
Abstract:
Recommender systems rely heavily on increasing computation resources to improve their business goal. By deploying computation-intensive models and algorithms, these systems are able to inference user interests and exhibit certain ads or commodities from the candidate set to maximize their business goals. However, such systems are facing two challenges in achieving their goals. On the one hand, fac…
▽ More
Recommender systems rely heavily on increasing computation resources to improve their business goal. By deploying computation-intensive models and algorithms, these systems are able to inference user interests and exhibit certain ads or commodities from the candidate set to maximize their business goals. However, such systems are facing two challenges in achieving their goals. On the one hand, facing massive online requests, computation-intensive models and algorithms are pushing their computation resources to the limit. On the other hand, the response time of these systems is strictly limited to a short period, e.g. 300 milliseconds in our real system, which is also being exhausted by the increasingly complex models and algorithms.
In this paper, we propose the computation resource allocation solution (CRAS) that maximizes the business goal with limited computation resources and response time. We comprehensively illustrate the problem and formulate such a problem as an optimization problem with multiple constraints, which could be broken down into independent sub-problems. To solve the sub-problems, we propose the revenue function to facilitate the theoretical analysis, and obtain the optimal computation resource allocation strategy. To address the applicability issues, we devise the feedback control system to help our strategy constantly adapt to the changing online environment. The effectiveness of our method is verified by extensive experiments based on the real dataset from Taobao.com. We also deploy our method in the display advertising system of Alibaba. The online results show that our computation resource allocation solution achieves significant business goal improvement without any increment of computation cost, which demonstrates the efficacy of our method in real industrial practice.
△ Less
Submitted 3 March, 2021;
originally announced March 2021.
-
Bid Optimization by Multivariable Control in Display Advertising
Authors:
Xun Yang,
Yasong Li,
Hao Wang,
Di Wu,
Qing Tan,
Jian Xu,
Kun Gai
Abstract:
Real-Time Bidding (RTB) is an important paradigm in display advertising, where advertisers utilize extended information and algorithms served by Demand Side Platforms (DSPs) to improve advertising performance. A common problem for DSPs is to help advertisers gain as much value as possible with budget constraints. However, advertisers would routinely add certain key performance indicator (KPI) cons…
▽ More
Real-Time Bidding (RTB) is an important paradigm in display advertising, where advertisers utilize extended information and algorithms served by Demand Side Platforms (DSPs) to improve advertising performance. A common problem for DSPs is to help advertisers gain as much value as possible with budget constraints. However, advertisers would routinely add certain key performance indicator (KPI) constraints that the advertising campaign must meet due to practical reasons. In this paper, we study the common case where advertisers aim to maximize the quantity of conversions, and set cost-per-click (CPC) as a KPI constraint. We convert such a problem into a linear programming problem and leverage the primal-dual method to derive the optimal bidding strategy. To address the applicability issue, we propose a feedback control-based solution and devise the multivariable control system. The empirical study based on real-word data from Taobao.com verifies the effectiveness and superiority of our approach compared with the state of the art in the industry practices.
△ Less
Submitted 24 May, 2019;
originally announced May 2019.