-
An Accurate and Explainable Deep Learning System Improves Interobserver Agreement in the Interpretation of Chest Radiograph
Authors:
Hieu H. Pham,
Ha Q. Nguyen,
Hieu T. Nguyen,
Linh T. Le,
Lam Khanh
Abstract:
Recent artificial intelligence (AI) algorithms have achieved radiologist-level performance on various medical classification tasks. However, only a few studies addressed the localization of abnormal findings from CXR scans, which is essential in explaining the image-level classification to radiologists. We introduce in this paper an explainable deep learning system called VinDr-CXR that can classi…
▽ More
Recent artificial intelligence (AI) algorithms have achieved radiologist-level performance on various medical classification tasks. However, only a few studies addressed the localization of abnormal findings from CXR scans, which is essential in explaining the image-level classification to radiologists. We introduce in this paper an explainable deep learning system called VinDr-CXR that can classify a CXR scan into multiple thoracic diseases and, at the same time, localize most types of critical findings on the image. VinDr-CXR was trained on 51,485 CXR scans with radiologist-provided bounding box annotations. It demonstrated a comparable performance to experienced radiologists in classifying 6 common thoracic diseases on a retrospective validation set of 3,000 CXR scans, with a mean area under the receiver operating characteristic curve (AUROC) of 0.967 (95% confidence interval [CI]: 0.958-0.975). The VinDr-CXR was also externally validated in independent patient cohorts and showed its robustness. For the localization task with 14 types of lesions, our free-response receiver operating characteristic (FROC) analysis showed that the VinDr-CXR achieved a sensitivity of 80.2% at the rate of 1.0 false-positive lesion identified per scan. A prospective study was also conducted to measure the clinical impact of the VinDr-CXR in assisting six experienced radiologists. The results indicated that the proposed system, when used as a diagnosis supporting tool, significantly improved the agreement between radiologists themselves with an increase of 1.5% in mean Fleiss' Kappa. We also observed that, after the radiologists consulted VinDr-CXR's suggestions, the agreement between each of them and the system was remarkably increased by 3.3% in mean Cohen's Kappa.
△ Less
Submitted 6 August, 2022;
originally announced August 2022.
-
On proportional fairness of uplink spectral efficiency in cell-free massive MIMO systems
Authors:
Viet Quoc Pham,
Ha Hoang Kha,
Le Ty Khanh
Abstract:
This paper is concerned with the proportional fairness (PF) of the spectral efficiency (SE) maximization of uplinks in a cell-free (CF) massive multiple-input multiple-output (MIMO) system in which a large number of single-antenna access points (APs) connected to a central processing unit (CPU) serve many single-antenna users. To detect the user signals, the APs use matched filters based on the lo…
▽ More
This paper is concerned with the proportional fairness (PF) of the spectral efficiency (SE) maximization of uplinks in a cell-free (CF) massive multiple-input multiple-output (MIMO) system in which a large number of single-antenna access points (APs) connected to a central processing unit (CPU) serve many single-antenna users. To detect the user signals, the APs use matched filters based on the local channel state information while the CPU deploys receiver filters based on knowledge of channel statistics. We devise the maximization problem of the SE PF, which maximizes the sum of the logarithm of the achievable user rates, as a jointly nonconvex optimization problem of receiver filter coefficients and user power allocation subject to user power constraints. To handle the challenges associated with the nonconvexity of the formulated design problem, we develop an iterative algorithm by alternatively finding optimal filter coefficients at the CPU and transmit powers at the users. While the filter coefficient design is formulated as a generalized eigenvalue problem, the power allocation problem is addressed by a gradient projection (GP) approach. Simulation results show that the SE PF maximization not only offers approximately the achievable sum rates as compared to the sum-rate maximization but also provides an improved trade-off between the user rate fairness and the achievable sum rate.
△ Less
Submitted 15 October, 2021;
originally announced October 2021.
-
Deep Learning for Uplink Spectral Efficiency in Cell-Free Massive MIMO Systems
Authors:
Le Ty Khanh,
Pham Quoc Viet,
Ha Hoang Kha,
Nguyen Minh Hoang
Abstract:
In this paper, we introduce a Deep Neural Network (DNN) to maximize the Proportional Fairness (PF) of the Spectral Efficiency (SE) of uplinks in Cell-Free (CF) massive Multiple-Input Multiple-Output (MIMO) systems. The problem of maximizing the PF of the SE is a non-convex optimization problem in the design variables. We will develop a DNN which takes pilot sequences and large-scale fading coeffic…
▽ More
In this paper, we introduce a Deep Neural Network (DNN) to maximize the Proportional Fairness (PF) of the Spectral Efficiency (SE) of uplinks in Cell-Free (CF) massive Multiple-Input Multiple-Output (MIMO) systems. The problem of maximizing the PF of the SE is a non-convex optimization problem in the design variables. We will develop a DNN which takes pilot sequences and large-scale fading coefficients of the users as inputs and produces the outputs of optimal transmit powers. By consisting of densely residual connections between layers, the proposed DNN can efficiently exploit the hierarchical features of the input and motivates the feed-forward nature of DNN architecture. Experimental results showed that, compared to the conventional iterative optimization algorithm, the proposed DNN has excessively lower computational complexity with the trade-off of approximately only 1% loss in the sum rate and the fairness performance. This demonstrated that our proposed DNN is reasonably suitable for real-time signal processing in CF massive MIMO systems.
△ Less
Submitted 10 October, 2021;
originally announced October 2021.
-
The effects of institutional quality on formal and informal borrowing across high-, middle-, and low-income countries
Authors:
Lan Chu Khanh
Abstract:
This paper examines the effects of institutional quality on financing choice of individual using a large dataset of 137,160 people from 131 countries. We classify borrowing activities into three categories, including formal, constructive informal, and underground borrowing. Although the result shows that better institutions aids the uses of formal borrowing, the impact of institutions on construct…
▽ More
This paper examines the effects of institutional quality on financing choice of individual using a large dataset of 137,160 people from 131 countries. We classify borrowing activities into three categories, including formal, constructive informal, and underground borrowing. Although the result shows that better institutions aids the uses of formal borrowing, the impact of institutions on constructive informal and underground borrowing among three country sub-groups differs. Higher institutional quality improves constructive informal borrowing in middle-income countries but reduces the use of underground borrowing in high- and low-income countries.
△ Less
Submitted 19 March, 2019;
originally announced March 2019.
