-
Acute Lymphoblastic Leukemia Detection from Microscopic Images Using Weighted Ensemble of Convolutional Neural Networks
Authors:
Chayan Mondal,
Md. Kamrul Hasan,
Md. Tasnim Jawad,
Aishwariya Dutta,
Md. Rabiul Islam,
Md. Abdul Awal,
Mohiuddin Ahmad
Abstract:
Acute Lymphoblastic Leukemia (ALL) is a blood cell cancer characterized by numerous immature lymphocytes. Even though automation in ALL prognosis is an essential aspect of cancer diagnosis, it is challenging due to the morphological correlation between malignant and normal cells. The traditional ALL classification strategy demands experienced pathologists to carefully read the cell images, which i…
▽ More
Acute Lymphoblastic Leukemia (ALL) is a blood cell cancer characterized by numerous immature lymphocytes. Even though automation in ALL prognosis is an essential aspect of cancer diagnosis, it is challenging due to the morphological correlation between malignant and normal cells. The traditional ALL classification strategy demands experienced pathologists to carefully read the cell images, which is arduous, time-consuming, and often suffers inter-observer variations. This article has automated the ALL detection task from microscopic cell images, employing deep Convolutional Neural Networks (CNNs). We explore the weighted ensemble of different deep CNNs to recommend a better ALL cell classifier. The weights for the ensemble candidate models are estimated from their corresponding metrics, such as accuracy, F1-score, AUC, and kappa values. Various data augmentations and pre-processing are incorporated for achieving a better generalization of the network. We utilize the publicly available C-NMC-2019 ALL dataset to conduct all the comprehensive experiments. Our proposed weighted ensemble model, using the kappa values of the ensemble candidates as their weights, has outputted a weighted F1-score of 88.6 %, a balanced accuracy of 86.2 %, and an AUC of 0.941 in the preliminary test set. The qualitative results displaying the gradient class activation maps confirm that the introduced model has a concentrated learned region. In contrast, the ensemble candidate models, such as Xception, VGG-16, DenseNet-121, MobileNet, and InceptionResNet-V2, separately produce coarse and scatter learned areas for most example cases. Since the proposed kappa value-based weighted ensemble yields a better result for the aimed task in this article, it can experiment in other domains of medical diagnostic applications.
△ Less
Submitted 9 May, 2021;
originally announced May 2021.
-
Transient Stability Assessment for Current Constrained and Unconstrained Fault Ride-Through in Virtual Oscillator Controlled Converters
Authors:
M A Awal,
Iqbal Husain
Abstract:
Unified virtual oscillator controller (uVOC) inherits the rigorous analytical foundation offered by oscillator based grid-forming (GFM) controllers and enables fast over-current limiting and fault ride-through (FRT). Control design for effective FRT requires transient stability analysis. Existing transient stability analysis methods and studies are limited in either considering only current uncons…
▽ More
Unified virtual oscillator controller (uVOC) inherits the rigorous analytical foundation offered by oscillator based grid-forming (GFM) controllers and enables fast over-current limiting and fault ride-through (FRT). Control design for effective FRT requires transient stability analysis. Existing transient stability analysis methods and studies are limited in either considering only current unconstrained scenarios or neglecting the simultaneous power angle and voltage dynamics. Under current-constrained faults, the voltage and power angle dynamics are strongly coupled and both play critical roles in determining transient stability. Therefore, decoupled analysis of the two, typically used in transient stability studies, does not offer comprehensive insight into the system dynamics. In this work, the overall FRT method for uVOC is presented and a comprehensive modeling and analysis method for transient stability is developed under both current-saturated and unsaturated symmetrical AC faults. We utilize phase-plane analysis of the overall system in a single graphical representation to obtain holistic insights into the coupled voltage and power angle dynamics. The FRT controller and the analysis method have been validated through simulations and hardware experiments. The results demonstrate that uVOC is not constrained by a critical clearing angle unlike droop and virtual synchronous machine (VSM) type second order controllers.
△ Less
Submitted 18 November, 2020;
originally announced November 2020.
-
Unified Virtual Oscillator Control for Grid-Forming and Grid-Following Converters
Authors:
M A Awal,
Iqbal Husain
Abstract:
A unified virtual oscillator controller (uVOC) is proposed, which enables a unified analysis, design, and implementation framework for both grid-forming (GFM) and grid-following (GFL) voltage source converters (VSCs). Oscillator based GFM controllers, such as dispatchable virtual oscillator control (dVOC), offer rigorous analytical framework with enhanced synchronization, but lack effective fault…
▽ More
A unified virtual oscillator controller (uVOC) is proposed, which enables a unified analysis, design, and implementation framework for both grid-forming (GFM) and grid-following (GFL) voltage source converters (VSCs). Oscillator based GFM controllers, such as dispatchable virtual oscillator control (dVOC), offer rigorous analytical framework with enhanced synchronization, but lack effective fault handling capability which severely limits practical application. The proposed uVOC facilitates synchronization with an arbitrarily low grid voltage and fast over-current limiting; this enables effective fault ride-through unlike existing GFM controllers which typically switch to a back-up controller during fault. GFM operation with uVOC is achieved in both grid connected and islanded modes with seamless transition between the two. In GFL converters, bidirectional power flow control and DC bus voltage regulation is achieved with uVOC. No phase-locked-loop (PLL) is required for either GFL or GFM operation circumventing the synchronization issues associated with PLLs in weak grid applications. Detail small signal models for GFM and GFL operation have been developed and systematic design guidelines for controller parameters are provided. The proposed controller is validated through hardware experiments in a hybrid AC-DC microgrid.
△ Less
Submitted 12 July, 2020;
originally announced July 2020.
-
Modular Medium Voltage AC to Low Voltage DC Converter for Extreme Fast Charging Applications
Authors:
M A Awal,
Iqbal Husain,
Md Rashed Hassan Bipu,
Oscar Andreas Montes,
Fei Teng,
Hao Feng,
Mehnaz Khan,
Srdjan Lukic
Abstract:
A modular and scalable converter for medium voltage (MV) AC to low voltage (LV) DC power conversion is proposed; single-phase-modules (SPMs), each consisting of an active-front-end (AFE) stage and an isolated DC-DC stage, are connected in input-series-output-parallel (ISOP) configuration to reach desired voltage and power capacity. In prior art, high-speed bidirectional communication among modules…
▽ More
A modular and scalable converter for medium voltage (MV) AC to low voltage (LV) DC power conversion is proposed; single-phase-modules (SPMs), each consisting of an active-front-end (AFE) stage and an isolated DC-DC stage, are connected in input-series-output-parallel (ISOP) configuration to reach desired voltage and power capacity. In prior art, high-speed bidirectional communication among modules and a centralized controller is required to ensure module-level voltage and power balancing, which severely limits the scalability and practical realization of higher voltage and higher power systems. Moreover, large capacitors are used to suppress double-line-frequency voltage variations on the common MV DC bus shared by the AFE and the DC-DC stage originating from AC power pulsations through the SPMs. We propose a comprehensive controller which achieves voltage and power balancing using complete decentralized control of the DC-DC stages based on only local sensor feedback and the AFE stages are controlled using feedback of only the LV DC output. Furthermore, reduced capacitor requirement on the MV DC bus is achieved through design and control. The proposed method is validated through simulation and experimental results.
△ Less
Submitted 8 July, 2020;
originally announced July 2020.
