A TinyML based Portable, Low-Cost Microwave Head Imaging System for Brain Stroke Detection

Hashir, Muhammad; Khalid, Nazish; Mahmood, Nasir; Rehman, Muhammad Attiq; Asad, Muhammad Usman; Mehmood, Muhammad Qasim; Zubair, Muhammad; Massoud, Yehia

doi:10.1109/iscas46773.2023.10181746

Cited by 12 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, it has shown success in the prediction of the BGLs of patients with T1DM with a CGM sensor and a recurrent neural network that builds on long-short term memory (LSTM) [ 61 ]. This tool has proven effectiveness in the broader health domain [ 62 ], facilitating predictions of vital metrics such as blood pressure, cough detection, the pre-screening of oral tongue lesions, and employing a head imaging system for brain stroke detection [ 63 , 64 , 65 , 66 ].…”

Section: Introductionmentioning

confidence: 99%

Noninvasive Diabetes Detection through Human Breath Using TinyML-Powered E-Nose

Gudiño-Ochoa,

García-Rodríguez,

Ochoa-Ornelas

et al. 2024

Sensors

View full text Add to dashboard Cite

Volatile organic compounds (VOCs) in exhaled human breath serve as pivotal biomarkers for disease identification and medical diagnostics. In the context of diabetes mellitus, the noninvasive detection of acetone, a primary biomarker using electronic noses (e-noses), has gained significant attention. However, employing e-noses requires pre-trained algorithms for precise diabetes detection, often requiring a computer with a programming environment to classify newly acquired data. This study focuses on the development of an embedded system integrating Tiny Machine Learning (TinyML) and an e-nose equipped with Metal Oxide Semiconductor (MOS) sensors for real-time diabetes detection. The study encompassed 44 individuals, comprising 22 healthy individuals and 22 diagnosed with various types of diabetes mellitus. Test results highlight the XGBoost Machine Learning algorithm’s achievement of 95% detection accuracy. Additionally, the integration of deep learning algorithms, particularly deep neural networks (DNNs) and one-dimensional convolutional neural network (1D-CNN), yielded a detection efficacy of 94.44%. These outcomes underscore the potency of combining e-noses with TinyML in embedded systems, offering a noninvasive approach for diabetes mellitus detection.

show abstract

Section: Introductionmentioning

confidence: 99%

Noninvasive Diabetes Detection through Human Breath Using TinyML-Powered E-Nose

Gudiño-Ochoa,

García-Rodríguez,

Ochoa-Ornelas

et al. 2024

Sensors

View full text Add to dashboard Cite

show abstract

“…In various application scenarios of wearable devices, such as heart rate estimation [11] and human activity recognition [12], it is necessary to apply the Fourier Transform (FT) to convert sensor sampling data from the time-domain signals into the frequency-domain signals before performing intelligent inference. Moreover, the two-dimensional Fourier Transform has been widely used in edge intelligence scenarios such as road object detection [13] and stroke detection [14]. With the advancement in signal sampling rate and resolution, the computational consumption of the FT has notably increased, especially for image Fourier Transforms.…”

Section: Introductionmentioning

confidence: 99%

Research on High-Performance Fourier Transform Algorithms Based on the NPU

Li,

Zuo,

Feng

et al. 2024

Applied Sciences

View full text Add to dashboard Cite

Backpack computers require powerful, intelligent computing capabilities for field wearables while taking energy consumption into careful consideration. A recommended solution for this demand is the CPU + NPU-based SoC. In many wearable intelligence applications, the Fourier Transform is an essential, computationally intensive preprocessing task. However, due to the unique structure of the NPU, the conventional Fourier Transform algorithms cannot be applied directly to it. This paper proposes two NPU-accelerated Fourier Transform algorithms that leverage the unique hardware structure of the NPU and provides three implementations of those algorithms, namely MM-2DFT, MV-2FFTm, and MV-2FFTv. Then, we benchmarked the speed and energy efficiency of our algorithms for the gray image edge filtering task on the Huawei Atlas200I-DK-A2 development kits against the Cooley-Tukey algorithm running on CPU and GPU platforms. The experiment results reveal MM-2DFT outperforms OpenCL-based FFT on NVIDIA Tegra X2 GPU for small input sizes, with a 4- to 8-time speedup. As the input image resolution exceeds 2048, MV-2FFTv approaches GPU computation speed. Additionally, two scenarios were tested and analyzed for energy efficiency, revealing that cube units of the NPU are more energy efficient. The vector and CPU units are better suited for sparse matrix multiplication and small-scale inputs, respectively.

show abstract

“…A novel approach has been adopted to model chiral metamaterials by employing an on-demand deep learning model. This method offers an alternative to the traditional techniques and can potentially enhance the accuracy of the modeling process [53][54][55][56]. In this paper, we developed broadband metasurfaces that generate polygonal PV beams with different energy distribution shapes, like hexagons and octagons.…”

Section: Introductionmentioning

confidence: 99%

Versatile broadband generation of polygonal perfect vortex beams in the visible spectrum

Danial Shafqat,

Cabrera,

Abbasi

et al. 2024

Nanophotonics X

View full text Add to dashboard Cite

Optical vortex (OV) beams characterized by their twisted wave phase and the orbital angular momentum (OAM) have garnered significant interest and found practical utility in diverse applications like optical communication, particle manipulation, and many more. The topological charge plays a crucial role in defining and recognizing the OV beams, which are controlled by the number of twisted waves and the radius of the intensity profile. The reliance of OV beams on the topological charge poses limitations in numerous applications that necessitate coupling and higher topological charge values. The perfect vortex beams integrated with the Metasurface are the nanoscale solution for the abovementioned issues. Previous studies have controlled the shape and size of the ideal vortex beam. This study demonstrates the broadband generation of polygonal perfect vortex beams through all-glass metasurfaces, aiming to manipulate the beam's shape while preserving its orbital angular momentum (OAM) and radius. The presented metasurfaces carry an array of nanoantennas made of zinc sulfide material to generate a broadband perfect vortex beam within the visible spectrum, from 475 nm to 650 nm. For proof of concept, we have generated OV beams exhibiting multiple shapes like hexagons and octagons, and these shapes helped us make asymmetric intensity profiles. These irregular intensity distributions will help to create flexible optical traps for nanoparticles. The implications of our study include nano-optical trapping, optical manipulation, and optical communication.

show abstract

A TinyML based Portable, Low-Cost Microwave Head Imaging System for Brain Stroke Detection

Cited by 12 publications

References 12 publications

Noninvasive Diabetes Detection through Human Breath Using TinyML-Powered E-Nose

Noninvasive Diabetes Detection through Human Breath Using TinyML-Powered E-Nose

Research on High-Performance Fourier Transform Algorithms Based on the NPU

Versatile broadband generation of polygonal perfect vortex beams in the visible spectrum

Contact Info

Product

Resources

About