Balanced Deep Learning-based Bubble Segmentation: Model Comparison, Optimization, and Application in Microbubble Detection

Flow Measurement and Instrumentation, Год журнала: 2025, Номер unknown, С. 102907 - 102907

Опубликована: Апрель 1, 2025

Язык: Английский

---

A Lightweight Neural Network for Cell Segmentation Based on Attention Enhancement

Information, Год журнала: 2025, Номер 16(4), С. 295 - 295

Опубликована: Апрель 8, 2025

Deep neural networks have made significant strides in medical image segmentation tasks, but their large-scale parameters and high computational complexity limit applicability on resource-constrained edge devices. To address this challenge, paper introduces a lightweight nuclear network called Attention-Enhanced U-Net (AttE-Unet) for cell segmentation. AttE-Unet enhances the network’s feature extraction capabilities through an attention mechanism combines strengths of deep learning with traditional filtering algorithms, while substantially reducing storage demands. Experimental results PanNuke dataset demonstrate that AttE-Unet, despite its reduction model size—with number floating-point operations per second reduced to 1.57% 0.1% original model, respectively—still maintains level performance. Specifically, F1 score Intersection over Union (IoU) are 91.7% 89.3% model’s scores. Furthermore, deployment MCU consumes only 2.09 MB Flash 1.38 RAM, highlighting nature potential practical as solution

Язык: Английский

---

Advancing Breast Cancer Diagnosis: Integrating Deep Transfer Learning and U-Net Segmentation for Precise Classification and Delineation of Ultrasound Images

Results in Engineering, Год журнала: 2025, Номер unknown, С. 105047 - 105047

Опубликована: Апрель 1, 2025

Язык: Английский

---

A Unified Multi-Task Learning Model with Joint Reverse Optimization for Simultaneous Skin Lesion Segmentation and Diagnosis

Bioengineering, Год журнала: 2024, Номер 11(11), С. 1173 - 1173

Опубликована: Ноя. 20, 2024

Classifying and segmenting skin cancer represent pivotal objectives for automated diagnostic systems that utilize dermoscopy images. However, these tasks present significant challenges due to the diverse shape variations of lesions inherently fuzzy nature images, including low contrast presence artifacts. Given robust correlation between classification their segmentation, we propose employing a combined learning method holds promise considerably enhancing performance both tasks. In this paper, unified multi-task strategy concurrently classifies abnormalities allows joint segmentation lesion boundaries. This approach integrates an optimization technique known as reverse learning, which fosters mutual enhancement through extracting shared features limiting task dominance across two The effectiveness proposed was assessed using publicly available datasets, ISIC 2016 PH

Язык: Английский

---

Enhanced Panoramic Radiograph-Based Tooth Segmentation and Identification Using an Attention Gate-Based Encoder–Decoder Network

Diagnostics, Год журнала: 2024, Номер 14(23), С. 2719 - 2719

Опубликована: Дек. 3, 2024

Dental disorders are one of the most important health problems, affecting billions people all over world. Early diagnosis is for effective treatment planning. Precise dental disease segmentation requires reliable tooth numbering, which may be prone to errors if performed manually. These steps can automated using artificial intelligence, provide fast and accurate results. Among AI methodologies, deep learning has recently shown excellent performance in image processing, allowing numbering.

Язык: Английский

---

Relationship between pancreatic morphological changes and diabetes in autoimmune pancreatitis: Multimodal medical imaging assessment has important potential

World Journal of Radiology, Год журнала: 2024, Номер 16(11), С. 703 - 707

Опубликована: Ноя. 26, 2024

Autoimmune pancreatitis (AIP) is a special type of chronic with clinical symptoms obstructive jaundice and abdominal discomfort; this condition caused by autoimmunity marked pancreatic fibrosis dysfunction. Previous studies have revealed close relationship between early atrophy the incidence rate diabetes in 1 AIP patients receiving steroid treatment. Shimada et al performed long-term follow-up study reported that volume (PV) these initially exponentially decreased but then slowly decreased, which was considered to be an important factor related diabetes; moreover, serum IgG4 levels were positively correlated PV during follow-up. In letter, regarding original presented , we present our insights discuss how multimodal medical imaging artificial intelligence can used better assess morphological changes AIP.

Язык: Английский

---

Detection of Pneumonia in Children through Chest Radiographs using Artificial Intelligence in a Low-Resource Setting: A Pilot Study

medRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Дек. 2, 2024

Pneumonia is a leading cause of death among children under 5 years in low-and-middle-income-countries (LMICs), causing an estimated 700,000 deaths annually. This burden compounded by limited diagnostic imaging expertise. Artificial intelligence (AI) has potential to improve pneumonia diagnosis from chest radiographs (CXRs) through enhanced accuracy and faster time. However, most AI models lack validation on prospective clinical data LMICs, limiting their real-world applicability. study aims develop validate model for childhood detection using Nigerian CXR data. In multi-center cross-sectional Ibadan, Nigeria, CXRs were prospectively collected University College Hospital (a tertiary hospital) Rainbow-Scans private center) radiology departments via cluster sampling (November 2023–August 2024). An was developed open-source paediatric dataset the USA, classify local as either normal or pneumonia. Two blinded radiologists provided consensus classification reference standard. The model's accuracy, precision, recall, F1-score, area-under-the-curve (AUC) evaluated. 5,232 CXRs, divided into training (1,349 normal, 3,883 pneumonia) internal test (234 390 sets, externally tested 190 radiologist-labeled (93 97 pneumonia). achieved 86% 0.83 0.98 0.79 0.93 AUC test, 58% 0.62 0.48 0.68 0.65 external test. illustrates AI’s but reveals challenges when applied across diverse healthcare environments, revealed discrepancies between evaluations. performance gap likely stems differences protocols/equipment LMICs high-income settings. Hence, public health priority should be developing robust, locally relevant datasets Africa facilitate sustainable independent development within African healthcare.

Язык: Английский

---

Improving Medical Image Segmentation Using Test-Time Augmentation with MedSAM

Mathematics, Год журнала: 2024, Номер 12(24), С. 4003 - 4003

Опубликована: Дек. 20, 2024

Medical image segmentation is crucial for diagnostics and treatment planning, yet traditional methods often struggle with the variability of real-world clinical data. Deep learning models, like Segment Anything Model (SAM), have been proposed as a powerful tool that helps to delimit regions using prompt. This work proposes methodology improve quality by integrating test-time augmentation (TTA) SAM medical applications (MedSAM) random circular shifts, addressing challenges such misalignments imaging variability. The method generates several input variations during inference are combined after, improving robustness accuracy without requiring retraining. Evaluated across diverse computed tomography (CT) datasets, including Segmentation Decathlon (MSD), KiTS, COVID-19-20, demonstrated consistent improvements in Dice Similarity Coefficient (DSC) Normalized Surface (NSD) metrics. highest performances were 93.6% DSC 97% NSD. Notably, it achieved superior boundary precision surface alignment complex pancreas colon, outperforming baseline MedSAM DeepLabv3+. approach computationally feasible, leveraging balance intensity accuracy.

Язык: Английский

---

Profiling Peripheral Artery Structures Using CT Scan Data with Curve Planar Reformation and Gaussian Mixture Model in Polar Coordinates

2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Год журнала: 2024, Номер unknown, С. 7185 - 7187

Опубликована: Дек. 3, 2024

Язык: Английский

---