Fetal Diagnostics using Vision Transformer for Enhanced Health and Severity Prediction in Ultrasound Imaging.

IF 1.1 4区 医学 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING
Eshika Jain, Pratham Kaushik, Vinay Kukreja, Sakshi, Ayush Dogra, Bhawna Goyal
{"title":"Fetal Diagnostics using Vision Transformer for Enhanced Health and Severity Prediction in Ultrasound Imaging.","authors":"Eshika Jain, Pratham Kaushik, Vinay Kukreja, Sakshi, Ayush Dogra, Bhawna Goyal","doi":"10.2174/0115734056360199250227053012","DOIUrl":null,"url":null,"abstract":"<p><strong>Aim: </strong>This research aims to develop and evaluate a novel health classification and severity detection system based on Vision Transformers (ViTs) for fetal ultrasound imagery. This contributes to improved precision in fetal health status detection and abnormalities with more accurate results than other traditional models.</p><p><strong>Background: </strong>Amidst the other imperatives of resource-deficient developing nations, mitigating neonatal mortality rates is a challenge that demands precisionbased solutions in the era of artificial intelligence. Though the advent of machine learning models has added an optimal dimension to deal with emerging complexity in fetal ultrasound imagery, there is a call to address the huge gap in the demanded precision for prediction than the existing interpretation.</p><p><strong>Objective: </strong>This research strives to formulate and access a novel health classification and severity detection system based on the implementation of the Vision Transformers frameworks. This pioneering investigation represents an unparalleled exploration into the efficacy of ViTs for discerning intricate patterns within fetal ultrasonographic imagery, facilitating precise categorization of fetal well-being and prognosticating the magnitude of potential anomalies.</p><p><strong>Methodology: </strong>A private and confidential dataset of 500 fetal ultrasound images has been collected from diverse hospitals. Each image has been annotated by radiologists according to two main labels: the health status of the fetus, which includes healthy, mild, moderate, or severe, and the severity of abnormalities as a continuous measure. At different levels, the dataset underwent pre-processing via distinct techniques. Then, the composite loss function Cross-Entropy has been deployed to train the optimized VIT model using the Adam algorithm.</p><p><strong>Results: </strong>The classification accuracy of the proposed model is 90% for detecting the severity with an F1-score of 0.87 and MAE of 0.30. The research ascertained that the model ViT evinced a superlative efficacy for the capturing of fine-grained spatial relations in ultrasound images to produce revolutionary predictions.</p><p><strong>Conclusion: </strong>These results emphasize that ViTs have the potential to revolutionize fetal health monitoring and will contribute significantly to reducing neonatal mortality by supplying clinicians with accurate and reliable predictions for early interventions. This work stands as a yardstick for further diagnostic applications using AI in fetal health care.</p>","PeriodicalId":54215,"journal":{"name":"Current Medical Imaging Reviews","volume":" ","pages":""},"PeriodicalIF":1.1000,"publicationDate":"2025-03-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Current Medical Imaging Reviews","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.2174/0115734056360199250227053012","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0

Abstract

Aim: This research aims to develop and evaluate a novel health classification and severity detection system based on Vision Transformers (ViTs) for fetal ultrasound imagery. This contributes to improved precision in fetal health status detection and abnormalities with more accurate results than other traditional models.

Background: Amidst the other imperatives of resource-deficient developing nations, mitigating neonatal mortality rates is a challenge that demands precisionbased solutions in the era of artificial intelligence. Though the advent of machine learning models has added an optimal dimension to deal with emerging complexity in fetal ultrasound imagery, there is a call to address the huge gap in the demanded precision for prediction than the existing interpretation.

Objective: This research strives to formulate and access a novel health classification and severity detection system based on the implementation of the Vision Transformers frameworks. This pioneering investigation represents an unparalleled exploration into the efficacy of ViTs for discerning intricate patterns within fetal ultrasonographic imagery, facilitating precise categorization of fetal well-being and prognosticating the magnitude of potential anomalies.

Methodology: A private and confidential dataset of 500 fetal ultrasound images has been collected from diverse hospitals. Each image has been annotated by radiologists according to two main labels: the health status of the fetus, which includes healthy, mild, moderate, or severe, and the severity of abnormalities as a continuous measure. At different levels, the dataset underwent pre-processing via distinct techniques. Then, the composite loss function Cross-Entropy has been deployed to train the optimized VIT model using the Adam algorithm.

Results: The classification accuracy of the proposed model is 90% for detecting the severity with an F1-score of 0.87 and MAE of 0.30. The research ascertained that the model ViT evinced a superlative efficacy for the capturing of fine-grained spatial relations in ultrasound images to produce revolutionary predictions.

Conclusion: These results emphasize that ViTs have the potential to revolutionize fetal health monitoring and will contribute significantly to reducing neonatal mortality by supplying clinicians with accurate and reliable predictions for early interventions. This work stands as a yardstick for further diagnostic applications using AI in fetal health care.

胎儿诊断使用视觉转换器增强健康和严重程度的超声成像预测。
目的:开发并评价一种基于视觉变换的胎儿超声图像健康分类与严重程度检测系统。这有助于提高胎儿健康状况检测和异常的精度,结果比其他传统模型更准确。背景:在资源匮乏的发展中国家的其他当务之急中,降低新生儿死亡率是一项挑战,需要人工智能时代基于精确的解决方案。尽管机器学习模型的出现为处理胎儿超声图像中出现的复杂性增加了一个最佳维度,但有人呼吁解决预测精度要求与现有解释之间的巨大差距。目的:基于视觉变形框架的健康分类与严重程度检测系统的构建与实现。这项开创性的研究代表了对ViTs在胎儿超声图像中识别复杂模式的功效的无与伦比的探索,促进了胎儿健康的精确分类和预测潜在异常的程度。方法:从不同的医院收集了500个胎儿超声图像的私人和机密数据集。每张图像都由放射科医生根据两个主要标签进行注释:胎儿的健康状况,包括健康、轻度、中度或严重,以及作为连续测量的异常严重程度。在不同的层次上,数据集通过不同的技术进行预处理。然后,利用复合损失函数Cross-Entropy利用Adam算法对优化后的VIT模型进行训练。结果:所提模型对严重程度的分类准确率为90%,f1得分为0.87,MAE为0.30。该研究确定了模型ViT在超声图像中捕获细粒度空间关系以产生革命性预测方面具有最高功效。结论:这些结果强调了ViTs有可能彻底改变胎儿健康监测,并将为临床医生提供准确可靠的早期干预预测,从而显著降低新生儿死亡率。这项工作为人工智能在胎儿保健中的进一步诊断应用提供了一个标准。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
2.60
自引率
0.00%
发文量
246
审稿时长
1 months
期刊介绍: Current Medical Imaging Reviews publishes frontier review articles, original research articles, drug clinical trial studies and guest edited thematic issues on all the latest advances on medical imaging dedicated to clinical research. All relevant areas are covered by the journal, including advances in the diagnosis, instrumentation and therapeutic applications related to all modern medical imaging techniques. The journal is essential reading for all clinicians and researchers involved in medical imaging and diagnosis.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信