A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation

Abstract An effective diagnosis system and suitable treatment planning require the precise segmentation of thyroid nodules in ultrasound imaging. The advancement of imaging technologies has not resolved traditional imaging challenges, which include noise issues, limited contrast, and dependency on o...

Full description

Saved in:
Bibliographic Details
Main Authors: Tathagat Banerjee, Davinder Paul Singh, Debabrata Swain, Shubham Mahajan, Seifedine Kadry, Jungeun Kim
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-12602-6
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849332841161162752
author Tathagat Banerjee
Davinder Paul Singh
Debabrata Swain
Shubham Mahajan
Seifedine Kadry
Jungeun Kim
author_facet Tathagat Banerjee
Davinder Paul Singh
Debabrata Swain
Shubham Mahajan
Seifedine Kadry
Jungeun Kim
author_sort Tathagat Banerjee
collection DOAJ
description Abstract An effective diagnosis system and suitable treatment planning require the precise segmentation of thyroid nodules in ultrasound imaging. The advancement of imaging technologies has not resolved traditional imaging challenges, which include noise issues, limited contrast, and dependency on operator choices, thus highlighting the need for automated, reliable solutions. The researchers developed TATHA, an innovative deep learning architecture dedicated to improving thyroid ultrasound image segmentation accuracy. The model is evaluated using the digital database of thyroid ultrasound images, which includes 99 cases across three subsets containing 134 labelled images for training, validation, and testing. It incorporates data pre-treatment procedures that reduce speckle noise and enhance contrast, while edge detection provides high-quality input for segmentation. TATHA outperforms U-Net, PSPNet, and Vision Transformers across various datasets and cross-validation folds, achieving superior Dice scores, accuracy, and AUC results. The distributed thyroid segmentation framework generates reliable predictions by combining results from multiple feature extraction units. The findings confirm that these advancements make TATHA an essential tool for clinicians and researchers in thyroid imaging and clinical applications.
format Article
id doaj-art-c395d13a4b2b4ad9b379d4d601ec2907
institution Kabale University
issn 2045-2322
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-c395d13a4b2b4ad9b379d4d601ec29072025-08-20T03:46:04ZengNature PortfolioScientific Reports2045-23222025-07-0115112110.1038/s41598-025-12602-6A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentationTathagat Banerjee0Davinder Paul Singh1Debabrata Swain2Shubham Mahajan3Seifedine Kadry4Jungeun Kim5Department of Computer Science and Engineering, IIT PatnaDepartment of Computer Science and Engineering, Pandit Deendayal Energy UniversityDepartment of Computer Science and Engineering, Pandit Deendayal Energy UniversityAmity School of Engineering and Technology, Amity University HaryanaDepartment of Computer Science and Mathematics, Lebanese American UniversityDepartment of Computer Engineering, Inha UniversityAbstract An effective diagnosis system and suitable treatment planning require the precise segmentation of thyroid nodules in ultrasound imaging. The advancement of imaging technologies has not resolved traditional imaging challenges, which include noise issues, limited contrast, and dependency on operator choices, thus highlighting the need for automated, reliable solutions. The researchers developed TATHA, an innovative deep learning architecture dedicated to improving thyroid ultrasound image segmentation accuracy. The model is evaluated using the digital database of thyroid ultrasound images, which includes 99 cases across three subsets containing 134 labelled images for training, validation, and testing. It incorporates data pre-treatment procedures that reduce speckle noise and enhance contrast, while edge detection provides high-quality input for segmentation. TATHA outperforms U-Net, PSPNet, and Vision Transformers across various datasets and cross-validation folds, achieving superior Dice scores, accuracy, and AUC results. The distributed thyroid segmentation framework generates reliable predictions by combining results from multiple feature extraction units. The findings confirm that these advancements make TATHA an essential tool for clinicians and researchers in thyroid imaging and clinical applications.https://doi.org/10.1038/s41598-025-12602-6Medical imagingTATHASegmentationNeural networksUltrasound imaging
spellingShingle Tathagat Banerjee
Davinder Paul Singh
Debabrata Swain
Shubham Mahajan
Seifedine Kadry
Jungeun Kim
A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
Scientific Reports
Medical imaging
TATHA
Segmentation
Neural networks
Ultrasound imaging
title A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
title_full A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
title_fullStr A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
title_full_unstemmed A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
title_short A novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
title_sort novel hybrid deep learning approach combining deep feature attention and statistical validation for enhanced thyroid ultrasound segmentation
topic Medical imaging
TATHA
Segmentation
Neural networks
Ultrasound imaging
url https://doi.org/10.1038/s41598-025-12602-6
work_keys_str_mv AT tathagatbanerjee anovelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT davinderpaulsingh anovelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT debabrataswain anovelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT shubhammahajan anovelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT seifedinekadry anovelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT jungeunkim anovelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT tathagatbanerjee novelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT davinderpaulsingh novelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT debabrataswain novelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT shubhammahajan novelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT seifedinekadry novelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation
AT jungeunkim novelhybriddeeplearningapproachcombiningdeepfeatureattentionandstatisticalvalidationforenhancedthyroidultrasoundsegmentation