Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer

Joanna Loja; Armando Mendez; Kuan Huang

doi:10.1109/URTC60662.2023.10534928

Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer

Joanna Loja, Armando Mendez, Kuan Huang

Computer Information Systems

Kean University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Breast ultrasound (BUS) imaging offers a non-invasive and radiation-free method to examine breast tissues. Automated BUS image segmentation and classification can help doctors identify lesions and possible abnormalities early, enabling healthcare professionals to detect breast cancer or other conditions in time for early intervention. In this research, we first conduct a comprehensive performance comparison between transformer networks and convolutional networks; secondly, we propose a novel approach by merging segmentation and classification networks, creating a multitask network tailored explicitly for BUS image segmentation and classification; thirdly, we thoroughly investigate network performance and refine training parameters to prevent overfitting. Finally, we create a user-friendly GUI demo showing our classification and segmentation results. The results demonstrate that the ResNet-50 Multi-Task model exhibits the best overall performance for both segmentation and classification tasks.

Original language	English
Title of host publication	IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9798350308600
DOIs	https://doi.org/10.1109/URTC60662.2023.10534928
State	Published - 2023
Event	2023 IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Hybrid, Cambridge, United States Duration: 6 Oct 2023 → 8 Oct 2023

Publication series

Name	IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings

Conference

Conference	2023 IEEE MIT Undergraduate Research Technology Conference, URTC 2023
Country/Territory	United States
City	Hybrid, Cambridge
Period	6/10/23 → 8/10/23

Keywords

breast ultrasound imaging
deep learning
image classification
image segmentation
multitask neural network

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/URTC60662.2023.10534928

Cite this

Loja, J., Mendez, A., & Huang, K. (2023). Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer. In IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings (IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/URTC60662.2023.10534928

Loja, Joanna ; Mendez, Armando ; Huang, Kuan. / Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer. IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. (IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings).

@inproceedings{a7e565bbf3c043dc9953136b48b3e8dd,

title = "Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer",

abstract = "Breast ultrasound (BUS) imaging offers a non-invasive and radiation-free method to examine breast tissues. Automated BUS image segmentation and classification can help doctors identify lesions and possible abnormalities early, enabling healthcare professionals to detect breast cancer or other conditions in time for early intervention. In this research, we first conduct a comprehensive performance comparison between transformer networks and convolutional networks; secondly, we propose a novel approach by merging segmentation and classification networks, creating a multitask network tailored explicitly for BUS image segmentation and classification; thirdly, we thoroughly investigate network performance and refine training parameters to prevent overfitting. Finally, we create a user-friendly GUI demo showing our classification and segmentation results. The results demonstrate that the ResNet-50 Multi-Task model exhibits the best overall performance for both segmentation and classification tasks.",

keywords = "breast ultrasound imaging, deep learning, image classification, image segmentation, multitask neural network",

author = "Joanna Loja and Armando Mendez and Kuan Huang",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE MIT Undergraduate Research Technology Conference, URTC 2023 ; Conference date: 06-10-2023 Through 08-10-2023",

year = "2023",

doi = "10.1109/URTC60662.2023.10534928",

language = "English",

series = "IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings",

}

Loja, J, Mendez, A & Huang, K 2023, Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer. in IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings. IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings, Institute of Electrical and Electronics Engineers Inc., 2023 IEEE MIT Undergraduate Research Technology Conference, URTC 2023, Hybrid, Cambridge, United States, 6/10/23. https://doi.org/10.1109/URTC60662.2023.10534928

Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer. / Loja, Joanna; Mendez, Armando; Huang, Kuan.
IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. (IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer

AU - Loja, Joanna

AU - Mendez, Armando

AU - Huang, Kuan

PY - 2023

Y1 - 2023

N2 - Breast ultrasound (BUS) imaging offers a non-invasive and radiation-free method to examine breast tissues. Automated BUS image segmentation and classification can help doctors identify lesions and possible abnormalities early, enabling healthcare professionals to detect breast cancer or other conditions in time for early intervention. In this research, we first conduct a comprehensive performance comparison between transformer networks and convolutional networks; secondly, we propose a novel approach by merging segmentation and classification networks, creating a multitask network tailored explicitly for BUS image segmentation and classification; thirdly, we thoroughly investigate network performance and refine training parameters to prevent overfitting. Finally, we create a user-friendly GUI demo showing our classification and segmentation results. The results demonstrate that the ResNet-50 Multi-Task model exhibits the best overall performance for both segmentation and classification tasks.

AB - Breast ultrasound (BUS) imaging offers a non-invasive and radiation-free method to examine breast tissues. Automated BUS image segmentation and classification can help doctors identify lesions and possible abnormalities early, enabling healthcare professionals to detect breast cancer or other conditions in time for early intervention. In this research, we first conduct a comprehensive performance comparison between transformer networks and convolutional networks; secondly, we propose a novel approach by merging segmentation and classification networks, creating a multitask network tailored explicitly for BUS image segmentation and classification; thirdly, we thoroughly investigate network performance and refine training parameters to prevent overfitting. Finally, we create a user-friendly GUI demo showing our classification and segmentation results. The results demonstrate that the ResNet-50 Multi-Task model exhibits the best overall performance for both segmentation and classification tasks.

KW - breast ultrasound imaging

KW - deep learning

KW - image classification

KW - image segmentation

KW - multitask neural network

UR - http://www.scopus.com/inward/record.url?scp=85195446685&partnerID=8YFLogxK

U2 - 10.1109/URTC60662.2023.10534928

DO - 10.1109/URTC60662.2023.10534928

M3 - Conference contribution

AN - SCOPUS:85195446685

T3 - IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings

BT - IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2023 IEEE MIT Undergraduate Research Technology Conference, URTC 2023

Y2 - 6 October 2023 through 8 October 2023

ER -

Loja J, Mendez A, Huang K. Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer. In IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2023. (IEEE MIT Undergraduate Research Technology Conference, URTC 2023 - Proceedings). doi: 10.1109/URTC60662.2023.10534928

Multi-Task Breast Ultrasound Image Segmentation and Classification Using Convolutional Neural Network and Transformer

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this