Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification

BackgroundPhotoacoustic spectral analysis has been demonstrated to be efficacious in the diagnosis of prostate cancer (PCa). With the incorporation of deep learning, its discrimination accuracy is progressively enhancing. Nevertheless, individual heterogeneity persists as a significant factor that i...

Full description

Saved in:
Bibliographic Details
Main Authors: Yingna Chen, Feifan Li, Zhuoheng Dai, Ying Liu, Shengsong Huang, Qian Cheng
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-07-01
Series:Frontiers in Oncology
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fonc.2025.1592815/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850116507375239168
author Yingna Chen
Yingna Chen
Feifan Li
Zhuoheng Dai
Ying Liu
Shengsong Huang
Qian Cheng
Qian Cheng
author_facet Yingna Chen
Yingna Chen
Feifan Li
Zhuoheng Dai
Ying Liu
Shengsong Huang
Qian Cheng
Qian Cheng
author_sort Yingna Chen
collection DOAJ
description BackgroundPhotoacoustic spectral analysis has been demonstrated to be efficacious in the diagnosis of prostate cancer (PCa). With the incorporation of deep learning, its discrimination accuracy is progressively enhancing. Nevertheless, individual heterogeneity persists as a significant factor that impacts discrimination performance.ObjectiveExtracting more reliable features from intricate biological tissue and augmenting discrimination accuracy of the prostate cancer.MethodsSupervised contrastive learning is introduced to explore its performance in photoacoustic spectral feature extraction. Three distinct models, namely the CNN-based model, the supervised contrastive (SC) model, and the supervised contrastive loss adjust (SCL-adjust) model, have been compared, along with traditional feature extraction and machine learning-based methods.ResultsThe outcomes have indicated that the SCL-adjust model exhibits the optimal performance, its accuracy rate has increased by more than 10% compared with the traditional method. Besides, the features extracted from this model are more resilient, regardless of the presence of uniform or Gaussian noise and model transfer. Compared with CNN model, the transfer performance of the proposed model has improved by approximately 5%.ConclusionsSupervised contrast learning is integrated into photoacoustic spectrum analysis and its effectiveness is verified. A comprehensive analysis is conducted on the performance improvement of the proposed SCL-adjust model in photoacoustic prostate cancer diagnosis, its resistance to noise, and its adaptability to the data heterogeneity of different systems.
format Article
id doaj-art-8894c62d43a84b41b3f7cd2d24689f31
institution OA Journals
issn 2234-943X
language English
publishDate 2025-07-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Oncology
spelling doaj-art-8894c62d43a84b41b3f7cd2d24689f312025-08-20T02:36:19ZengFrontiers Media S.A.Frontiers in Oncology2234-943X2025-07-011510.3389/fonc.2025.15928151592815Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identificationYingna Chen0Yingna Chen1Feifan Li2Zhuoheng Dai3Ying Liu4Shengsong Huang5Qian Cheng6Qian Cheng7School of Information Engineering, College of Science & Technology Ningbo University, Ningbo, Zhejiang, ChinaSchool of Physics Science and Engineering, Tongji University, Shanghai, ChinaSchool of Information Engineering, College of Science & Technology Ningbo University, Ningbo, Zhejiang, ChinaSchool of Information Engineering, College of Science & Technology Ningbo University, Ningbo, Zhejiang, ChinaDepartment of Urology, Tongji Hospital, Tongji University School of Medicine, Shanghai, ChinaDepartment of Urology, Tongji Hospital, Tongji University School of Medicine, Shanghai, ChinaSchool of Physics Science and Engineering, Tongji University, Shanghai, ChinaThe National Key Laboratory of Autonomous Intelligent Unmanned Systems, China, The Frontiers Science Center for Intelligent Autonomous Systems, Ministry of Education, Shanghai, ChinaBackgroundPhotoacoustic spectral analysis has been demonstrated to be efficacious in the diagnosis of prostate cancer (PCa). With the incorporation of deep learning, its discrimination accuracy is progressively enhancing. Nevertheless, individual heterogeneity persists as a significant factor that impacts discrimination performance.ObjectiveExtracting more reliable features from intricate biological tissue and augmenting discrimination accuracy of the prostate cancer.MethodsSupervised contrastive learning is introduced to explore its performance in photoacoustic spectral feature extraction. Three distinct models, namely the CNN-based model, the supervised contrastive (SC) model, and the supervised contrastive loss adjust (SCL-adjust) model, have been compared, along with traditional feature extraction and machine learning-based methods.ResultsThe outcomes have indicated that the SCL-adjust model exhibits the optimal performance, its accuracy rate has increased by more than 10% compared with the traditional method. Besides, the features extracted from this model are more resilient, regardless of the presence of uniform or Gaussian noise and model transfer. Compared with CNN model, the transfer performance of the proposed model has improved by approximately 5%.ConclusionsSupervised contrast learning is integrated into photoacoustic spectrum analysis and its effectiveness is verified. A comprehensive analysis is conducted on the performance improvement of the proposed SCL-adjust model in photoacoustic prostate cancer diagnosis, its resistance to noise, and its adaptability to the data heterogeneity of different systems.https://www.frontiersin.org/articles/10.3389/fonc.2025.1592815/fullsupervised contrastive learningphotoacoustic spectral analysisprostate cancerrobust featureCNN
spellingShingle Yingna Chen
Yingna Chen
Feifan Li
Zhuoheng Dai
Ying Liu
Shengsong Huang
Qian Cheng
Qian Cheng
Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
Frontiers in Oncology
supervised contrastive learning
photoacoustic spectral analysis
prostate cancer
robust feature
CNN
title Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
title_full Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
title_fullStr Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
title_full_unstemmed Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
title_short Supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
title_sort supervised contrastive loss helps uncover more robust features for photoacoustic prostate cancer identification
topic supervised contrastive learning
photoacoustic spectral analysis
prostate cancer
robust feature
CNN
url https://www.frontiersin.org/articles/10.3389/fonc.2025.1592815/full
work_keys_str_mv AT yingnachen supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT yingnachen supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT feifanli supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT zhuohengdai supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT yingliu supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT shengsonghuang supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT qiancheng supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification
AT qiancheng supervisedcontrastivelosshelpsuncovermorerobustfeaturesforphotoacousticprostatecanceridentification