Contrastive Clustering-Based Patient Normalization to Improve Automated In Vivo Oral Cancer Diagnosis from Multispectral Autofluorescence Lifetime Images.
Kayla Caughlin, Elvis Duran-Sierra, Shuna Cheng, Rodrigo Cuenca, Beena Ahmed, Jim Ji, Mathias Martinez, Moustafa Al-Khalil, Hussain Al-Enazi, Javier A Jo, Carlos Busso
{"title":"Contrastive Clustering-Based Patient Normalization to Improve Automated In Vivo Oral Cancer Diagnosis from Multispectral Autofluorescence Lifetime Images.","authors":"Kayla Caughlin, Elvis Duran-Sierra, Shuna Cheng, Rodrigo Cuenca, Beena Ahmed, Jim Ji, Mathias Martinez, Moustafa Al-Khalil, Hussain Al-Enazi, Javier A Jo, Carlos Busso","doi":"10.3390/cancers16234120","DOIUrl":null,"url":null,"abstract":"<p><p><b>Background:</b> Multispectral autofluorescence lifetime imaging systems have recently been developed to quickly and non-invasively assess tissue properties for applications in oral cancer diagnosis. As a non-traditional imaging modality, the autofluorescence signal collected from the system cannot be directly visually assessed by a clinician and a model is needed to generate a diagnosis for each image. However, training a deep learning model from scratch on small multispectral autofluorescence datasets can fail due to inter-patient variability, poor initialization, and overfitting. <b>Methods:</b> We propose a contrastive-based pre-training approach that teaches the network to perform patient normalization without requiring a direct comparison to a reference sample. We then use the contrastive pre-trained encoder as a favorable initialization for classification. To train the classifiers, we efficiently use available data and reduce overfitting through a multitask framework with margin delineation and cancer diagnosis tasks. We evaluate the model over 67 patients using 10-fold cross-validation and evaluate significance using paired, one-tailed <i>t</i>-tests. <b>Results:</b> The proposed approach achieves a sensitivity of 82.08% and specificity of 75.92% on the cancer diagnosis task with a sensitivity of 91.83% and specificity of 79.31% for margin delineation as an auxiliary task. In comparison to existing approaches, our method significantly outperforms a <i>support vector machine</i> (SVM) implemented with either <i>sequential feature selection</i> (SFS) (<i>p</i> = 0.0261) or L1 loss (<i>p</i> = 0.0452) when considering the average of sensitivity and specificity. Specifically, the proposed approach increases performance by 2.75% compared to the L1 model and 4.87% compared to the SFS model. In addition, there is a significant increase in specificity of 8.34% compared to the baseline autoencoder model (<i>p</i> = 0.0070). <b>Conclusions:</b> Our method effectively trains deep learning models for small data applications when existing, large pre-trained models are not suitable for fine-tuning. While we designed the network for a specific imaging modality, we report the development process so that the insights gained can be applied to address similar challenges in other non-traditional imaging modalities. A key contribution of this paper is a neural network framework for multi-spectral fluorescence lifetime-based tissue discrimination that performs patient normalization without requiring a reference (healthy) sample from each patient at test time.</p>","PeriodicalId":9681,"journal":{"name":"Cancers","volume":"16 23","pages":""},"PeriodicalIF":4.5000,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Cancers","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.3390/cancers16234120","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ONCOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Background: Multispectral autofluorescence lifetime imaging systems have recently been developed to quickly and non-invasively assess tissue properties for applications in oral cancer diagnosis. As a non-traditional imaging modality, the autofluorescence signal collected from the system cannot be directly visually assessed by a clinician and a model is needed to generate a diagnosis for each image. However, training a deep learning model from scratch on small multispectral autofluorescence datasets can fail due to inter-patient variability, poor initialization, and overfitting. Methods: We propose a contrastive-based pre-training approach that teaches the network to perform patient normalization without requiring a direct comparison to a reference sample. We then use the contrastive pre-trained encoder as a favorable initialization for classification. To train the classifiers, we efficiently use available data and reduce overfitting through a multitask framework with margin delineation and cancer diagnosis tasks. We evaluate the model over 67 patients using 10-fold cross-validation and evaluate significance using paired, one-tailed t-tests. Results: The proposed approach achieves a sensitivity of 82.08% and specificity of 75.92% on the cancer diagnosis task with a sensitivity of 91.83% and specificity of 79.31% for margin delineation as an auxiliary task. In comparison to existing approaches, our method significantly outperforms a support vector machine (SVM) implemented with either sequential feature selection (SFS) (p = 0.0261) or L1 loss (p = 0.0452) when considering the average of sensitivity and specificity. Specifically, the proposed approach increases performance by 2.75% compared to the L1 model and 4.87% compared to the SFS model. In addition, there is a significant increase in specificity of 8.34% compared to the baseline autoencoder model (p = 0.0070). Conclusions: Our method effectively trains deep learning models for small data applications when existing, large pre-trained models are not suitable for fine-tuning. While we designed the network for a specific imaging modality, we report the development process so that the insights gained can be applied to address similar challenges in other non-traditional imaging modalities. A key contribution of this paper is a neural network framework for multi-spectral fluorescence lifetime-based tissue discrimination that performs patient normalization without requiring a reference (healthy) sample from each patient at test time.
期刊介绍:
Cancers (ISSN 2072-6694) is an international, peer-reviewed open access journal on oncology. It publishes reviews, regular research papers and short communications. Our aim is to encourage scientists to publish their experimental and theoretical results in as much detail as possible. There is no restriction on the length of the papers. The full experimental details must be provided so that the results can be reproduced.