Detecting Vasodilation as Potential Diagnostic Biomarker in Breast Cancer Using Deep Learning-Driven Thermomics

Yousefi, Bardia; Akbari, Hamed; Maldague, Xavier

doi:10.3390/bios10110164

Cited by 22 publications

(30 citation statements)

References 79 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep radiomics is known for its tendency of having high dimensionality, thereby intensifying the possibility of overfitting a decision-making unit (i.e., the random forest model in this study) and the curse of dimensionality problem. The proposed ConvAE provided low-dimensional deep radiomics by spanning the imaging features to a lower-dimensional space [ 54 ]. We applied dimensionality reduction for the conventional radiomics by following the traditional way to shrink features while preserving the image characteristics.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Impartially Validated Multiple Deep-Chain Models to Detect COVID-19 in Chest X-ray Using Latent Space Radiomics

Yousefi

Kawakita

Amini

et al. 2021

JCM

Self Cite

View full text Add to dashboard Cite

The COVID-19 pandemic continues to spread globally at a rapid pace, and its rapid detection remains a challenge due to its rapid infectivity and limited testing availability. One of the simply available imaging modalities in clinical routine involves chest X-ray (CXR), which is often used for diagnostic purposes. Here, we proposed a computer-aided detection of COVID-19 in CXR imaging using deep and conventional radiomic features. First, we used a 2D U-Net model to segment the lung lobes. Then, we extracted deep latent space radiomics by applying deep convolutional autoencoder (ConvAE) with internal dense layers to extract low-dimensional deep radiomics. We used Johnson–Lindenstrauss (JL) lemma, Laplacian scoring (LS), and principal component analysis (PCA) to reduce dimensionality in conventional radiomics. The generated low-dimensional deep and conventional radiomics were integrated to classify COVID-19 from pneumonia and healthy patients. We used 704 CXR images for training the entire model (i.e., U-Net, ConvAE, and feature selection in conventional radiomics). Afterward, we independently validated the whole system using a study cohort of 1597 cases. We trained and tested a random forest model for detecting COVID-19 cases through multivariate binary-class and multiclass classification. The maximal (full multivariate) model using a combination of the two radiomic groups yields performance in classification cross-validated accuracy of 72.6% (69.4–74.4%) for multiclass and 89.6% (88.4–90.7%) for binary-class classification.

show abstract

Section: Resultsmentioning

confidence: 99%

“…It lowers the accuracy while showing pseudo improvement in the overall accuracy of the system due to the increased collinearity among the features. To alleviate this problem, one potential solution is to use deep-learning feature selection for deep radiomics [ 54 ].…”

Section: Methodsmentioning

confidence: 99%

Impartially Validated Multiple Deep-Chain Models to Detect COVID-19 in Chest X-ray Using Latent Space Radiomics

Yousefi

Kawakita

Amini

et al. 2021

JCM

Self Cite

View full text Add to dashboard Cite

show abstract

“…The authors 8,9 in previous work demonstrated use of equivalent wave field transform (EWFT) applied to dynamic data to detect increased perfusion associated with the tumor.…”

Section: Breast Thermographymentioning

confidence: 99%

Application of components analysis in dynamic thermal breast imaging to identify pathophysiologic mechanisms of heat transfer

Gershenson

2021

Preprint

View full text Add to dashboard Cite

Significance: Of the most compelling unsolved issues in the paradigm to create success in the field of breast cancer infrared imaging is localization of direct internal heat of the tumor. The contribution of differential heat production related to metabolism versus perfusion is not understood. Previous work until now has not shown progress beyond identifying veins which are fed by the hot cancer. Employing signal analysis techniques, we probe important questions which may lead to further understanding pathophysiology of heat transfer occurring in the setting of malignancy. Aim: When using thermal imaging to detect breast cancer, the dominant heat signature is that of indirect heat transported in gradient away from the tumor location. Unprocessed images strikingly display vasculature which acts to direct excess heat superficially towards the skin surface before dissipating. In current clinical use, interpretation of thermogram images considers abnormal vascular patterns and overall temperature as indicators of disease. The goal of this work is to present a processing method for dynamic external stimulus thermogram images to isolate and separate the indirect vascular heat while revealing the desired direct heat from the tumor. Approach: In dynamic thermal imaging of the breast, a timed series of images are taken following application of external temperature stimulus (most often cooled air). While the tumor heat response is thought to be independent of the external stimulus, the secondary heat of the veins is known to be affected by vasomodulation. The recorded data is analyzed using independent component analysis (ICA) and principal component analysis (PCA) methods. ICA separates the image sequence into new independent images having a common characteristic time behavior. Resulting individual components are analyzed for correspondence to the presence or lack of vasomodulation. Results: Using the Brazilian visual lab mastology data set containing dynamic thermograms, applying components analysis resulted in three corresponding images: 1. Minimum change as a function of applied temperature or time (suggests correlation with the cancer generated heat), 2. Moderate temperature dependence (suggests correlation with veins affected by vasomodulation) and 3. Complex time behavior (suggests correlation with heat absorption due to high tumor perfusion). All components appear clear and distinct. Conclusions: Applying signal processing methods to the dynamic infrared data, we found three distinct components with correspondence to understood physiologic processes. The two cases shown are self-evident of the capability of the method but are lacking supporting ground truth that is unavailable with such a limited data set. Validation of this proposed paradigm and studying furthering clinical applications has potential to create significant achievement for IR modality in diagnostic imaging.

show abstract

“…In [24], a new method for detecting the immunohistochemical response of breast cancer was proposed. The proposed method works by finding heterogeneous thermal patterns in the target area.…”

Section: Related Workmentioning

confidence: 99%

Classification of Breast Cancer Using Neutrosophic Techniques and Deep Neural Network

Shaban¹

2021

Preprint

View full text Add to dashboard Cite

Breast cancer is one of the most common types of cancer that affects women globally and it is the primary cause of death. Early detection of breast cancer is a vital process that can facilitate appropriate treatment, stop the progression of cancer cells, and reduce morbidity and mortality. Artificial Intelligence (AI) and Machine Learning (ML) are the most popular methods that can be used to detect and classify breast cancer accurately. In this paper, a new strategy for classifying breast cancer using Neutrosophic Techniques (NTs) and machine learning techniques is introduced, which is called Breast Cancer Classification Strategy (BC2S). The proposed BC2S consists of two phases, which are; Data Preprocessing Phase (DP2) and Breast Cancer Classification Phase (BC2P). The main aim of the data preprocessing phase is to; (i) extract features from mammogram images and then remove the outlier items, (ii) select the most effective and informative features from those extracted features using new feature selection method called Efficient Ant Colony Optimization (EACO), and (iii) convert the selected features from classical domain into neutrosophic domain using NTs to give accurate classification through the next classification phase BC2P. The proposed classification model uses Deep Neural Network (DNN) to determine whether the patient is normal or infected with benign or malignant cancer. According to experimental results, the proposed strategy outperforms other competitors in terms of accuracy, precision, recall, and F-measure.

show abstract

Detecting Vasodilation as Potential Diagnostic Biomarker in Breast Cancer Using Deep Learning-Driven Thermomics

Cited by 22 publications

References 79 publications

Impartially Validated Multiple Deep-Chain Models to Detect COVID-19 in Chest X-ray Using Latent Space Radiomics

Impartially Validated Multiple Deep-Chain Models to Detect COVID-19 in Chest X-ray Using Latent Space Radiomics

Application of components analysis in dynamic thermal breast imaging to identify pathophysiologic mechanisms of heat transfer

Classification of Breast Cancer Using Neutrosophic Techniques and Deep Neural Network

Contact Info

Product

Resources

About