Transfer Learning with Feature Extraction Modules for Improved Classifier Performance on Medical Image Data

Jha, Ritesh; Bhattacherjee, Vandana; Mustafi, Abhijit

doi:10.1155/2022/4983174

Cited by 4 publications

(5 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The goal is to combine the transfer learning technique with the FIN to generate imitated feature maps from the significant features revealed in this study to improve the predictive power of the adaptive models. As future work, we believe that if the K-SOM is combined with the Transfer Learning with Feature Extraction Modules Network (TrFEMNet) 156 , Feature representations from General Feature Extraction Module (GFEM) 156 , and Specific Feature Extraction Module (SFEM) 156 , it would generate more stable and in-depth results for further modeling. The major role for the transfer learning will come into play when the significant radiomics features/biomarkers discovered by this study (in form of K-SOM topology or feature space) are used as a learned representation for physiological knowledge transfer into the clinical setting that uses the same imaging concept with a slightly different DCE-MRI pulse sequence (SPGRE).…”

Section: Discussionmentioning

confidence: 99%

Radiomics characterization of tissues in an animal brain tumor model imaged using dynamic contrast enhanced (DCE) MRI

Bagher‐Ebadian

Brown

Ghassemi

et al. 2023

Sci Rep

View full text Add to dashboard Cite

Here, we investigate radiomics-based characterization of tumor vascular and microenvironmental properties in an orthotopic rat brain tumor model measured using dynamic-contrast-enhanced (DCE) MRI. Thirty-two immune compromised-RNU rats implanted with human U-251N cancer cells were imaged using DCE-MRI (7Tesla, Dual-Gradient-Echo). The aim was to perform pharmacokinetic analysis using a nested model (NM) selection technique to classify brain regions according to vasculature properties considered as the source of truth. A two-dimensional convolutional-based radiomics analysis was performed on the raw-DCE-MRI of the rat brains to generate dynamic radiomics maps. The raw-DCE-MRI and respective radiomics maps were used to build 28 unsupervised Kohonen self-organizing-maps (K-SOMs). A Silhouette-Coefficient (SC), k-fold Nested-Cross-Validation (k-fold-NCV), and feature engineering analyses were performed on the K-SOMs’ feature spaces to quantify the distinction power of radiomics features compared to raw-DCE-MRI for classification of different Nested Models. Results showed that eight radiomics features outperformed respective raw-DCE-MRI in prediction of the three nested models. The average percent difference in SCs between radiomics features and raw-DCE-MRI was: 29.875% ± 12.922%, p < 0.001. This work establishes an important first step toward spatiotemporal characterization of brain regions using radiomics signatures, which is fundamental toward staging of tumors and evaluation of tumor response to different treatments.

show abstract

Section: Discussionmentioning

confidence: 99%

Radiomics characterization of tissues in an animal brain tumor model imaged using dynamic contrast enhanced (DCE) MRI

Bagher‐Ebadian

Brown

Ghassemi

et al. 2023

Sci Rep

View full text Add to dashboard Cite

show abstract

“…This research aimed to combine respiratory sound, CNN classification, and time-series feature extraction from pre-trained images. We also assessed the performance of several feature extractors [34]. The LSTM model is utilized to obtain the findings after the MFCC model has been used for feature extraction [35].…”

Section: Related Work Historymentioning

confidence: 99%

Transfer Learning based Diagnosis and Analysis of Lung Sound Aberrations

Gulzar¹,

Li²,

Manzoor³

et al. 2023

IJBB

View full text Add to dashboard Cite

With the development of computer -systems that can collect and analyze enormous volumes of data, the medical profession is establishing several non-invasive tools. This work attempts to develop a non-invasive technique for identifying respiratory sounds acquired by a stethoscope and voice recording software via machine learning techniques. This study suggests a trained and proven CNN-based approach for categorizing respiratory sounds. A visual representation of each audio sample is constructed, allowing resource identification for classification using methods like those used to effectively describe visuals. We used a technique called MelFrequency Cepstral Coefficients (MFCCs). Here, features are retrieved and categorized via VGG16 (transfer learning) and prediction is accomplished using 5-fold cross-validation. Employing various data splitting techniques, Respiratory Sound Database obtained cutting- edge results, including accuracy of 95%, precision of 88%, recall score of 86%, and F1 score of 81%. The ICBHI dataset is used to train and test the model.

show abstract

“…Chunfeng Lian et al, proposed an attention-guided deep learning framework for dementia diagnosis using structural MRI. This framework combines a fully convolutional network (FCN) and a multi-branch hybrid network (HybNet) to jointly learn and fuse sMRI features for CAD model development [6] proposed the TrFEMNet model for medical image classification, showing comparable performance to other models, especially on complex datasets [7]. Mohammad Monirujjaman Khan et al, explored the application of convolutional neural networks (CNNs) for diagnosing brain cancers using medical images.…”

Section: Literature Reviewmentioning

confidence: 99%

“…For multiclass classification the score of each class is calculated and then aggregated as defined in Eq. (7). to assess the overall performance of the model.…”

Section: F1-scorementioning

confidence: 99%

See 1 more Smart Citation

Deep Learning Techniques in Classification of Stages in Dementia: An Ensemble Approach

Pandiyarajan Mariappan,

R. S. Valarmathi

2023

ARASET

View full text Add to dashboard Cite

Dementia is the severe brain disease that affects the memory system of the humans which requires early diagnosis. It progresses through different stages from very mild dementia, which motivates to find the stage of dementia for treating the affected persons. A novel ensemble-based classification model to classify the stages in dementia such as mild-dementia, non-dementia, very-mild dementia has been proposed in this work. The MRI images were pre-processed with techniques such as resizing, data augmentation and Contrast Limited Adaptive Histogram Equalization (CLAHE) and different four forms of datasets such as Image Data-1, Image Data-2, Image Data-3, Image Data-4 are framed to justify preprocessing techniques. The pre-trained models namely ResNet, Inception, MobileNet, VGG, DenseNet, Xception, Inception Resnet, AlexNet, EfficientNet and NasNet Mobile are trained with all four forms of dataset and performance of each model is analysed. The models DenseNet201, MobileNetV2, VGG19, ResNet152 achieved accuracy of 93%,93%,93%.90% respectively with Image Data-4. The proposed model is the ensemble of DenseNet201, MobileNetV2, VGG19, ResNet152 which provides accuracy precision, recall and f1-score of 95%, 93%, 100% and 96% respectively. This ensemble model has undergone testing using a brain tumour dataset, and the experimental findings indicate that the proposed framework enhances the classifier's generalization capabilities for multiclass problems.

show abstract

Transfer Learning with Feature Extraction Modules for Improved Classifier Performance on Medical Image Data

Cited by 4 publications

References 34 publications

Radiomics characterization of tissues in an animal brain tumor model imaged using dynamic contrast enhanced (DCE) MRI

Radiomics characterization of tissues in an animal brain tumor model imaged using dynamic contrast enhanced (DCE) MRI

Transfer Learning based Diagnosis and Analysis of Lung Sound Aberrations

Deep Learning Techniques in Classification of Stages in Dementia: An Ensemble Approach

Contact Info

Product

Resources

About