Multichannel CNN for Facial Expression Recognition

Trivedi, Prapti; Mhasakar, Purva; Sujata, .; Mitra, Suman K.

doi:10.1007/978-3-030-34869-4_27

Cited by 1 publication

(6 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared to the single-input CNN model, MI-CNNs will provide more CNN features for the classification, which could improve the accuracy of the entire system [ 33 , 34 ]. However, most previous studies focused on whole images with different formats as the CNN inputs [ 33 , 34 ], and few studies attempted to segment a medical image into different regions as the CNN inputs and evaluate their contributions to the final classification of disease images. Therefore, our MI-CNNs could independently extract features from different regions; more importantly, the contribution of each region could be evaluated through the MI-CNN models.…”

Section: Discussionmentioning

confidence: 99%

“…In most of these previous studies, the CNN models were trained with whole-image CXRs as a single input for the classification [9,11,[13][14][15][16]18,[26][27][28][29][30][31]. Other studies also attempted to develop new CNN models that accept multiple inputs; such multiple-input CNNs (MI-CNN) could effectively improve the classification accuracy and demonstrated better performance than single-input CNNs [32][33][34]. Because an MI-CNN could provide different features, fusing these network features together could improve the accuracy of the entire system [34].…”

Section: Prior Workmentioning

confidence: 99%

“…Because an MI-CNN could provide different features, fusing these network features together could improve the accuracy of the entire system [ 34 ]. To date, MI-CNNs have been applied in the fields of facial expression and gender recognition [ 33 , 34 ] or flower grading [ 35 ].…”

Section: Introductionmentioning

confidence: 99%

“…However, most of the MI-CNN models developed in previous studies used whole images as at least one of the CNN inputs, and the prefeatured images were used as the other inputs [ 33 , 34 ]. To our best knowledge, few studies have reported using MI-CNNs to detect and analyze COVID-19 CXRs.…”

Section: Introductionmentioning

confidence: 99%

“…According to the latest WHO reports, by July 2021, 190,671,330 people have been confirmed to have COVID-19, contributing to 4,098,758 deaths. In the United States, there have been 33,741,532 confirmed cases with 603,880 deaths [2]. At the time of writing, the United States, India, and Brazil have the highest numbers of confirmed cases globally, followed by France, Russia, Turkey, and the United Kingdom [2].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Multiple-Inputs Convolutional Neural Network for COVID-19 Classification and Critical Region Screening From Chest X-ray Radiographs: Model Development and Performance Evaluation

Li¹,

Li²,

Yao³

et al. 2022

JMIR Bioinform Biotech

View full text Add to dashboard Cite

Background The COVID-19 pandemic is becoming one of the largest, unprecedented health crises, and chest X-ray radiography (CXR) plays a vital role in diagnosing COVID-19. However, extracting and finding useful image features from CXRs demand a heavy workload for radiologists. Objective The aim of this study was to design a novel multiple-inputs (MI) convolutional neural network (CNN) for the classification of COVID-19 and extraction of critical regions from CXRs. We also investigated the effect of the number of inputs on the performance of our new MI-CNN model. Methods A total of 6205 CXR images (including 3021 COVID-19 CXRs and 3184 normal CXRs) were used to test our MI-CNN models. CXRs could be evenly segmented into different numbers (2, 4, and 16) of individual regions. Each region could individually serve as one of the MI-CNN inputs. The CNN features of these MI-CNN inputs would then be fused for COVID-19 classification. More importantly, the contributions of each CXR region could be evaluated through assessing the number of images that were accurately classified by their corresponding regions in the testing data sets. Results In both the whole-image and left- and right-lung region of interest (LR-ROI) data sets, MI-CNNs demonstrated good efficiency for COVID-19 classification. In particular, MI-CNNs with more inputs (2-, 4-, and 16-input MI-CNNs) had better efficiency in recognizing COVID-19 CXRs than the 1-input CNN. Compared to the whole-image data sets, the efficiency of LR-ROI data sets showed approximately 4% lower accuracy, sensitivity, specificity, and precision (over 91%). In considering the contributions of each region, one of the possible reasons for this reduced performance was that nonlung regions (eg, region 16) provided false-positive contributions to COVID-19 classification. The MI-CNN with the LR-ROI data set could provide a more accurate evaluation of the contribution of each region and COVID-19 classification. Additionally, the right-lung regions had higher contributions to the classification of COVID-19 CXRs, whereas the left-lung regions had higher contributions to identifying normal CXRs. Conclusions Overall, MI-CNNs could achieve higher accuracy with an increasing number of inputs (eg, 16-input MI-CNN). This approach could assist radiologists in identifying COVID-19 CXRs and in screening the critical regions related to COVID-19 classifications.

show abstract