Research Article
A Lightweight Deep Learning-Based Pneumonia Detection Approach for Energy-Efficient Medical Systems
Table 11
Comparison of results of current related work.
| | Related work (year) | Approach/model | No. of images | Accuracy |
| | Ayan and Ünver [50] (2019) | VGG16 | 5856 | 84.50% | | Abiyev and Ma’aitah [41] (2018) | CNN | 1000 | 92.40% | | Cohen et al. [43] (2019) | DenseNet-121 | 5232 | 92.80% | | Stephen et al. [42] (2019) | CNN | 5856 | 93.73% | | Saraiva et al. [49] (2019) | CNN | 5840 | 94.40% | | Rajaraman et al. [44] (2018) | Customized CNN | 5856 | 96.20% | | Chouhan et al. [56] (2020) | Ensemble of pretrained CNNs | 5232 | 96.39% | | Toğaçar et al. [57] (2019) | Combined features of pretrained AlexNet, VGG-16, and VGG-19 with mRMR feature selection algorithm | 5849 | 96.84% | | Rahman et al. [51] (2020) | Transfer learning with deep CNN | 5247 | 98.00% | | Hashmi et al. [58] (2020) | Combined the weighted predictions from ResNet18, Xception, InceptionV3, DenseNet121, and MobileNetV3 | 5856 | 98.43% | | This work (2021) | Pretrained DenseNet-121+ optimized DNN using random search algorithm | 5856 | 98.90% |
|
|
