RESUMO
The outbreak of COVID-19 has shocked the entire world with its fairly rapid spread, and has challenged different sectors. One of the most effective ways to limit its spread is the early and accurate diagnosing of infected patients. Medical imaging, such as X-ray and computed tomography (CT), combined with the potential of artificial intelligence (AI), plays an essential role in supporting medical personnel in the diagnosis process. Thus, in this article, five different deep learning models (ResNet18, ResNet34, InceptionV3, InceptionResNetV2, and DenseNet161) and their ensemble, using majority voting, have been used to classify COVID-19, pneumoniæ and healthy subjects using chest X-ray images. Multilabel classification was performed to predict multiple pathologies for each patient, if present. Firstly, the interpretability of each of the networks was thoroughly studied using local interpretability methods-occlusion, saliency, input X gradient, guided backpropagation, integrated gradients, and DeepLIFT-and using a global technique-neuron activation profiles. The mean micro F1 score of the models for COVID-19 classifications ranged from 0.66 to 0.875, and was 0.89 for the ensemble of the network models. The qualitative results showed that the ResNets were the most interpretable models. This research demonstrates the importance of using interpretability methods to compare different models before making a decision regarding the best performing model.
RESUMO
Tropical cities are more susceptible to the suggested fall outs from projected global warming scenarios as they are located in the Torrid Zone and growing at rapid rates. Therefore, research on the mitigation of urban heat island (UHI) effects in tropical cities has attained much significance and increased immensely over recent years. The UHI mitigation strategies commonly used for temperate cities need to be examined in the tropical context since the mechanism of attaining a surface energy balance in the tropics is quite different from that in the mid-latitudes. The present paper evaluates the performance of four different mitigation strategies to counterbalance the impact of UHI phenomena for climate resilient adaptation in the Kolkata Metropolitan Area (KMA), India. This has been achieved by reproducing the study sites, selected from three different urban morphologies of open low-rise, compact low-rise and mid-rise residential areas, using ENVI-met V 4.0 and simulating the effects of different mitigation strategies- cool pavement, cool roof, added urban vegetation and cool city (a combination of the three former strategies), in reducing the UHI intensity. Simulation results show that at a diurnal scale during summer, the green city model performed best at neighborhood level to reduce air temperature (Ta) by 0.7⯰C, 0.8⯰C and 1.1⯰C, whereas the cool city model was the most effective strategy to reduce physiologically equivalent temperature (PET) by 2.8° - 3.1⯰C, 2.2° - 2.8⯰C and 2.8° - 2.9⯰C in the mid-rise, compact low-rise and open low-rise residential areas, respectively. It was observed that (for all the built environment types) vegetation played the most significant role in determining surface energy balance in the study area, compared to cool roofs and cool pavements. This study also finds that irrespective of building environments, tropical cities are less sensitive to the selected strategies of UHI mitigation than their temperate counter parts, which can be attributed to the difference in magnitude of urbanness.