Multi-Task Driven Explainable Diagnosis of COVID-19 using Chest X-ray Images

Abstract

With increasing number of COVID-19 cases globally, all the countries are ramping up the testing numbers. While the RT-PCR kits are available in sufficient quantity in several countries, others are facing challenges with limited availability of testing kits and processing centers in remote areas. This has motivated researchers to find alternate methods of testing which are reliable, easily accessible and faster. Chest X-Ray is one of the modalities that is gaining acceptance as a screening modality. Towards this direction, the paper has two primary contributions. Firstly, we present the COVID-19 Multi-Task Network which is an automated end-to-end network for COVID-19 screening. The proposed network not only predicts whether the CXR has COVID-19 features present or not, it also performs semantic segmentation of the regions of interest to make the model explainable. Secondly, with the help of medical professionals, we manually annotate the lung regions of 9000 frontal chest radiographs taken from ChestXray-14, CheXpert and a consolidated COVID-19 dataset. Further, 200 chest radiographs pertaining to COVID-19 patients are also annotated for semantic segmentation. This database will be released to the research community.

Biases associated with database structure for COVID-19 detection in X-ray images

“… In these parts, for most of the papers, we do not consider the article in which the dataset was presented. We found 46 papers in total; 39 correspond to the classification task 8 – 10 , 14 , 21 – 55 , 5 to the triage task 56 – 60 , and two do both tasks 61 , 62 .…”

“… Database Subjects Metadata References Download page Cohen 20 332 Sex-(Image Amount), Age-(Included), Location-(Europe>Others), Dates-(Included), Others-(ICU admission-Survival model). 8 – 10 , 24 – 31 , 55 32 – 38 , 41 , 42 , 44 , 45 , 52 – 54 Cohen BIMCV 63 4899 positive 5242 negative Sex-(Patient amount), Age-(Included), Location-(Spain) Dates-(Included),Others-(Classification test). 14 , 22 , 43 , 46 , 47 , 62 BIMCV-COVID19 Cancer Image Archive 64 105 Sex-(Image/patient amount), Age-(Included), Location-(EE.UU) Dates-(Not included),Others-(ICU admit-Mortality).…”

“… 40 – 42 , 45 , 46 , 48 , 49 Qatar university SIRM 68 65 Sex-(Patient amount), Age-(Included) Location-(Italy),Dates-(Not included), Other-(Some symthoms). 10 , 60 , 69 SIRM CHUAC dataset 51 Privated not validated Sex-(Private not validated),Age-(Private not validated),Location-(Spain). 51 Not available for downloading Radiopaedia 70 16 Sex-(Patient amount), Age-(Included) Location-(Not mentioned), Dates-(Not mentioned).…”

“…For their lower cost compared to computed tomography, X-ray images are valuable assets for COVID-19 recognition (classification) and prognosis (Triage analysis for knowing the best treatment) 7 . Research, in particular, has focused on developing artificial intelligence (AI) models to support the diagnosis of COVID-19 using medical images 8 – 10 .…”

Enhance Portable Radiograph for Fast and High Accurate COVID-19 Monitoring

“…COVID-19 detection using CXR has been well studied in [24][25][26][27][28][29]. Furthermore, the classification of COVID-19 from CXR has been well studied in [30][31][32][33]. Most of these methods use off-the-shelf networks, including VGG-16 or VGG-19 [34,35], ResNet [36] variants such as ResNet-18, ResNet-50, ResNet-101, ResNet-151 [30,37,38], Inception [33], EfficientNet [39,40], DenseNet [41,42].…”