Google Drive Link : https://drive.google.com/drive/folders/1aTHmENrN4bBCy4GNhofpe4DHhQUPlcyg?usp=sharing
About The Project:
Pneumonia affects a large number of individuals, especially children, in developing countries where risk factors include poor sanitary conditions, malnutrition, and a lack of appropriate medical services. Pneumonia becomes the reason for the death of around three lakh children in India and to our astonishment half of the pneumonia deaths are of children in India. Most of the time the disease is being ignored. Many lives can be saved if it is detected on time. For a successful recovery, pneumonia must be detected early. X-ray scan examination is the most common way of diagnosis, albeit it is based on the radiologist's interpretation ability. An autonomous system with generalising capabilities is required to detect the condition. Some studies have shown that discrepancies in the interpretation of chest x-rays by radiologists are common. Chest X-rays are currently the most common imaging modality read by radiologists in hospitals and teleradiology practices. These lower resolution modalities, despite their widespread use, are not the easiest to interpret. To read them correctly, you must first understand the viewing constraints caused by patient positions, image quality, and tissue overlays. This disadvantage of human-based observation motivated to intervene in this area where the machine classifies normal and abnormal chest x-rays. This is complicated by the fact that there's been no major initiative in the radiology community to catalogue the wide range of possible findings seen among chest radiographs. Second, because many of the methods were trained on single hospital source datasets, their generalizability to different demographics around the world is dubious. Third, the evaluation methodology for these algorithms employs well-known AI metrics such as area under the curve (AUC) or F-scores for label-based precision and recall evaluations.