Answers are grounded in this paper and cite their sources.
This research proposes a comprehensive automated approach for the quick and precise identification of Chronic obstructive pulmonary disease (COPD) in chest computerized tomography (CT) scan images of patients. The collection includes 48,260 scanned photographs of 282 healthy persons and 15,591 photos from 95 COPD patients. An image processing method has been created to assess CT scans of the lungs and eliminate any images that lack clear visualization of the interior structures of the lungs. Doing this helps reduce processing time and minimize incorrect detections. We present a novel design that improves the classification accuracy of convolutional networks when applied to photos containing small yet meaningful objects. To maintain the data integrity of the ResNet50V2 model across various image resolutions, the architectural design integrates a novel feature pyramid network specifically designed for classification tasks. The proposed approach dramatically enhances the accuracy of classification, a crucial aspect given that COPD infections can manifest in various sizes, often being exceedingly small. Subsequently, the system employs a preestablished threshold to identify the patient's condition after completing these two procedures. The proposed system has evaluated on Xception and Inception utilizing both the proposed model and ResNet50V2. Throughout the single image classification phase, the proposed framework obtained an impressive 98.91% accuracy on a dataset of 7,996 test photos. During the quick identification phase, the system accurately detected the ailments of approximately 234 out of 245 patients.
No takes yet. Share an insight, caveat, or question.
Chawla et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: