Certain investigations on efficient and short time frame pathology localization techniques using image registration and learning algorithms

Abstract

Pathology Localization, the process of determining the anatomical newlinelocation of a pathology in an image is an important step in the diagnosis of a newlinedisease. Manual detection of abnormalities in medical images is tedious and newlinetime consuming. Automatic detection of pathology in images can help in newlinereducing the workload of pathologists and speed up the diagnosis. newlineAdvancements in medical imaging technologies have enabled high-quality newlinevisualization of tissue structures for anatomical and pathological newlineexaminations. The need for pathology localization is strongly felt in image guided newlineinterventions in the management of tumors, in which precise tumor newlinelocalization is essential for precise tumor targeting and dose escalation. newlineRecently, the outbreak of COVID19 has also witnessed the significance of newlineimage guided interventions in screening, diagnosis and management of this newlinevirus and comorbidities such as pulmonary embolism, cardiomegaly and newlineventricular enlargement. Evolution of machine learning and deep learning approaches, and newlinenovel anatomical and functional imaging modalities have resulted in several newlinecomputer-aided diagnosis and detection systems. These systems are centered newlinearound pathology localization, detection and classification of abnormalities. newlineConventional medical image analysis approaches such as classification and newlinesegmentation are tailored to the problem of detecting or classifying newlineabnormalities in pathology images. Machine learning and deep learning newlinemodels are trained on a large number of pathology images in the form of newlinelabeled image datasets, so that they generalize well with unseen data. newline This research is aimed at developing fast and reliable systems for newlinepathology localization towards improving diagnostic accuracy. Toward this newlineend, the following specific objectives have been defined: newlinei. To develop a machine learning model for tumor volume newlinemeasurement and segmentation ii newline