Abstract

Cancer is one kind of dangerous disease which can cause by the growth of unwanted cell. There are many different types of cancer and bone tumor is one of the part of it and this has to be detected at earlier stage. Here, Magnetic Resonance Images will be used as input data and then preprocessing operations will be done and then features will be extracted and given to the convolution neural network which is an algorithm used in order to classify the images as tumor and non tumor. This algorithm gives best accuracy and performance with minimum loss.

Keyword:
Benign,
malignant,
image processing,
AI tools
Introduction:
Cancer of the bones and joints was the 3rd leading cause of cancer related deaths in people under the age of 20 in the United States in 2016, with approximately 3,500 new bone cancer diagnoses in 2019. Outside of bone metastases (secondary bone tumors) and plasma cell myeloma, the most common bone malignancies are osteosarcoma, chondrosarcoma, and Ewing sarcoma . While the incidence of benign bone tumors is more difficult to determine because they are rarely fully evaluated or biopsied, osteochondroma, enchondroma and osteoid osteoma are among the most common benign tumors Radiographs are the recommended first line imaging modality for
the characterization of bone lesions, as it typically provides clear assessment of lesion location, internal matrix, margins, and associated periosteal reaction . These lesion characteristics in combination with patient age are often sufficient to provide differential diagnoses of bone lesions . However, radiographs have limitations. Superimpositions, poorly-visualized partial cortex destruction, and difficulties analyzing flat and short bones and soft tissues can make radiographic diagnosis more challenging . In addition, the clinical symptoms and radiographic appearance of infections (osteomyelitis) often mimic those of bone tumors . Image diagnosis of primary bone lesions can be further complicated by the presence of pathologic fractures, which can increase the amount of fluid, hemorrhage, or edema in and around a lesion; this is particularly relevant
for benign lesions such as non-ossifying fibroma, aneurysmal bonecyst, and fibrous dysplasia . Chondroblastoma, osteoid osteoma, and Langerhans cell histiocytosis are among the benign bone lesions that can present with extra-lesional edema-like signal even in the absence of pathological fracture .