International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
CLASSIFICATION OF BONE TUMOR USING EFFICIENT-NET B0 Sri P. Anil Kumar1, Thanuri Bhanu Harshitha2, Yarlagadda Rajeswari3, Sanaka Venkata Karthikeya4, Ramireddy Usha Kumari5 1 Assistant professor, Department of Electronics and Communication Engineering, Seshadri Rao Gudlavalleru
Engineering college, Andhra Pradesh, India 2,3,4,5 U.G Student, Department of Electronics and Communication Engineering, Seshadri Rao Gudlavalleru
Engineering college, Andhra Pradesh, India ---------------------------------------------------------------------***--------------------------------------------------------------------outcomes in computer vision, natural language processing, Abstract - Accurately diagnosing bone tumors is crucial for
and healthcare, among other domains. Deep learning models, and in particular convolutional neural networks (CNNs), have shown promising results in automated medical image interpretation applications. Compared to traditional methods, these models may offer benefits including improved accuracy, consistency, and efficiency.[2]
patient management and creating effective treatment plans. In this study, we propose a unique approach that uses deep learning algorithms to diagnose bone tumors using medical imaging data. Our method combines picture segmentation and the EfficientNetB0 architecture to enable high-performance tumor classification. First, pre-processing and segmentation of the input medical images are done to identify regions of interest that might be malignancies. The convolutional neural network EfficientNetB0 is then fed the segmented areas to carry out feature extraction and classification. EfficientNetB0 is known for its exceptional performance and computational efficiency, which enables strong learning from the separated tumor zones. We train and validate our model on a large dataset of annotated bone tumor images to guarantee generalizability and reliability. Our method's exceptional levels of sensitivity, specificity, and accuracy in diagnosing bone tumors are demonstrated by the results of our experiments. This method offers a scalable and efficient way to automatically identify tumors from medical images, which presents a potential way to improve clinical judgment in the diagnosis and treatment of bone malignancies.
Image segmentation is crucial for medical image analysis because it makes it easier to identify and separate areas of interest—like cancers—from the surrounding anatomical structures. By accurately segmenting bone tumor locations, clinicians can obtain valuable information on tumor features, such as size, shape, and texture, which are useful for diagnosis and therapeutic planning. Moreover, segmentation facilitates the extraction of quantitative elements from image data that serve as algorithmic input for classification, enabling automatic tumor categorization.[3]
2. LITERATURE SURVEY [1] Several scientists have seen trends in the information and data that they have obtained from large databases and pertinent websites. Learning vector quantization, fuzzy theory, probabilistic neural networks, association rule mining, and supporting vector machines are the most widely used methods for diagnosing and classifying bone cancer. In order to segment bone pictures, the k means clustering algorithm was used in this work. To process the segmented image further for the aim of identifying bone cancer, the mean intensity of the detected area is evaluated. It is recommended to classify medical images based on the presence or absence of bone cancer using threshold values.
Key Words:
Deep Learning, Bone Tumors, Image Segmentation, EfficientNetB0, Classification, Radiology.
1.INTRODUCTION Bone tumors are common growths or masses of tissue that develop inside the bones. These tumors can be benign or malignant; malignant tumors pose a major risk to health, including the possibility of spread and death, if treatment is not received. For patients with bone malignancies, timely planning of their treatment and an accurate diagnosis are crucial. Bone malignancies are usually diagnosed by a clinical examination, imaging techniques such as CT, MRI, and X-rays, and histological analysis of tissue samples obtained through biopsy.
[2] The basis of this work is the fusion of computer science with the biomedical field. Numerous image segmentation methods, including Sobel, Prewitt, Canny, K-means, and Region Growing, are described in this paper. These methods can be helpful in understanding MRI and X-ray images as well as in predicting the type of bone cancer. In order to use MATLAB to detect osteosarcoma cancer present on bone, the study also displays the outcomes of edge-based and regionbased image segmentation techniques applied to X-ray pictures.
In recent years, deep learning methods have become more and more popular for use in medical image analysis, particularly when dealing with issues like tumor segmentation and classification. The artificial intelligence field of deep learning has demonstrated remarkable
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 732