Detection Of Tumor Affected Part From Histopathological Bone Images Using Morphological Classification And Recurrent Convoluted Neural Networks

Authors

  • D. Anand, G. Arulselvi, G.N. Balaji

DOI:

https://doi.org/10.47750/pnr.2022.13.S09.617

Abstract

The bone tumor (BT) remains a sickness type that is marked by unlimited cell growth and results in numerous early demises worldwide. Hence, diagnosing BT in the earlier phases and classifying them turned into a vital job for healing the sick person. Springing out of the bone, BT disseminates throughout the body swiftly impacting sick persons. By assessing histopathological images (HpIs), BT’s swift and early prognosis could be started. Centered upon this, a survey has been performed on BT identification employing diverse approaches to image processing and observed that there remain several issues, and the adversities did by BT enhance while it is unidentified at the right time. The abnormalities in the bone image are identified from the region of interest. For these problems, many research works were carried out to develop several Computer-aided diagnosis systems, having as a major challenge to define the features that better represent the images to classify. To overcome the problem this paper aims to develop Recurrent CONVoluted neural networks (Rec-CONVnet) based classification for the assessment of tumor type based on analyzing MRI images. For the smoothing process wiener filter is used to reduce the mean square error (MSE) and hence image enhancement is done by using a convoluted Gabor filter. The performance analysis is done by comparing with three traditional methods such as Dense Convolutional neural network (DCNN), Random Forest (RF), and Decision Tree (DT) in terms of accuracy, sensitivity specificity, f1 score, recall, and is found that the proposed Rec-CONVnet achieves 98.34% of accuracy, 97.96% of precision, 98.72% of recall, 98.33% of specificity and 98.32% of  f1-score.

Downloads

Published

2022-12-10 — Updated on 2022-12-10

Versions

Issue

Section

Articles

How to Cite

Detection Of Tumor Affected Part From Histopathological Bone Images Using Morphological Classification And Recurrent Convoluted Neural Networks. (2022). Journal of Pharmaceutical Negative Results, 4992-5008. https://doi.org/10.47750/pnr.2022.13.S09.617