Osteoporosis

Published on 02 Jul 2021. Written by Akshay

Due to the advancement in the technology many miracles are seen in the field of engineering, medical sciences, etc.

Read more..

But many problems also arose due to the growth of many industries and many diseases are also created. Hence a solution should be developed to detect these problems.

Many infections, diseases are such that their detection procedure is not easy and osteoporosis is one among that kind. Hence its detection is a problem even after observing the symptoms.

Osteoporosis is a kind of skeletal system disease which can be categorized by changes in bone mass density. Its structure is so harmful that it leads to the fracture of the bone.

Bone radiograph images are majorly used to detect this type of diseases. But even using these type of instruments to detect the osteoporosis may be difficult.

The major difficulty in analyzing this type of disease is because of the textured image of osteoporosis.

Methodology used

• Artificial neural network (ANN)
• Stacked sparse autoencoder
• X-ray imaging technique
• Fuzzy neural network

Stacked Sparse Auto Encoder (SSAE)

• These consist of series of sparse autoencoder networks which is stacked one over the other
• The input layer here is the successive network which is wired to each of the hidden layers of an auto-encoder.
• The work of this auto-encoder is to minimize the reconstruction error between the input and output data so that high-level features are obtained.
• The main purpose of this method is to study the high-level features obtained from the original data by optimizing the cost function. These are encapsulated in weights.

The other processes which are applied to classify an image from its pixels are preprocessing, image sub-division and feature extraction, and pooling and classification.

Support Vector Machine(SVM)

• The main aim of SVM is to achieve good separation between classes by searching for a class of hyperplanes that can maximize the functional margin at a space where the nearest training example is highest
• It is observed that the linear SVM classifier perform better than other kernel operations
• There is an introduction of a deep network to the local features and hence signature of training images are constructed by the features which belong to that image
• There are labels attached to these signatures which help them to train the SVM classifier. These help in differentiating the Osteoporosis Population (OP) and Control Cases (CC).
• The diagnosis of this disease has given promising results and the accuracy rates were found to be 95.5%.

Kit required to develop Osteoporosis:

• X ray

Technologies you will learn by working on Osteoporosis:

• biotechnology

Any Questions?