Bone Fracture Detection Using Morphological and Comparing the Accuracy with Genetic Algorithm Y. Rakesh

OriginalArticle
 

---

Journal of Pharmaceutical Negative Results ¦Volume13¦SpecialIssue 4¦2022
271
 
Y. Rakesh
,
etal
.:
Bone Fracture Detection Using Morphological and Comparing the Accuracy with Genetic Algorithm 
team has extensive knowledge and research experience that has translate into high quality publications(Bhansali 
et al. 2021; Jayanth et al. 2021; Sudhakar, Ravel, and Perumal 2021; Sathiyamoorthi et al. 2021; Deepanraj et al. 
2021; Raju et al. 2021; Arun Prakash et al. 2020; Kamath et al. 2020; Shanmugam et al. 2021; Rajasekaran et al. 
2020; Adhinarayanan et al. 2020; Rajesh et al. 2020; Aurtherson et al. 2021) 
The limitations of the study of existing bone fracture detection is automatically discovered in MATLAB, 
resulting in a decline in accuracy (Santosh and Hegadi 2019). The goal of this work is to use a morphological 
algorithm to detect broken bones with a higher readability rate than a genetic algorithm. 
MATERIALS AND METHODS 
 
This study work is for the Department of Electronics and Communication Engineering at Saveetha School of 
Engineering, Chennai. Two groups noise images and image classification for the image from fracture detection, 
each dataset consists of 16 samples, in total 32 samples with threshold 0.05, 95% confidence and pretest power 
80% is taken for test purpose (Ryan 2013). 
For Morphological, two processes took the 16 samples from the dataset. The input x-ray image which is 
collected from the dataset is rescaled into color jpeg (200*180) resolution. By the help of a morphological 
algorithm (Shivahare 2020) the novel feature extension of the images is processed. These samples are stored in 
Microsoft Excel for statistical analysis. Morphology is used for sample preparation and the simulation is done in 
MATLAB. The images are helpful to store the data and these images are labeled on the basis of the folder. By 
deep learning the data is separated into testing and training sets. The input image of size 200*180 is tuned 
perfectly for the classification of the datasets by the novel feature extraction, the x-ray images with different 
sizes cause noise and to reduce this data size will increase and the validation, recognition of the data are done. 
Similar to Morphological, the two processes are done for Genetic with 16 samples. the input images from the 
dataset is rescaled into 24-bit color jpeg (227*227) resolution. The Novel Feature Extraction and the 
classification of the data is carried out by the Genetic algorithm. The sample values are stored in Microsoft 
Excel for further statistical analysis. The Factor values of mainly connected layers are increased for the training 
options in the transferred layers to set the initial rate for fully connected and transferred layers in order to 
validate the data from the stored database. 
The total experiment was done on a Windows platform with dual core processor, resolution of 1024×768 pixels, 
configuration of 10 th generation, intel i3, 8GB RAM, 1TB HDD, 256 SSD and MATLAB 2018 software with 
required add-ons for training and testing procedures. The scaling of the image is done at the pre-processing 
stage and morphologically detects the image, Novel Feature Extraction is done by selecting the factor values and 
the images are extracted. These factor values have best transferred layers which have greater dimension and this 
allows for better performance which leads to the successful detection of image. The removal of the artifacts 
from the dataset images is done in the pre-processing part. The dataset images are collected wrist fractures and 
leg fractures are marked in deep learning, The preprocess normalization is implemented in the training images 
and processed using a morphological algorithm for Novel Feature Extraction. These data samples differ from 
each other and they are like images with dim light, heavy light, etc. The position of these coordinates are 
tabulated in table-1. The blur images and the images with disturbance are the independent variables and 
accuracy is the dependent variable in this image recognition system. Independent sample T test is performed and 
the accuracy of approximate is given in Equation [1] as follow 
Accuracy = TP + TN / TP + TN + FP +FN
------------ [1] 
Where, 
TP - true positive 
TF - true negative 
FP - false positive 
FN - false negative