Abstract:
An established paradigm in the field of imaging in medical is fracture detection. These days,
computer-aided diagnostic-(CAD) systems are widely used because they help physicians and other
medical experts diagnose various ailments more accurately by interpreting medical pictures. In a
similar vein, pressure, accidents, and osteoarthritis are typical causes of bone fractures.
Furthermore, bone is a hard component that sustains the entire body. As a result, the significant
issue of the last year is considered to be the bone fracture. Machine vision-based bone fracture
identification is becoming more and more significant in CAD systems since it helps lessen
physician burden by weeding out cases that are simple to handle. This work develops multiple
image processing approaches for the detection of fracture types in the tibia and femur, the lower
leg bones. The aim of this study is to identify the type of fracture and determine if both the femur
and tibia are fractured from an x-ray picture. Numerous techniques and algorithms have been
developed to precisely identify and categorize photos according to whether or not fractures are
present in various body areas. Two class types—Fractured and Normal—as well as models based
on deep learning have been used in this specific experiment.MobileNetV2,DenseNet169,
InceptionV3, VGG16, VGG19, and RestNet50 are the six models used to predict and recognize
X-ray images for the categorization of bone fractures. Lastly, two types of evaluations of
performance are used to evaluate the technique's outputs. Using four potential outcomes—TP, TN,
FP, and FN—performance assessment for fracture and normal situations is the first of all accuracy
set. The following step is to use these models to analyze each fracture type's accuracy within error
situations. With the VGG16 model, which it emilite’s 97.77% reliability, my proposed technique
paves the way for autonomous identification of femur and tibia bone fractures.
