AU2021101325A4 - Methodology for the Detection of Bone Fracture in Humans Using Neural Network - Google Patents

Methodology for the Detection of Bone Fracture in Humans Using Neural Network Download PDF

Info

Publication number
AU2021101325A4
AU2021101325A4 AU2021101325A AU2021101325A AU2021101325A4 AU 2021101325 A4 AU2021101325 A4 AU 2021101325A4 AU 2021101325 A AU2021101325 A AU 2021101325A AU 2021101325 A AU2021101325 A AU 2021101325A AU 2021101325 A4 AU2021101325 A4 AU 2021101325A4
Authority
AU
Australia
Prior art keywords
image
fractures
images
bone
neural network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
AU2021101325A
Inventor
Lakshmi B.
Fareeza F.
Seetha J.
Rituraj Jain
Jayanthi Kannan M. K.
Madiajagan M.
Anand Shukla
Saurabh Shukla
Subramani T.
Thamari Thankam
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to AU2021101325A priority Critical patent/AU2021101325A4/en
Application granted granted Critical
Publication of AU2021101325A4 publication Critical patent/AU2021101325A4/en
Ceased legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/45—For evaluating or diagnosing the musculoskeletal system or teeth
    • A61B5/4504—Bones
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7253—Details of waveform analysis characterised by using transforms
    • A61B5/726—Details of waveform analysis characterised by using transforms using Wavelet transforms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
    • A61B6/505—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for diagnosis of bone
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
    • A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
    • A61B6/5217—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data extracting a diagnostic or physiological parameter from medical diagnostic data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30008—Bone

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Pathology (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Animal Behavior & Ethology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Surgery (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Veterinary Medicine (AREA)
  • Artificial Intelligence (AREA)
  • Theoretical Computer Science (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Physiology (AREA)
  • General Physics & Mathematics (AREA)
  • Radiology & Medical Imaging (AREA)
  • Evolutionary Computation (AREA)
  • Psychiatry (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Signal Processing (AREA)
  • Mathematical Physics (AREA)
  • Dentistry (AREA)
  • Data Mining & Analysis (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Optics & Photonics (AREA)
  • Orthopedic Medicine & Surgery (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Fuzzy Systems (AREA)
  • Computational Linguistics (AREA)

Abstract

Every day, new and rapidly evolving technologies appear in a number of fields, especially in the medical sector. However, some older strategies are still commonly used, reliable, and useful in this regard. One of these methods for identifying bone fractures is X-rays. However, often the scale of fractures is negligible and cannot be readily identified. As a consequence, powerful and intelligent systems should be developed. The aim of this innovation is to establish an intelligent classification device that can detect and diagnose bone fractures. The evolved method is split into two phases. The images of the fractures are analyzed using various image processing techniques in order to detect their location and shapes in the first stage, and then a back propagation neural network is trained and evaluated on the processed images in the second stage. The method was checked on multiple bone fracture images in the lab, and the findings demonstrated high performance and a high classification rate. 1 Ori inal mage Redced Size3 Layer BackBoeFatr Orgia Img eucdSz Propagation Neural BoeFatr (256X256) image (32X32) Network Detection Resultant Image Fig. 1

Description

Orgia ImgOri inal Redced mage Size3 Layer BackBoeFatr eucdSz Propagation Neural BoeFatr (256X256) image (32X32) Network Detection
Resultant Image
Fig. 1
TITLE OF THE INVENTION Methodology for the Detection of Bone Fracture in Humans Using Neural Network
FIELD OF THE INVENTION
[001]. The present disclosure is generally related to a Methodology for the Detection of Bone Fracture in Humans Using Neural Network.
BACKGROUND OF THE INVENTION
[002]. This invention aims to develop an intelligent classification system that would be capable of detecting and classifying the bone fractures. The images of the fractures are processed using different image processing techniques in order to detect their location and shapes and experimentally, the system was tested on different bone fracture images and the results show high efficiency and a classification rate. In this invention, image processing and neural network approaches for the detection of bone fractures were developed which will be useful and significant in the medical field such as in X-ray radiographs.
SUMMARY OF THE INVENTION
[001]. Any previous studies in the literature aimed to present a stable digital data compression and modulation for efficient voice data transmission. The primary aim was to achieve higher data speeds, smaller bit error rates, and lower bandwidth usage. The proposed technique's efficiency is contrasted to that of an existing technique for data transmission over speech channels.
[0021. An intellectual classification method for bone fractures is developed in this innovation. The production phase and the classification phase are the two major stages of the suggested method. The photographs are processed in the image processing phase using techniques such as Haar Wavelet transforms and SIFT as a function extractor. These procedures are used to increase the visual clarity and remove the broken section of the bone. The images are able to be fed into the next phase, which is the neural network, at the end of this phase. The proposed framework is implemented and simulated using the Matlab programming language.
[0031. The functional study of the proposed bone fracture detection method indicated high reliability and a high rate of outperformance. Last but not least, the findings of the proposed SIFT algorithm would help medical professionals in the diagnosis and assessment of bone fractures.
DETAILED DESCRIPTION OF THE INVENTION
[0041. Any previous studies in the literature aimed to present a stable digital data compression and modulation for efficient voice data transmission. The primary aim was to achieve higher data speeds, smaller bit error rates, and lower bandwidth usage. The proposed technique's efficiency is contrasted to that of an existing technique for data transmission over speech channels. In addition, from two-dimensional images, a Rotational Haar Wavelet Transform (RHWT) method is developed to describe the fabric of particulate assemblies. The fabric position and strength are shown by the Maximal Energy Ratio.
[005]. There are varying numbers of elements that can be labeled as "feature" portrayals of an image in any form of picture. These elements are isolated and documented using a range of techniques and methods. Filter picture highlights display how the elements are organized. Another well-known Scale invariant feature transform (SIFT) algorithm is used to highlight images. It takes a photo and transforms it into a wide set of neighborhood vectors. Scaling, pivoting, and reading artifacts are all considered invariant for these vectors. For primate vision, SIFT indicates a wide number of components. A separation process is used to remove these materials.
[006]. The device has been developed to diagnose bone fractures. This can be achieved by using a range of image processing processes and techniques that have been validated in broken environments. As a consequence, any given images are analyzed, and a fracture on the original image should be observed using one of the well-known Wavelet transform algorithms. The proposed method allows use of bone fracture photographs from a Northern Cyprus general hospital's orthopedics and traumatology department.
[007]. SIFT feature vectors are used in the proposed process. This suggested approach, in particular, considers the salient points and key-points for the task of feature extraction for the first time.
[0081. An intellectual classification method for bone fractures is developed in this innovation. The production phase and the classification phase are the two major stages of the suggested method. The photographs are processed in the image processing phase using techniques such as Haar Wavelet transforms and SIFT as a function extractor. These procedures are used to increase the visual clarity and remove the broken section of the bone. The images are able to be fed into the next phase, which is the neural network, at the end of this phase. The proposed framework is implemented and simulated using the Matlab programming language.
[0091. As previously mentioned, SIFT implements a method for identifying interest points in a grey-level image that depends on image intensities stored in image structures. The aim of this analysis is to see how well a back- propagation neural network can distinguish various forms of bone fractures. The established structure consists of picture and classification phases, as described in the previous section. The fracture images are preprocessed during the manufacturing phase. After that, SIFT is used to remove functions. These methods are used to improve image clarity and assess the extent of a fracture.
[00101. Haar Wavelet Transform: Wavelet theorems are commonly used in image manipulation, denoising, and compression, and image transformations are widely used in image filtering and related applications. The Haar wavelets are used in a number of discrete image transformations and processing processes. Previous research has looked at the use of compression techniques in general, as well as wavelet compression of medical images. More recently, by using Haar wavelet image compression on a range of images, a neural network model was used to achieve the best compression ratio.
[00111. The Haar wavelet transform technique was used to strengthen the bone fractures in this analysis. The original image was first converted to grayscale to save Processor time, and then image filtering was applied to eliminate any excess data from the image.
[00121. Smoothing, also known as blurring, is an image processing method for eliminating noise in Ulan photographs in order to create meaningful images for subsequent steps. Wavelet transformations, unlike the discrete cosine transform, are not based on the fourier transform, but they can accommodate image data discontinuities better.
[00131. Scale-Invariant Feature Transform (SIFT) algorithm: Another method for function extraction is the SIFT, which employs statistics of gradient directions of image intensities. These intensities are accumulated in each interest point's image structure. The descriptor is used to connect related interest points through images. The suggested algorithm was extended to various types of images as an attribute descriptor and extractor. SIFT descriptor uses a method to detect an interest point in a gray level image. The statistics of gradient directions of image intensities are used in this process.
[00141. These intensities are accumulated in each interest point's image structure. The descriptor is used to connect related interest points through images. This algorithm has been used by several researchers as a function extractor in conjunction with intelligent classifiers like neural networks. The algorithm was incredibly successful at obtaining the relevant characteristics.
[00151. The method starts by analyzing and manipulating the bone fracture images in order to obtain a free-noise image from the original image. The feature extractions stage is where the features are extracted using SIFT. Following the recognition and retrieval of the features, they are fed into a back-propagation neural network.
[0016]. To remove any features that can represent the whole bone structure, an algorithm was used. SIFT transforms an image into a "huge array of local feature vectors" in order to create image attributes. Scaling invariance, rotation, and translation are all properties of these local function vectors. These features were extracted using a five-stage filtering system. These steps are as follows:
[00171. Scale-Space Extrema Detection: Separation of efforts is conducted in this process to identify areas and sizes from different viewpoints of the input picture. Localization of key points: This stage uses the Laplacian distribution to find low differentiations on an edge from the key point run-down.
[00181. Orientation Assignment: The aim of this progression is to have a steady introduction to the key points related to image properties in the neighborhood. The main point descriptor, seen below, can then be spoken to in relation to this introduction, achieving rotation invariance.
[00191. Key point Descriptor: The above-mentioned neighborhood angle information is often used to build key point descriptors. The data for the slope is obtained by rotating the main point and then weighting it. A Gaussian fluctuation of 1.5 x key point scale is used to perform this weighted procedure.
[00201. Key-point Matching: The closest neighbors of key-points in two photos are marked and paired.
[00211. However, in some situations, the second nearest match can be exceptionally similar to the first, which may be due to noise or other factors. The closest-distance to second-closest-distance ratio is used in this situation. They are refused if it is greater than 0.8. It excludes nearly 90% of incorrect matches while discarding only 5% of right matches.
[00221. The sample photos SIFT added and the photos' attributes are extracted using the SIFT algorithm and used to train the back-propagation network. Following training and convergence, the testing images are used to assess the neural network after it has been through the SIFT feature extraction process.
[0023]. IBFDS: Intelligent Bone Fracture Detection System: Two conventional 3-layer back propagation neural networks (BPNN) with 1024 input neurons are used in the Intelligent Bone Fracture Detection System (IBFDS). The bone is identified by output neurons using binary coding: [1 0] for a broken bone and [0 1] for a bone without a fracture. Neurons in the secret and output layers are triggered using the sigmoid activation feature.
[0024]. The proposed BPNN's topology as seen in Figure 1. The use of a back propagation neural network, which is a supervised learner, has been favored due to its ease of deployment and the availability of an appropriate "input - goal" database for training. The architecture of back propagation neural networks is illustrated in Figure 1.
[0025]. This process involves both preparation and research (generalization). The picture collection of 30 pictures is used for training and 70 pictures are used for testing.
[0026]. Initial random weights of between -0.3 and 0.3 were used during the learning process. During different tests, the learning rate and momentum rate were modified in order to obtain the necessary minimum error value and realistic learning. For this application, a 0.002 error value was deemed appropriate.
[0027]. The results of the iterations using the Super computer indicate that BPNN finished the training in 25 seconds after 3742 iterations. After several tests, hidden layer nodes, learning speeds, and momentum rates for the neural network were discovered..
[0028]. Image recognition and neural network methods for the identification of bone fractures were built in this invention. In the medical field, such as in X-ray radiographs, image processing techniques are extremely useful and important. IBFDS recruits BPNNs, or back propagation neural networks. To detect bones with and without fractures, the built BPNN uses preprocessed, resized original images.
[0029]. The aim of this analysis was to compare and analyze the findings, as well as the efficiency of a back propagation neural network combined with a suggested SIFT algorithm in detecting various bone fractures. IBFDS outperforms the SIFT algorithm in terms of detection rations, as described in the findings and discussion section.
[0030]. Since it was qualified to detect bone fractures, the established method is a successful one. The image that was used to train and validate the proposed method came from the benchmark database, which has 100 images. For each subject, the photos contain various fracture sizes and lighting conditions, which helped us, improve the system's performance and reliability. Finally, the functional study of the proposed bone fracture detection method indicated high reliability and a high rate of outperformance.
[0031]. The findings of the proposed SIFT algorithm would help medical professionals in the diagnosis and assessment of bone fractures. Future studies could refine and develop the proposed algorithm for biorthogonal and daubechies wavelet transforms and their effects on fracture determination to improve the percentage ratios.

Claims (5)

  1. CLAIMS: We Claim: 1. This invention is related to a methodology for the detection of bone fracture in humans using Neural Network.
  2. 2. As we claimed in 1, this invention develops an intelligent classification system that would be capable of detecting and classifying the bone fractures.
  3. 3. As we claimed in 1 and 2, in this invention images of the fractures are processed using different image processing techniques in order to detect their location and shapes.
  4. 4. The proposed work of this invention achieves high efficiency and high classification rate in detecting the bone fractures.
  5. 5. We claim that this invention will help in the medical field such as in X-ray radiographs.
    3 Layer Back Original Image Reduced Size Bone Fracture Propagation Neural (256X256) Image (32X32) Detection Network 2021101325
    Resultant Image
    Fig. 1
AU2021101325A 2021-03-14 2021-03-14 Methodology for the Detection of Bone Fracture in Humans Using Neural Network Ceased AU2021101325A4 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
AU2021101325A AU2021101325A4 (en) 2021-03-14 2021-03-14 Methodology for the Detection of Bone Fracture in Humans Using Neural Network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
AU2021101325A AU2021101325A4 (en) 2021-03-14 2021-03-14 Methodology for the Detection of Bone Fracture in Humans Using Neural Network

Publications (1)

Publication Number Publication Date
AU2021101325A4 true AU2021101325A4 (en) 2021-05-13

Family

ID=75829025

Family Applications (1)

Application Number Title Priority Date Filing Date
AU2021101325A Ceased AU2021101325A4 (en) 2021-03-14 2021-03-14 Methodology for the Detection of Bone Fracture in Humans Using Neural Network

Country Status (1)

Country Link
AU (1) AU2021101325A4 (en)

Similar Documents

Publication Publication Date Title
Qian et al. Thinking in frequency: Face forgery detection by mining frequency-aware clues
Dimililer IBFDS: Intelligent bone fracture detection system
Kaur et al. Deep convolutional neural network for graphics forgery detection in video
US20020181786A1 (en) Intelligent systems and methods for processing image data based upon anticipated regions of visual interest
Abrahim et al. RETRACTED ARTICLE: Splicing image forgery identification based on artificial neural network approach and texture features
Luo et al. Fingerprint Classification Combining Curvelet Transform and Gray‐Level Cooccurrence Matrix
Wan et al. Generative adversarial multi-task learning for face sketch synthesis and recognition
Aria et al. QDL-CMFD: A Quality-independent and deep Learning-based Copy-Move image forgery detection method
Shirdhonkar et al. Off-line handwritten signature identification using rotated complex wavelet filters
Ganeshbabu Glaucoma image classification using discrete orthogonal stockwell transform
Tahir et al. Kidney stones detection based on deep learning and discrete wavelet transform
AU2021101325A4 (en) Methodology for the Detection of Bone Fracture in Humans Using Neural Network
Dinariyah Accuracy enhancement in face recognition using 1D-PCA & 2D-PCA based on multilevel reverse-biorthogonal wavelet transform with KNN classifier
Mr et al. Developing a novel technique to match composite sketches with images captured by unmanned aerial vehicle
JP2003533785A (en) Method and apparatus for determining an object in an image
Rezaei et al. A robust fingerprint recognition system based on hybrid DCT and DWT
Albu Face anti-spoofing based on Radon transform
Zhong et al. An efficient saliency detection model based on wavelet generalized lifting
Le-Tien et al. Combined Zernike moment and multiscale Analysis for tamper detection in digital images
Filius Interpol review of imaging and video 2016–2019
binti Ashari et al. Multi-scale texture analysis for finger vein anti-spoofing
Krishnan et al. Web-based remote sensing image retrieval using multiscale and multidirectional analysis based on Contourlet and Haralick texture features
Shaji et al. Brain Tumor Segmentation Using Modified Double U-Net Architecture
Shaju et al. Haar wavelet transform based histogram concatenation model for finger print spoofing detection
Serdaroglu et al. Defect detection in textile fabric images using subband domain subspace analysis

Legal Events

Date Code Title Description
FGI Letters patent sealed or granted (innovation patent)
MK22 Patent ceased section 143a(d), or expired - non payment of renewal fee or expiry