BrainTumorDetection CSIR.pdf - Aryaman Srivastava

Please enable JavaScript to view the full PDF

- ⬥Contents: WHAT IS A BRAIN TUMOR? - Introduction; types CAUSES OF BRAIN TUMOR - Risk factors SYMPTOMS OF THE BRAIN TUMOR - Indications and statistics CURRENT CURE FOR THE TUMOR - treatments made and yet to be terminated PREVIOUS PROPOSITIONS MADE - Past proposals on this topic MY INNOVATION/CONTRIVANCE - My solution to detect brain tumors using Artificial Intelligence (AI) - The program’s working ADVANTAGES OF AI -Advantages of AI in medical imaging LEGAL ASPECTS - Algorithm Bias - Data Privacy - CyberCrime CONCLUSION - Winding up and the Applicant’s details 1 I. What is a Brain Tumor? ❖ A brain tumor is an abnormal cell growth in the brain. There are two types of brain tumors: ➢ Benign (noncancerous) and malignant (cancerous). ❖ A malignant tumor is the cancerous one if it’s not spotted as well as treated in its premature grade; it may lead to the death of the patient. ❖ Tumors can begin in the brain, and can spread from another part of the body to the brain. ❖ The central nervous system (CNS) is made up of the brain and spinal column, and it is where all vital functions are regulated. ➢ These functions involve thinking, speaking, as well as body movements. ➢ This actively illustrates that a tumor in the CNS will affect a person's thought processes, speech, state, and movement patterns. ❖ Primary tumors are those that reach the brain first, while secondary tumors or metastatic tumors are those that originate elsewhere in the body as well as spread to the brain. ❖ The exact cause of tumors is yet to be known. II. Causes for Brain Tumor – Studies have found the following risk factors for brain tumors: 2 ❏ Ionizing radiation: Ionizing radiation from high dose x- rays (such as radiation therapy from a large machine aimed at the head) as well as other sources can cause cell damage that leads to a tumor. ⬔ Being exposed to ionizing radiation can increase the risk of having a brain tumor. ❏ Family history: It is rare for brain tumors to run in a family. ⬔ Only a very small number of families have several members with brain tumors. III. Symptoms of the Brain Tumor – ❖ Some symptoms include strong headaches, blurred vision, loss of balance, confusion in everyday matters, problems with memory, seizures, muscle jerking or twitching, etc. ❖ In some cases, there may be no symptoms. ■ HEADACHE: Can be acute or persistent ■ MUSCULAR: Difficulty walking, instability, muscle weakness, problems with coordination, weakness of one side of the body, or weakness of the arms and legs ■ WHOLE BODY: Dizziness, fatigue, or vertigo ■ GASTROINTESTINAL: Nausea or vomiting ■ SENSORY: Pins and needles or reduced sensation of touch 3 ■ COGNITIVE: Inability to speak or understand languages or mental confusion ❖ This year, an estimated 23,890 adults (13,590 men as well as 10,300 women) in the United States will be diagnosed with primary cancerous tumors of the brain and spinal cord. ➢ A person’s likelihood of developing this type of tumor in their lifetime is less than 1%. Brain tumors account for 85% to 90% of all primary central nervous system (CNS) tumors. ❖ Statistics state that approximately 78,990 newer victims of primary tumors, (Benign) non-malignant tumors, and other distinct tumors of the central nervous system are diagnosed. ➢ These include about 23,829 primary malignant tumors as well as 55,151 (Benign) non-malignant tumors. 4 ❖ According to the analysis, India is the country that deals with more number diagnoses [as shown in the figure below]. IV. Current Cure for the Tumor – ❖ Diagnosis of a brain tumor is done by a neurologic exam, CT scan and MRI, as well as other tests in a similar manner- an angiogram, spinal tap, and biopsy. ❖ Researchers are examining biomarkers that may help diagnose a brain tumor, estimate a patient’s prognosis, and predict the working of a specific treatment. 5 TREATMENTS INCLUDE ↝ ⭒ Surgery is the usual first treatment for a majority proportion of brain tumors. Surgery to open the skull, a craniotomy is done. ⭒ Radiation therapy kills brain tumor cells with high- energy x-rays, gamma rays, or protons. ⭒ Chemotherapy by the use of drugs to destroy tumor cells, usually by keeping the tumor cells from growing, dividing, and making more cells. ⭒ Immunotherapy, also called biological response modifier (BRM) therapy, is designed to boost the body's natural defenses to fight the tumor. Different methods are being studied for brain tumors, such as the use of dendritic cells or the use of vaccines aimed against a specific molecule on the surface of the tumor cells. ⭒ Oncolytic virus therapy. This therapy uses a virus that infects as well as destroys tumor cells, sparing healthy brain cells. It is currently being researched. ⭒ Targeted therapy treatment targets faulty genes or proteins that contribute to a tumor’s growth and development. ⭒ The blood-brain barrier disruption technique temporarily disrupts the brain’s natural protective barrier to allow chemotherapy to more easily enter the brain from the bloodstream. 6 V. Advantages of AI – ● Artificial Intelligence excels at recognizing complex patterns in images and thus offers the opportunity to transform image interpretation from a purely qualitative and subjective task to one that is quantifiable and effortlessly reproducible. ● In addition, AI may quantify information from images that are not detectable by humans and thereby complement clinical decision-making. ● AI also can enable the aggregation of multiple data streams into powerful integrated diagnostic systems spanning radiographic images, genomics, pathology, electronic health records, as well as social networks. ● With the advancement in technology, the future of healthcare will be transformed due to the generation of big digital datasets acquired utilizing next generation sequencing (NGS), the use of algorithms for image processing, patient-related health records, data arising from large clinical trials, and disease predictions. 7 Fig. 1. An overview of the applications of artificial intelligence in some major sectors Deep learning algorithms have been a powerful tool in healthcare for medical imaging used to monitor the disease, diagnosis, aid surgical procedures, and management of the disease. The use of artificial intelligence (AI) in diagnostic medical imaging is undergoing extensive evaluation. In most oncology-related diagnoses, the applications of AI are crucial in radiology for various modalities with improved quality such as X-rays, ultrasounds, computed tomography (CT/CAT), magnetic resonance imaging (MRI), positron emission tomography (PET), as well as digital pathology. Images are analyzed with highly specialized algorithms with increased speed and accuracy. Differentiating between normal and abnormal medical images is a key aspect of accurate diagnosis. This is especially essential for detecting cancers early as it will ensure a better prognosis. AI has contributed to medical imaging by improving the quality of images, computer-aided image interpretation, and 8 radionics, and the future of AI in medical imaging will focus on improving speed as well as cost reduction. VI. Legal Aspect – It is of utmost importance that AIs are safe and effective. Stakeholders can contribute to a successful implementation of AI in clinical practice by making sure that the datasets are reliable and valid, perform software updates at regular intervals, and being transparent about their product, including shortcomings such as data biases. In addition, an adequate level of oversight is needed to ensure the safety and effectiveness of AI. 1. Algorithmic fairness and biases Several real-world examples have demonstrated that algorithms can exhibit biases that can result in injustice concerning ethnic origins and skin color or gender. Biases can also occur regarding other features such as age or disabilities. 2. Data Privacy The value of health data can reach up to billions of dollars, and some evidence suggests that the public is uncomfortable with companies or the government selling patient data for profit. In the world of big data, it is of pivotal importance that there are data protection laws in place that adequately protect the privacy of individuals, especially patients. In the following, we will give an overview of relevant provisions and legal developments on data protection and privacy in the US and Europe. 9 3. CyberSecurity Cybersecurity is another important issue we are required to consider when addressing legal challenges to the use of AI in healthcare. In the future, much of the healthcare-related services, processes, and products will operate within the IoT. Unfortunately, much of the underlying infrastructure is vulnerable to both cyber and physical threats and hazards. For example, sophisticated cyber actors, criminals, and nation-states can exploit vulnerabilities to steal or influence the flow of money or essential (healthcare) information. VI. Previous Propositions Made – Sharma et al. [1] propose a system which applies Gray Level Cooccurrence Matrix (GLCM) for extraction of features, Multilayer Perceptron and Naïve Bayes is used for classification with maximum accuracy of 98.6% and 91.6% respectively, additionally referenced that the accuracy of the system could probably be improved by considering a larger data set and extracting intensity-based features with texture based features. Natarajan et al. [2] proposed a method using Median filter for the preprocessing of the Brian MRI image, for segmentation applied threshold segmentation and morphological operations, and then the image subtraction technique is used to get the region of interest in the brain MRI image. Joshi et al. [3] proposed a system for the detection and classification of tumors in Brain MR Images. Firstly extract the tumor region from the brain MRI, then Gray Level Cooccurrence Matrix (GLCM) is used for extracting the texture features, and then using neuro-fuzzy classifier tumor is classified. Amin and Mageed [4] proposed a system using a neural network and segmentation base system to automatically detect the tumor in brain MRI images. Principal Component Analysis (PCA) is used for the extraction as well as then Multi-Layer Perceptron (MLP) is used for classification for detection of tumor in the brain MRI image. 10 George and Karnan [5] proposed preprocessing approaches to enhance Brian’s MRI image. The labels are firstly removed from the MRI image and then Median Filter, Histogram Equalization, and Center Weighted Median (CWM) filter are applied to remove noise as well as for the enhancement of the Brian MRI high-frequency components removal. SharmilaAgnal A et al. [6] the study showed that the system can be achieved 100% accuracy even with a very small dataset of 121 training samples. For better output factors to work upon are feature extraction algorithms, data preprocessing, and larger datasets. In terms of overall performance and 5-fold cross- validation of the proposed system, it outshone the existing system with 80.3% and 88.8% respectively. DibyaJyoti Bora et al. [7]proposed that selection of a clustering algorithm is purely dependent on the purpose of clustering application and the type of data working on. K-means algorithm was suitable for tasks like exclusive clustering whereas fuzzy clustering algorithms were suitable for overlapping clustering tasks. Comparing their computational time it was found that the K-means algorithm was faster as compared with the fuzzy clustering algorithm and the complexity of the fuzzy clustering algorithm is higher than that of the K-means algorithm. Rajeshwari and Sharmila [8] proposed pre-processing techniques used to improve the quality of MRI images before using them. For noise removal, the average, median, and Wiener filters are used. Resolution enhancement is done by the interpolation-based Discrete Wavelet Transform (DWT) technique. Applying Peak Signal to Noise Ratio (PSNR), evaluation of these techniques is done. 11 Sanjeev Kumar et al. [9] proposed a hybrid approach for the classification of tumors in a brain MRI image, which will reduce the time for manual labeling and overcome human errors. DWT (Discrete Wavelet Transform) for the feature extraction and Principal Component Analysis (PCA) for feature reduction and for classification they have used Support Vector Machine (SVM). Sukumar Mehta et al. [10] Proposed a fuzzy-based median filter for de-noising impulsive noise from an image, this method gives a clearer and better image quality as compared with the standard median filter. VIII. My Innovation/Contrivance – 12 HOW WILL THE PROGRAM BE CREATED? ❖ The program will detect and mark the tumor directly from the brain scans using Python, Machine Learning, OpenCV, and Numpy. The Program Can be Found Here (Click here) WHAT ARE ITS ADVANTAGES? 13 ❖ As discussed above, The application utilizes Magnetic Resonance Imaging (MRI) brain scans to automatically detect the majority of tumors. Also, those that go undetected and lead to the patient's uncertain demise. HOW DOES IT WORK? ❖ The program accepts the brain scan from one end and sends the enhanced brain scan with tumor visible, which will make it easier for the doctor to plan the course of action. WHY THIS? ❖ This program is easy to run which means any medical specialist can use it at their convenience anytime anywhere. ❖ It will help thousands of people all over the world by bringing reform in the history of Medicines. HOW SOON CAN IT BE BROUGHT INTO PLAY? ❖ The software must be accessible and convenient to use for all, this further concludes that the only challenge will be to integrate it with MRI run efficiently; once resolved, it will be up and running in no time. Before it is used for professional reasons, it should also be assessed with certain real-life scenarios. IX. My Innovation’s Working– 14 15 1. DATA ACQUISITION The most important part of the program is going to be data acquisition. The process of sampling signals that measure real- world physical conditions and converting the samples into digital numeric values that can be manipulated by a computer is known as data acquisition. The data plays a vital role in the accuracy of the end result. Fig 1: Test Brain Scan used for Testing 16 2. PRE-PROCESSING The function of pre-processing is for the program to process the dataset of real-life brain tumors into feasible data for accurate analysis of the input brain scan. The real-time images that have not processed may contain some noise or distortion, and hence, the image has to be enhanced before applying the algorithm to get better performance further. 17 3. THRESHOLDING Thresholding is a type of image segmentation, where we change the pixels of an image to make the image easier to analyze. In thresholding, we convert an image from color or grayscale into a binary image. 18 4. SEGMENTATION The separating and studying of an individual pixel in an image to identify each pixel uniquely through their pixel values are known as segmentation. 19 This step is carried out to read each pixel of the input image and process it with the dataset to identify the tumor. Fig 5: Code Performing Segmentation on the test Brain Scan taken 20 X. The Difference - Fig 6: Normal Brain Scan Fig 7: Result of my Program with tumor marked clearly 21 XI. Conclusion – ● The main notion of the program I built was to help as many people as it can, who are currently fighting for their lives, due to late detection of their illness. ● “A tumor can grow in the brain and go relatively unnoticed for a great period of time,” says Dr. Dunbar. ● When symptoms do arise, they are often very generalized and most likely be caused by other conditions. Many areas don’t have technologies like MRI or CT-SCANS and their tumor goes undetected for years and results in the death of the patient. ● This is why I created this program. My goal is to make tumor detection more efficient and effective with the right technology. I want to provide the best possible options that will help everyone. ● This innovation of brain tumor detection has been possible due to the enhancement of technologies like Artificial Intelligence and Machine Learning. ● AI and ML are the technologies which if used properly can be used to change the world. They are nothing less than a blessing and a weapon to reform this world. With proper knowledge as well as tools, we can put these technologies to great use for better living. ● These technologies are those promising technologies that will change how we detect brain tumors. ● The idea of creating a program to detect brain tumors came from statistics that showed it’s causing more than 10000 deaths yearly and I hope that my method might create change by early detection and save many lives which are at risk. 22 ★ REFERENCES → ▾ https://www.medicinenet.com/brain_tumor/article.htm ▾ https://www.cancer.net/cancer-types/brain- tumor/introduction ▾ https://medium.com/@mohamedalihabib7/brain-tumor- detection-using-convolutional-neural-networks- 30ccef6612b0 ▾ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7490765/ ▾ Artificial intelligence (AI) and big data in cancer and precision oncology Zodwa Dlamini ⇑, Flavia Zita Francies, Rodney Hull, Rahaba Marima ▾ https://www.thelancet.com/journals/landig/article/PIIS2589- 7500(20)30160-6/fulltext ▾ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7332220/ 23