— Smart Brain Tumour Detection and Diagnosis
I think its quite obvious that medical imaging totally changed the landscape of the healthcare space, creating an area for new forms of early detection, diagnosis, and treatment. This has worked to aid prevention and optimize patient care.
This technology is one that we heavily rely on. It gives us the ability to make informed decisions with patient care in mind. But can we make it better? 🤔
Current medical practices revolving around Brain Tumour diagnosis rely solely on the physician’s experience and knowledge, as well as their interpretation of MRI’s (Magnetic resonance imaging).
And yet we often see cases of misdiagnosed brain tumours that result in incorrect treatment, or worse, no treatment at all.
This can be a result of errors made by clinicians, but moreover, we can prevent these misdiagnosed cases by improving the very system we rely on so heavily!
Living in a constantly developing world, why do we still rely on outdated methods of medical imaging as a vital method for saving a patient lives?
Crazy Right!
Take this for example.
With new development and extensive research, we’ve come to understand how common we misdiagnosis certain cancerous brain tumours in children. Some children with these particular rare tumours have been getting the wrong diagnoses and, in some cases, the wrong treatment.
Interested in learning more about this story: https://www.fredhutch.org/en/news/center-news/2018/10/pediatric-brain-cancer-misdiagnosis.html
The goal of modern medicine to provide relief of pain and suffering from all patients, working to promote disease prevention.
Health care is to help each person achieve four major goals: prevention of premature death and disability, maintenance and enhancement of quality of life, personal growth and development and a good death.
There needs to be some sort of change to aid physicians in the diagnosis process, allow for decisions to be made faster and more accurately, and properly highlight areas of the brain affected.
Oh, wait! 💡
Medical Imaging has been a pivotal advancement in the healthcare space, shifting the way we look at the human body and treat patients. But even some of our most reliable systems like MRI scans can miss out on some of the important details completely changing the direction for patient care.
MRI’s are widely used medical technology for diagnosis of various tissue abnormalities including the detection and diagnosis of brain tumours. The active development in the computerized medical image segmentation has played a vital role in scientific research, helping doctors understand and visualize abnormalities to take the necessary steps for optimized treatment with fast decision making.
Regular systems of MRI scans is a tremendous improvement and growth in the space of medical technology and saves millions of lives yearly. However, many still have been misdiagnosed and even lost their life due to inadequate detection models.
Question: So, how can we solve this problem?
Answer: Advanced Brain Tumour Segmentation from MRI Images
It's okay, relax. Don’t worry, I gotcha.
I assure you, Brain Tumour Segmentation isn’t as complicated as it may seem. But before we even start thinking about what it is and how it works, let's take a step back and understand brain tumours as a whole.
Keeping it Simple;
A Tumour is basically an uncontrolled growth of cells in any part of the body, whereas a brain tumour is a collection, or mass, of these abnormal cells concentrated in the brain.
The tumours can cause local damage by growing and pushing on crucial areas of the brain. They can also cause problems if they block the flow of fluid around the brain, which can lead to an increase in the pressure inside the skull. Some types of tumours can spread through the spinal fluid to distant areas of the brain or the spine, if not detected and treated.
Brain tumours are categorized as primary or secondary.
A primary brain tumour originates in your brain. Whereas a secondary brain tumour (metastasis) is a tumour growing within the brain that has arisen from the spread of a malignant tumour (cancer) elsewhere in the body.
- A malignant tumour is more dangerous because it can grow quickly and may grow into or spread to other parts of the brain or to the spinal cord. Malignant tumours are also sometimes called brain cancer. Metastatic brain tumours are always malignant because they have spread to the brain from other areas of cancer in the body.
- A benign primary brain tumour is not cancer. Benign tumours can cause damage by growing and pressing on other parts of the brain, but they don’t spread. In some cases, a benign tumour can turn into a malignant tumour.
Key Takeaway → Tumours are BAD (Get that in your 🧠)
Just to show the severity of these brain tumours, they can pretty much affect every single part of our brain. LIKE EVERY PART. Acoustic Neuroma Center, Glioma Center, Meningioma Center, Metastatic Brain Tumor Center, Neurofibromatosis (NF) Center, Pituitary Tumor Center, etc. Are just a few types of tumours originating in different areas of the brain, with a varying of symptoms and detrimental, even fatal effects.
— Provided some more information on these areas, feel free to check it out and learn more 😉
Now that you have a high-level understanding of what Brain Tumours are, let's get into understanding the segmentation process and how it can completely shift current methods of detection and diagnosis.
Taking Our Brain Apart One ̶S̶e̶g̶m̶e̶n̶t̶a̶t̶i̶o̶n̶ Bit at a Time
On a high-level, Brain tumour segmentation consists of extracting the tumour region from healthy brain tissues. The goal of image segmentation is to divide an image into a set of semantically meaningful, homogeneous, and nonoverlapping regions of similar attributes such as intensity, depth, colour, or texture.
However, accurate and effective segmentation of tumours remains a challenging task, since the tumours can have different sizes and locations. Their structures are often nonrigid and complex in shape and have various appearance properties.
The segmentation result is an image of labels identifying each homogeneous region or a set of contours which describe the region boundaries. Which is how we are to detect irregularity within sets of segmented imaging to identify tumours.
But let's get something straight. MRI segmentation isn’t new in practice, its been around for years, but we need to understand the challenges associated with current methods of analysis and diagnosis to make it even better than it already is.
The analysis of these large and complex MRI datasets has become a tedious and complex task for clinicians, who have to manually extract important information. This manual analysis is often time-consuming and prone to errors due to various inter- or intraoperator variability studies.
These difficulties in brain MRI data analysis required inventions in computerized methods to improve disease diagnosis and testing. Nowadays, computerized methods for MR image segmentation, registration, and visualization have been extensively used to assist doctors in qualitative diagnosis.
*Deep Learning Has Entered the Chat*
*Deep Learn* — WHA!? How!?
So How Does Deep Learning Actually Apply to Brain Segmentation and What Are the Benefits?
“The most important difference between deep learning and traditional machine learning is its performance as the scale of data increases.”
When taking on the challenge of leveraging Artificial Intelligence for brain segmentation, the data needs to adequately and accurately predict brain tumours through MRI scans, which going to require TONS of data.
The ability to process such data is most definitely a challenge, requiring not only the ability to get your hands of patient scans but also the computational power need to actually be successful in prediction and diagnosis.
Some of the major challenges associated with training machine learning models on medical imaging are the high-costs of obtaining each dataset, receiving patient approval, and post-analysis of the image by an expert.
To build medical machine learning systems with limited data, researchers apply extensive data augmentation, including stretching, gray-scaling, applying elastic deformations, and more, generating a large amount of synthetic training data.
In following such principles we’ve come to see a tremendous benefit and improvement in identifying abnormal tissues from patient MR images:
The experimental results of the proposed technique have been evaluated and validated for performance and quality analysis on magnetic resonance brain images, based on accuracy, sensitivity, specificity, and dice similarity index coefficient. The experimental results achieved 96.51% accuracy, 94.2% specificity, and 97.72% sensitivity, demonstrating the effectiveness of the proposed technique for identifying normal and abnormal tissues from brain MR images.
— Leveraging deep learning for tumour detection and diagnosis experimental results
The key issue was the detection of the brain tumour in very early stages so that proper treatment can be adopted. Based on this information, the most suitable therapy, radiation, surgery or chemotherapy can be decided. As a result, it is evident that the chances of survival of a tumour-infected patient can be increased significantly if the tumour is detected accurately in its early stage.
Hmmm. A better system of medical imaging that saves time and lives?!
DOPE!
Linkedin: https://www.linkedin.com/in/alim-bhatia-081a48155/
Medium: https://medium.com/@Alim.bhatia
Newsletter: https://mailchi.mp/291f9d2b6bfb/alim-monthly-newsletter