Medical imaging, by definition, refers to several different technologies used to view the human body to diagnose, monitor, or treat medical conditions. Each type of technology gives different information about the area of the body being studied or treated, related to possible disease, injury, or the effectiveness of medical treatment. Starting with the microscope, where cells were viewed for the first time, medical imaging has evolved dramatically over the years. It has reached new frontiers concerning both diagnosis of diseases as well as planning and monitoring treatment efficacy by using the application of artificial intelligence (AI). “The medical imaging market is worth USD 34 billion and is expanding, owing to the increased demand for remote diagnostics due to COVID-19. Another driving force is the rapidly growing geriatric population. There is a shortage of radiologists, neurologists, and psychiatrists, so rapid and objective medical diagnostics are essential for healthcare systems,” says Dr. David H. Nguyen, Ph.D., CEO and Co-Founder, BrainScanology.
Dr. Nguyen is a computational biologist who invented an algorithm to measure shape without measuring area or volume. He has written five novel algorithms to quantify “randomness” in the architecture of tumors (ww.TSG-Lab.org), an approach no one else has computationally conceived before. In 2020, Dr. Nguyen established BrainScanology, Inc, collaborating with his former research intern, Harini Kumar, MBA, who is experienced in both scientific and consumer research with a profound understanding of the current market space, key players and initiatives, and the overall business climate. Now the team is working on launching BrainScanology’s software, called ShapeGenie (www.ShapeGenie.net), which can measure shapes in ways that Deep Learning techniques in the field of computer vision cannot. It allows for the re-inclusion of human creativity in measuring biological shapes that correlate with sickness and disease. “Deep Learning requires the user to define ‘ground truth’ categories, meaning what is considered healthy and what is considered diseased. ShapeGenie, on the other hand, helps the user discover new ground truth categories hidden from the naked eye and Deep Learning. These new categories can then be taught to Deep Learning for rapid detection,” explains Dr. Nguyen.
ShapeGenie allows the user to specify which organ shapes to measure. The results are highly interpretable, allowing the physician to explain to a patient what features were measured and why. “Deep Learning models struggle with interpretability because they are excellent at distinguishing between healthy and diseased patients but do so in a way humans do not comprehend,” adds Dr. Nguyen.
BrainScanology: What Does It Do?
BrainScanology, Inc is a technology startup developing ground-breaking shape analysis software that does not require area or volume measurements. It develops software that allows doctors to measure complex biological shapes to better understand diseases imaged by MRI, CT, X-ray, and ultrasound. Dr. Nguyen says, “We are developing our software as a SaaS product available for all fields of science to do shape analysis: from agricultural engineering to home appliance design, to medical diagnostics, to aerospace engineering.”
BrainScanology’s ShapeGenie software is scheduled to release a minimum viable product (MVP) version in January 2023. ShapeGenie has already provided early adopters with consulting services to test its efficacy for their research. “ShapeGenie will be a cloud-based software (a “SaaS service”) to which users will be able to subscribe, upload their images, and then download the analysis results. In addition to a ShapeGenie license, customers can purchase consulting services from BrainScanology staff on how to best apply ShapeGenie to specific research questions,” shares Harini. BrainScanology also offers a service for developing machine learning models that predict specific diseases based on customer image data.
BrainScanology distinguishes itself from competitors in the market since it has filed a PCT patent application for the use of the LCPC Transform in medical diagnostics ranging from the cellular to the organ to the organismal level (example: finger shapes). Dr. Nguyen states, “While we can build in traditional measures of area and volume in our software, our competitors cannot build in the LCPC Transform into their software.”
BrainScanology also collaborates with individual academic labs or medtech companies. Partners benefit from access to its shape analysis and research expertise. This is useful for creating custom machine learning models and applying for research grants. The company also provides free pilot studies to clients to demonstrate that they can measure what other techniques cannot. Dr. Nguyen adds, “They compare our results with their previous results to see ShapeGenie’s unmatched level of precision and efficacy.”
Big Decisions Result from Big Incidents
The idea of creating BrainScanology came after a series of tragic incidents in Dr. Nguyen’s life. It started with the death of Dr. Nguyen’s close college friend, named Thuan Trinh, who took his own life due to bipolar disorder. During the years of struggle that led to Trinh’s death, Dr. Nguyen was going through deep depression in his own life. At Trinh’s funeral, he made a promise to do something about bipolar disorder. Years after that, Dr. Nguyen’s favorite high school biology teacher, Duane Nichols, died of an aggressive form of colon cancer that had recurred three times. To commemorate his former mentor, who inspired him to study science in college, Dr. Nguyen spent many sleepless nights examining histopathology images of colon polyps to derive a statistical method to measure the subtleties of complex biological shapes.
After a week of scribbling and crumpled paper, the Linearized Compressed Polar Coordinates (LCPC) Transform was founded to honor Nichols at his memorial service. Informally, the LCPC Transform was dubbed the “Nguyen-Nichols Transform.” Nguyen eventually realized that the LCPC Transform could be used to quantify brain folds and thus measure the difference between healthy and diseased brains. Thus was born the concept of BrainScanology, David Nguyen and Harini Kumar’s first startup.
Exceptional Performance Leads to Striking Progress
BrainScanology is only two years old, but it already has early adopters from the world’s top research institutions, with more joining every week. Harini highlights, “Once our SaaS software is released in January of 2023, the number of subscribers will skyrocket because we will offer free subscriptions under a freemium model that allows users to test our software at a limited capacity.” The National Science Foundation invited the BrainScanology team to apply for an SBIR grant for their machine learning model that detects Alzheimer’s disease by analyzing brain MRIs. “This Alzheimer’s model will likely be our first go-to-market clinical decision support product for the USA in the coming years,” added Harini.
Providing Recognition to Encourage Employees
BrainScanology acknowledges its employees, even if they are interns. This manifests as co-authorship of presentations and reports. They also hold regular meetings to keep the team abreast of everyone’s progress and contributions. Nguyen mentions, “As a small company, everyone’s contribution has a positive impact, so we make sure that everyone understands this, which helps them feel ownership of their work.” Individual interns can also pursue independent projects over the summer, allowing them to pursue different topics based on the company’s technology.
Poised to Discover Unanticipated Subtypes of Diseases in the Future
BrainScanology is preparing to discover unknown subtypes of many diseases in the coming years. Harini points out, “Once we get our SaaS software into the hands of many scientists, it will only be a matter of time when other labs make the same groundbreaking discoveries that we have been making over the past two years.” They are taking their technology directly to African countries to develop race- and sex-specific medical diagnostic AI in collaboration with local clinics. This is called the Medical AI for African Nations Initiative, or MAfoAN (“mah-foh-anh”) Initiative (https://africamedicalai.wordpress.com/). “Our technology is amazing. We don’t think that other countries should wait for the U.S. FDA to approve our technology before those countries can start benefiting from it if they have the imaging data and are willing to share it now,” added Harini.
Planning a Global Footprint
Harini says, “Because we measure the shape and not signal intensity, our disease-specific machine learning models can work on images from older MRI machines and CT scans that provide low-resolution images. For example, our Alzheimer’s model is based on the shape of the lateral ventricles in the brain, which is a liquid space that appears obviously different than brain tissue in MRI scans.” Since low-resolution CT scans clearly show the shape of the lateral ventricles, their Alzheimer’s detection model may work well with a low-resolution CT scan, allowing potential Alzheimer’s patients easier access to brain imaging. CT scans are less expensive and faster than MRIs, both of which are advantages.
The software developed by BrainScanology is more effective at detecting organ shape differences than area and volume differences, which can be used for machine learning models that can be applied across different populations of people. “This applies to many psychiatric diseases for which the current method of diagnosis is highly subjective and prolonged. We can cut diagnostic wait times by 10-50X,” underlined Dr. Nguyen.
Their software can generate diagnostic models based on ultrasound sonogram images, making it ideal for remote and rural areas. Since ultrasound images are so low in resolution, any machine learning models developed for MRI images will not work on ultrasound images. “However, this is not a problem for our technology if the shape of interest is visible in both an MRI and a sonogram. The science of machine learning and diagnostics is an iterative, self-correcting process that takes time to optimize, so now is the time to start,” concludes Dr. Nguyen.
For More Info: https://www.brainscanology.com/