How This Dad Used AI to Fight His Child’s Brain Tumor

Hey there, I'm Jim Carter III, and welcome back to The Prompt. As a father, this story hit me hard. Today I'm going to tell you about Siki Chen, his daughter Mira, and how a weekend coding project is helping families get through some of the scariest days of their lives. Mira was diagnosed with craniopharyngioma. It's a rare brain tumor in kids that grows near the pituitary gland. These tumors make up about 5 to 10% of childhood brain tumors. They are often benign, but where they grow makes them very risky because they sit close to the optic nerves, the pituitary gland, and other important parts of the brain. Unfortunately, the hard truth is that the tumor comes back in about 30 to 50% of cases after surgery. More surgeries can also cause hormone problems in about 40 to 60% of cases. So families end up in a tough loop of MRIs every 3 to 6 months, hoping the tumor stays quiet. Mira's tumor grew after the first treatment, and that meant her family had to choose between another surgery with a real risk of lasting brain injury or or a different plan. They chose a different plan. With researchers, they tried a mix of two drugs. One was tocilizumab, a drug used for arthritis that blocks IL-6 receptors. The other was Avastin, a cancer drug that blocks VEGF, which helps grow blood vessels that can feed tumors. The idea was to block both signals that help the tumor area stay active. In Mira's case, it worked. Small case reports on this mix have shown cyst shrinkage around 20 to 50%, even though big trials are not here yet. For her family, the results have been strong. Every few months, Mira gets an MRI to see if the treatment is still working. Comparing MRI scans from different dates is very hard. The patient is lined up a little differently each time. Brightness and contrast can change between machines and visits. You end up flipping through hundreds of image slices trying to match the same spot in the brain across scans. For a radiologist, it's slow work. For a parent carrying that fear, it can feel crushing. Siki, determined to help his daughter, built a tool to fix that. In one weekend, he made an open-source MRI viewer for this exact problem. He called it MiraViewer. You load your DICOM files, the standard medical image files, right on your computer. The files do not get sent anywhere. There is no cloud step and no outside server. The scans stay on your machine and run in the browser using WebGL and WebAssembly. That matters when the images are a child's brain scans. Privacy is handled in the simplest way possible by keeping the data local. After the scans load, you draw a box around the area you care about, like the tumor on one scan. Then the tool finds the closest matching slices from the other scan dates you loaded. It helps line things up and makes basic fixes like rotation, zoom, skew, and matching brightness. It cuts down the handwork that usually needs a trained specialist or costly software. Then it gives you two ways to look. One is a grid where you see scan dates side by side at the same spot in the brain. The other is an overlay where you switch between two scans to see what changed. For families tracking tumor growth, that can be the difference between guessing and seeing the change clearly. Sikhi used Gemini and ChatGPT while building the early version to help write parts of the code. The app itself does not call any outside AI tool when you use it. It runs on your computer. What's really cool is that he made it open source and free so anyone can use it, check it, and improve it. This matters because most medical imaging tools are made for clinics and researchers, not for parents at home. The medical imaging software market is about $3.3 billion. Still, many tools need installs or cloud processing, and many non-experts hit a wall when they try to use them. Tools like 3DSlicer, itkSnap, and OhioViewer are strong, but they can feel like too much for a parent trying to make sense of an MRI at midnight. MiraViewer fills that gap. It does not try to replace hospital tools. It helps families see what's happening with their own eyes, using their own data, in private, at no cost. The story goes beyond the software. Sitchi pushed ChatGPT hard with long research prompts, pulled together a small research group, and raised $1.4 million through a crypto token called Mira. The money was donated to the Hank Mitra Lab at the University of Colorado, which studies craniopharyngioma. This is what it looks like when someone with strong tech skills refuses to sit back during a medical crisis. It also reminds me of another story about a tech founder named Sid who stepped away from being a CEO during his own cancer treatment and built what he called a care stack. It was a system for collecting info, keeping records, working with researchers, and using computers to help plan his care. AI gets talked about all the time for business, content, and speed at work. This story shows a more human use. A father built an MRI tool in a weekend that would have been out of reach a few years ago. He used language models to move faster in research. He used open-source code so other families can use the tool. And he used crypto to fund lab work. I love that AI tools can make this kind of work possible for more people than before. Medical AI is moving fast. New methods can help with image tasks like marking tumor areas with less hand labeling. Imaging models are also getting better at spotting findings and helping predict how disease might respond to treatment. There's one detail about this story that really stuck with me. Most families facing rare disease do not have Sikhi's skills. They cannot build a tool in a weekend or raise a million dollars for research. That is why open-source tools like MiraViewer matter. Each free tool that runs locally makes things a little more fair for the next family. If you want more stories like this and more practical talk about how people are using AI on real problems, you can join me and a group of builders in my Slack community, Ctrl+Alt+Build. I share tools, workflows, and projects there, and we talk through real challenges together. I'll drop the link in the show notes for you to check it out. Thanks so much for tuning in and listening today. I think stories like Mira's are a reminder that this work is not just tech talk. It's personal and it matters. I'll catch you in the next episode.